Abstracts from the 37th Annual Meeting of the Society of General Internal Medicine (2024)

SCIENTIFIC ABSTRACTS

“A ONE-STOP SHOP:” PERCEIVED BENEFITS OF DIABETES GROUP VISITS IN THE SAFETY NET CLINIC SETTINGArshiya A. Baig¹; Amanda Benitez¹; Amanda Campbell²; Cynthia T. Schaefer^3,2; Loretta J. Heuer^4,2; Michael T. Quinn¹; Deborah L. Burnet¹; Marshall Chin¹. ¹University of Chicago, Chicago, IL; ²MidWest Clinicians’ Network, Lansing, MI; ³University of Evansville, Evansville, IN; ⁴North Dakota State University, Fargo, ND. (ID #1934849)

BACKGROUND: Diabetes group visits, shared appointments in which patients with diabetes receive self-management education in a group setting and have a medical visit, are an innovative and promising way to deliver diabetes care. The group visit model may be especially promising in safety net clinics where many patients with diabetes in underserved settings receive their care. However, little is known about safety net providers’ perceptions of diabetes group visits in the community health center setting.

METHODS: The research team conducted site visits at community health centers across the Midwest to assess their experiences in providing diabetes group visits. Site visits were conducted at centers that had implemented diabetes group visits in the past or currently had them. Two members of the research team interviewed health center personnel at each site regarding their diabetes group visit program. Health center personnel included chief executive officers, medical and nursing directors, care coordinators, physicians, nurses, physician assistants, diabetes educators, mental health professionals, dietitians, medical assistants, pharmacists, and other recommended staff members who were familiar with the diabetes group visits at the site. Interviewees were asked about the benefits of having diabetes group visits at their site. All interviews were audio-recorded and transcribed. Systematic qualitative analysis techniques were used to identify the range and consistency of opinions and experiences across interviewees.

RESULTS: The research team visited five health center sites across four states in the Midwest and conducted a total of 26 interviews with health center personnel. One site was rural, three were urban, and one was suburban. Health center personnel noted many benefits of group visits to the health center, providers, and patients. The benefits of group visits to the health center included an alignment of the group visit model with the mission of becoming a patient centered medical home, the ability to bill for group visits, and an efficient way to improve guideline-driven care for patients. Provider benefits from group visits included boost to provider morale, opportunities to collaborate with multidisciplinary colleagues, and having more time to focus on other medical concerns during patient’s routine follow-up visits. Patients derived many benefits, including receiving education and medical care in a single appointment, e.g. “the one-stop shop” or “best bang for your buck,” the opportunity to obtain social support and enhanced motivation through peers, and the potential to improve their clinical outcomes by attending group visits.

CONCLUSIONS: Diabetes group visits can offer many unique benefits to safety net clinics by providing patient-centered care, boosting provider morale, increasing multidisciplinary collaboration, and offering patients the convenience of combining an educational session with a primary care appointment. Further studies need to assess best practices in implementing group visits in health centers and assess their impact on patient outcomes.

“A PLACE AT THE TABLE:” EVALUATION OF COMMUNITY MEMBERS’ EXPERIENCES AND EXPECTATIONS FOR ACADEMIC-COMMUNITY PARTNERSHIPS IN HIV/AIDS RESEARCHStella Safo¹; Chinazo Cunningham^3,2; Alice Beckman²; Joanna L. Starrels^3,2. ¹Montefiore Medical Center, Bronx, NY; ²Albert Einstein College of Medicine, Bronx, NY; ³Montefiore Medical Center, Bronx, NY. (Tracking ID #1937992)

BACKGROUND: The foundation of community based participatory research (CBPR) is collaboration between academic researchers and members of a given community. Community advisory boards (CABs) are one mechanism through which academic-community partnerships are formed, but current research about CAB members’ opinions on barriers to collaborations with academics is limited. This qualitative study examined CAB members’ expectations and experiences in working with academic researchers in the field of HIV/AIDS.

METHODS: We conducted 10 semi-structured one-on-one interviews with individuals serving on a CAB for HIV-related research at an urban academic medical center. Participating CAB members were leaders of HIV/AIDS community organizations in Bronx, NY and had at least 5years of experience working in the field of HIV/AIDS. Interview questions focused on participants’ current and previous experiences with academic research and researchers, trust in these relationships, and best practices for collaboration. Interviews were professionally transcribed and data was analyzed using a grounded theory approach. Transcript data were coded by two independent researchers using NVivo 10 software and analyzed in an iterative process to identify emergent themes.

RESULTS: CAB members described positive aspects of inclusion on a CAB, including improved access to information about current HIV topics and the opportunity to help shape HIV research in their communities. However, CAB members also described negative previous interactions with researchers, and a lack of trust in researchers themselves or in the process of conducting research with academic institutions. A major reason for distrust was that power was perceived as unequal, often resulting in miscommunications that led to lasting negative impressions. For example, several members recounted researchers approaching subjects directly, without informing the home organization about the intended study focus. In addition, CAB members felt that expectations for collaboration were not clearly established. CAB members also shared suggestions for improving the collaborative process, including outlining expectations from the outset of the partnership and holding CAB meetings in the community rather than at the academic center.

CONCLUSIONS: We identified complex interpersonal issues in relationships between CAB members and academic partners, particularly around trust and power-sharing in conducting HIV/AIDS research. Our findings suggest that academic-community collaborations could be enhanced by improving transparency, explicitly addressing issues of distrust from previous negative experiences, and ensuring equitable power-sharing among researchers and community members. Further research should be conducted to determine the best ways to engage and maximize the impact of CABs in the field of HIV/AIDS.

“ARE YOU AFRAID OF PNEUMONIA?” PATIENTS’ FEEDBACK ON THE DEVELOPMENT OF AN EDUCATIONAL VIDEO ABOUT PNEUMOCOCCAL VACCINATIONKenzie A. Cameron; Tiffany Brown; Shira N. Goldman. Northwestern University, Chicago, IL. (Tracking ID #1926320)

BACKGROUND: Invasive pneumococcal disease causes a mortality burden among people age 65 and older. Consequently, the Advisory Committee on Immunization Practices recommends pneumococcal vaccination (PnVx) for these adults. However, national rates of PnVx for those age 65 and older are just over 60%. Further, racial and ethnic disparities in vaccination remain: 2011 estimates reported 66.5% vaccination among whites age 65 and older, as compared to only 47.6% among blacks. Informed by our previous research on perceptions toward PnVx among unvaccinated adults, we developed a 2min patient education video targeted at unvaccinated 65 and 66year olds to encourage uptake of PnVx. Our objective was to discuss patient attitudes toward vaccination and beliefs about pneumonia, and obtain feedback regarding suggested revisions for the video.

METHODS: At an academic general internal medicine clinic, we queried the electronic health record to identify 65 and 66year old black and white patients who had no record of receipt or refusal of PnVx. In early 2013, patients were recruited to participate in race-concordant focus groups. Prior to the start of each group, participants completed a brief demographic survey, which included questions regarding past experience with pneumonia and PnVx. During the audio-recorded focus group, we asked participants to describe pneumonia, tell us what they knew about PnVx, and specify what they would want included in educational messages about the vaccine.

RESULTS: Four white and two black patient focus groups were conducted, with a total of 26 participants. Mean age was 65.9 (3.1), 81% were female, 88% had attended college, and 39% reported receipt of PnVx. Based on our previous research, the video included a strong focus on individual susceptibility to pneumonia, as well as personal testimonial from a black man about his late wife who did not think she needed PnVx. Black patients in particular noted that the testimonial was a persuasive part of the video. Participant response to the video was positive, with some individuals indicating a newly found motivation to consider vaccination (e.g., “Even though I refused the pneumonia vaccine already, next time I see my provider, I’m going to tell her I want to get shot,” and “I’m going to go get my shot now. I didn’t realize how potent and prevalent it is.”). However, some participants felt that the video focused too heavily on severity and fear (e.g., “I find it dire. Very uncomfortable watching it.”). The video included an appeal to consistency and the need for vaccination across the lifespan (“Just as you got vaccinated as a child, you need to receive important lifesaving vaccines as an adult”), which was met with favorable responses from participants. Several participants highlighted the importance of discussing PnVx with their providers and indicated that viewing this video prior to an appointment could provide an opportunity for such discussion (e.g., “I might talk to my doctor. I don’t know if she could persuade me to get it. Maybe she could explain it to me differently”).

CONCLUSIONS: Feedback from the focus group participants led to revisions to the patient education video, specifically with regard to reducing the number of times the severity of pneumonia was stressed. The personal narrative was retained as numerous individuals spoke of its persuasiveness, as was a focus on personal susceptibility to pneumonia and the need for vaccination across the lifespan. The revised video is under 2min long, has been integrated into our electronic health record system, and is being tested at the point-of-care to assess its effect on receipt of PnVx among patients eligible for vaccination.

“BEYOND OUR CONTROL”: HOW ORGANIZATIONAL CONTEXT IMPACTS PERFORMANCE MEASUREMENTMolly Harrod¹; Jane Forman¹; Claire Robinson¹; Adam Tremblay^2,3; Leo Greenstone^2,3; Eve A. Kerr^1,3. ¹VA Ann Arbor Health Care System, Ann Arbor, MI; ²VA Ann Arbor Health Care System, Ann Arbor, MI; ³University of Michigan, Ann Arbor, MI. (Tracking ID #1937968)

BACKGROUND: The Patient Centered Medical Home (PCMH) model requires a shift from a physician centric to a team-based approach to care, as well as improvement in process measures that reflect PCMH goals. As a result, as primary care (PC) providers are learning to work in teams, they also must learn how to change their clinic processes to meet PCMH-specific performance measures. The Veterans Health Administration (VHA) transitioned over 900 PC sites to a PCMH model beginning in 2010 and has implemented national metrics of PCMH success that include same day access and continuity with the patient’s usual provider. One large VHA health system implemented a coaching model to help newly formed PC teams redesign their delivery processes to improve access and continuity measures. This study examines how these PC teams discussed and responded to these measures during their coaching sessions.

METHODS: Nine of 20 PC teams were longitudinally observed during coaching sessions (25+ hours) that entailed discussion of both performance measures and processes to improve measure results. Conversations and interactions were recorded via hand-written field notes. We analyzed data using a grounded approach.

RESULTS: As PC teams reviewed their access and continuity measures, two predominant, and often overlapping, themes emerged: lack of control and lack of consistency across the organization. PC teams felt they did not have control over processes both within and outside PC. For example, a centralized call center with variable understandings of PC redesign continued to refer patients to the Emergency Department even though PC providers had open appointments within their schedules, thus impacting continuity. A lack of consistency across the organization as a whole was apparent given that many of the other departments did not change their patient care processes to align with PCMH goals. For example, inpatient physicians followed a rule that patients discharged from the inpatient setting needed “follow-up with primary care physician in 7–10days”. Because this 7–10day mandate meant fewer open slots in a PC provider’s clinic, it decreased same day access. Thus, organizational processes were often working against one another resulting measures that were not reflective of all the changes.

CONCLUSIONS: Improving even straightforward performance measures, like same-day access, requires an understanding of the entire practice context. In our study, measurement results that fell short of goals reflected the organizational inconsistency of processes that impacted individual measures. It is important to develop processes designed to meet PACT goals across the organization rather than within each team or department. As more PC practices become part of Accountable Care Organizations and are assessed using these and other performance measures, a better understanding of what is being measured, how these measures reflect patient-centered care, unintended consequences of measures and other organizational actions, and how to align organizational performance goals is needed so that organizational performance reflects true quality of care and patient-centeredness, and not just performance on individual metrics.

“CODE R”: INTRODUCTION OF A HOSPITAL-WIDE PEER REVIEW PROCESS TO ASSESS RESIDENTS’ POTENTIAL IMPACT ON PATIENT SAFETY AND QUALITYDaniel I. Kim^1,3; Lawrence Loo²; Huy Au^1,3; Ramiz Fargo^1,3; Roger C. Garrison^1,3; Gary Thompson^1,3; Minho Yu^2,3. ¹University of California, Riverside, Riverside, CA; ²Loma Linda University Medical Center, Loma Linda, CA; ³Riverside County Regional Medical Center, Moreno Valley, CA. (Tracking ID #1928321)

BACKGROUND: The hospital physician peer review process is a valuable method of reviewing physician competency and safety, but it is rarely used to evaluate residents, who frequently work under indirect supervision. In 2008, our institution introduced “Code R” to our peer review process to highlight the potential resident involvement in cases undergoing peer review. To our knowledge, there are no published studies that utilize a hospital-wide peer review process to evaluate and track residents in training. Our study sought to characterize resident errors using the ACGME’s core competencies to determine whether interns are more prone to errors than senior residents and to see if there were patterns of resident errors by time of day or month of year.

METHODS: The study is a retrospective review of all peer review cases recorded from 2008 to 2011 by the Quality Management department at Riverside County Regional Medical Center, an academic safety net institution. The institutional peer review process spans all the departments within the hospital and reviews the cases involving patient deaths, unexpected adverse outcomes, incident reports, quality measure fallouts and random physician quality reviews. Cases designated a Code R were identified and the following information was collected using a standardized form: peer review level assigned, area of deficiency, resident post-graduate year, originating department, month, and time of day. Care issues were characterized into the five core competencies of patient care, medical knowledge, professionalism, communication/interpersonal skills and systems-based practice. The chi-square and Fisher’s exact tests were used to make comparisons and an a priori statistical level of significance of p value ≤ 0.01 was used to adjust for multiple comparisons.

RESULTS: Between 2008 and 2011, 8052 individual cases were peer reviewed by the medical staff and 258 (or 3.2%) had “Code R” assigned. Of the total charts initially identified, nine cases were determined to not be resident related, nine charts were unavailable and two cases had incorrect medical record numbers. Two hundred thirty-eight charts qualified for final data analysis. The peer reviewed designations and prevalence in each category were Level 1 (Meets standard)- 13 (5.5%), Level 2 (Meets standard - opportunity for improvement) −196 (82.4%), Level 3 (Deviation from standard) −26 (10.9%), and Level 4 (Unacceptable care) −3 (1.3%). Compared with the non-Code R, or attending physician cases, there were significantly more errors in all the competencies (5.5% vs. 89.3% Level 1 assignments for resident vs. attending physicians). In the Code R cases, 60% (143/238) of the charts had documentation issues. One-third (79/238) of the charts had deficiencies in patient care issues, which include clinical judgment, decision-making and lack of follow-up. The next most common errors were in the competencies of Systems-based Practice and Interpersonal Communication, 9.2% and 8.0% respectively. We did not find any significant differences in resident errors by time of day or calendar year. However, interns (PGY-1) were more prone to errors than senior residents (PGY ≥ 2) (p = 0.002).

CONCLUSIONS: The introduction of “Code R” helped identify cases in which resident involvement potentially impacted patient care and safety. Our analysis of Code R cases suggest specific areas for focused resident supervision and education are needed in the areas of documentation and patient care. This simple peer review process can be readily adopted by other institutions to help improve resident education, facilitate attending supervision, and potentially improve patient safety. Use of this instrument can be used to track the individual resident and attending performance that has not been previously described.

“HAVE YOU THOUGHT ABOUT SENDING THAT AS AN ECONSULT?”: PRIMARY CARE PROVIDERS’ EXPERIENCES WITH ELECTRONIC CONSULTATIONS AT AN ACADEMIC MEDICAL CENTERSara Ackerman¹; Gina Intinarelli⁴; Nathaniel Gleason²; Michael Wang³; Sereina Catschegn²; Delthia McKinney⁴; Ralph Gonzales². ¹University of California, San Francisco, San Francisco, CA; ²University of California, San Francisco, San Francisco, CA; ³University of California, San Francisco, San Francisco, CA; ⁴University of California, San Francisco, San Francisco, CA. (Tracking ID #1936109)

BACKGROUND: Electronic consultations (eConsults) are a recent innovation enabling primary care clinicians to request advice from specialists, via a shared electronic health record (EHR), for lower-complexity clinical problems that do not require an in-person evaluation of the patient by the specialist. Prior studies suggest that eConsults can improve the time-to-access to specialty care and reduce specialty care office visit utilization. Using in-depth interviews, we sought to elucidate primary care providers’ (PCPs) experiences with a recently adopted eConsult program at an academic medical center.

METHODS: We recruited a purposive sample of PCPs from an academic medical center with a recently implemented EHR-based program in which specialists and PCPs receive reimbursement for completed eConsults. Semi-structured interviews were conducted to explore the benefits and limitations of eConsults and their impact on clinical work - including information transfer and clinician-patient and PCP-specialist interactions. Interviews were audiotaped, professionally transcribed, and checked for accuracy. Each interview transcript was reviewed by at least two authors in addition to the interviewer. The constant comparative method was used by the team to collaboratively and iteratively develop themes related to the study objectives.

RESULTS: Nine interviews have been completed to date, with participants representing a range of primary care professions and levels of experience, including faculty physicians, medical residents and nurse practitioners. A total of 20 interviews will be completed by March 2014. We identified three key thematic domains related to eConsult benefits, limitations, and influence on clinical work. 1) Access to expert advice: Most participants reported that eConsults were preferable to the use of informal, or “curbside”, consultations because they offer timely and standardized access to specialist advice without requiring PCPs to call on social networks or ask for favors, and because the consultation is stored in the electronic medical record and is conveniently linked to information about related tests and treatments. As one PCP explained, “I like that it is more formal and that it gets into the patient’s chart and it is recorded and I can refer to it later.” Nonetheless, the occasional use of curbside consultations persisted even among PCPs who routinely use eConsults, and was not seen as incompatible with eConsult use. The perception that an eConsult renders the specialist part of a patient’s care team, while simultaneously obviating the need for an office visit, was also reported as a source of increased patient and PCP satisfaction. 2) Providing better care: The use of eConsults was associated with increased clarity about specialists’ and PCPs’ roles, more transparent and efficient care coordination procedures, and PCPs’ enhanced participation in determining a patient’s follow-up plan. “I just feel like I’m providing better primary care”, said one PCP. Some participants reported that patient management using an eConsult is more time-consuming, and entails greater responsibility on the part of the PCP, than a traditional consultation, but most felt that this was an acceptable tradeoff for the satisfying learning opportunities and increased sense of mastery accompanying the incorporation of eConsults in their workflow. 3) Uneven adoption patterns: A minority of participants was less enthusiastic about eConsults and the frequency of use varied widely among individual providers. Some PCPs described a reluctance to initiate an eConsult because of difficulty with technical aspects of the program, frustration with slower-than-promised responses from consultants, or concern about how and when to communicate with patients regarding a consultant’s recommendations. Others reported “pushback” from patients who wanted in-person advice from a specialist. Most participants reported weighing both clinical and non-clinical factors when deciding when to request an eConsult. “That is why I choose the right patients…”, said a PCP about the need to balance clinical appropriateness and patient preference when deciding between an eConsult and a conventional referral.

CONCLUSIONS: Our results suggest that eConsults present an effective strategy for improving PCPs’ ability to obtain specialist expertise, and that they increase PCPs’ engagement and satisfaction with their clinical work. However, eConsults do not uniformly replace informal “curbside” consultation practices, and decisions about their use are influenced by non-clinical factors such as convenience, usability of the electronic interface, and patient preference. eConsult program designers should elicit continual feedback from providers in order to increase the ease of use and appeal of this mode of consultation.

“IN THE MILITARY, YOUR BODY AND YOUR LIFE AREN’T YOUR OWN:” UNIQUE FACTORS INFLUENCING HEALTH BEHAVIOR CHANGE IN OVERWEIGHT AND OBESE VETERANSMelanie Jay^1,3; Katrina F. Mateo^1,2; Molly Horne^2,1; Allison Squires^4,2; Adina Kalet²; Scott Sherman^1,3. ¹VA NY Harbor Healthcare System, New York, NY; ²NYU School of Medicine, New York, NY; ³NYU School of Medicine, New York, NY; ⁴NYU College of Nursing, New York, NY. (Tracking ID #1920967)

BACKGROUND: Obesity affects 36% of the Veteran population. The Veterans Affairs (VA) Healthcare System offers an intensive weight management program called MOVE!, but only approximately 10% of eligible patients attend even one session. Since veterans see their PCPs 3.6 times per year, developing primary care-based weight management interventions may improve MOVE! attendance. To inform intervention development, we conducted a single-center qualitative study to explore unique VA patient 1) attitudes, barriers, and facilitators to healthy behavior change; 2) uses and understandings of goal-setting; and 3) weight management-related experiences with health care providers, technology, and MOVE!.

METHODS: We sent recruitment letters to overweight and obese patients at the New York Harbor VA. Eligibility criteria included Veterans aged 18–75, having a BMI ≥ 30 or ≥ 25 with at least one co-morbidity, and self-reported ability to read and participate in focus groups. We organized patients into focus groups of 6–11 participants, grouped by gender and MOVE! attendance. The interview guide was informed by the Theory of Planned Behavior (TPB). Sessions were audio-recorded, contextually supplemented with field notes, and transcribed by a professional company. Initial codes were determined based on session recollection and field notes. Using an iterative coding approach, two coders separately reviewed segmented transcripts, modified the codebook as new codes emerged, and met to negotiate codes. We utilized NVivo software to assist with coding and subsequent thematic analysis.

RESULTS: We screened 161 patients for eligibility, 77 were scheduled to attend, and 54 attended one of six focus groups (two female and four male). Participants were predominantly male (63%), African-American (46%) or White/Caucasian (32%), college-educated or higher (74%), and reported having attended MOVE! (61%). We found that although we attempted to stratify focus groups by MOVE! attendance, the VA electronic health record did not accurately capture this information, and some study participants were confused or unaware of their attendance. Thematic analyses revealed how military service informed participants’ attitudes, social norms, and perceived behavioral control with regards to weight management. Participants described how the structured environment and strict standards of the military motivated them to maintain weight control while enlisted, but that lack of autonomy was a barrier to learning how to manage their health when they left service. The military also influenced their perceived need for more personalized and tailored health counseling. Physical activity was largely considered the predominant method to lose weight, and participants felt the MOVE! program lacked a physical activity-related component. While participants acknowledged the potential usefulness of technology to help identify and achieve weight management goals, they expressed the need to have a knowledgeable person (i.e. health coach, PCP, dietician, etc.) provide guidance and support.

CONCLUSIONS: This study revealed how military service may influence veterans’ intentions to lose weight and their weight management-related experiences. This will guide the development of a computer-assisted intervention to treat obesity within primary care and improve access and quality of weight management services. The implications from this study may be limited as it was conducted at a single site within a specific and a mostly self-selected patient population.

“NONE OF THEM APPLY TO ME:” A USABILITY STUDY OF THE VA’S MOVE!23 ONLINE WEIGHT MANAGEMENT SOFTWARE IN LATINA WOMENHector R. Perez⁴; Michael W. Nick⁵; Katrina F. Mateo^1,2; Scott Sherman^1,3; Adina Kalet²; Melanie Jay^1,3. ¹VA NY Harbor Healthcare System, New York, NY; ²NYU School of Medicine, New York, NY; ³NYU School of Medicine, New York, NY; ⁴NYU School of Medicine, New York, NY; ⁵NYU School of Medicine, New York, NY. (Tracking ID #1920971)

BACKGROUND: The obesity epidemic disproportionately affects Latina women, but there are few interventions in this population. The MOVE!23 software program is a publicly available, English-only, online questionnaire used by the Veteran’s Affairs (VA) Administration to facilitate weight management. It evaluates patients’ current behaviors and stage of change and provides computer-generated, individualized advice. Completion of MOVE!23 during an intensive weight management program is associated with greater 6-month weight loss outcomes. Thus, adapting this tool for use in Latina women may be beneficial. However, the MOVE!23 is not specifically tailored for use in non-VA populations and its usability in this population is uncertain. The International Organization for Standardization (ISO) defines usability as “the extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency, and satisfaction in a specified context of use.” In order to determine the utility of tailoring the MOVE!23 for Latina women and translating it into Spanish, we conducted usability testing of this software.

METHODS: We recruited English-speaking, adult Latina women. A researcher observed each participant individually as they used the MOVE!23 . Participants were asked to verbalize their thoughts and reactions as part of a “Think-Aloud” protocol. A semi-structured interview followed that elicited reflections on this experience and how it might influence lifestyle behavior change. We utilized ScreenFlow (Version 4.0.3) to capture on-screen interactions, and each session was digitally recorded and professionally transcribed. A coding guide was derived from the ISO definition of usability and focused on positive and negative interactions between four domains (Tool, Task, User, Context). Two researchers reviewed and coded the transcripts independently using the guide and secondary codes were derived inductively. The researchers met to review the codes and resolve discrepancies. NVivo software was used to facilitate data analysis.

RESULTS: Six female Latina participants completed the study (mean age 39, SD = 17); all had lived in the US for more than 20years. Analysis indicated general satisfaction with tool navigation and the content, sequencing and segmentation of materials. Participants wanted individualized patient handouts to be more accessible and easily available on paper and suggested that images and icons might enhance visual appeal. Significant tool-user interactions included problems with navigation, difficulty understanding questions and answer choices, and sensitivity to the personal nature of the topics. User-context interactions included lack of experience with technology and requests for more support during completion of the questionnaire. In addition, subjects described personal experiences or beliefs while interacting with the materials. The extent to which questions or advice approximated their own experiences influenced the ease with which they were understood.

CONCLUSIONS: This study supports the utility of translating and tailoring the MOVE!23 for use in Latina women and identified several areas for improvement. Findings suggest that task execution may be enhanced when users are able to compare and reconcile the online materials with personal experiences and beliefs, highlighting the need to take these into consideration when tailoring software. Further research will aim to tailor and pilot the MOVE!23 software in populations of Latina women.

“SOMEDAY IT WILL BE THE NORM”: PHYSICIAN PERCEPTIONS OF THE CLINICAL UTILITY OF WHOLE GENOME SEQUENCINGJason L. Vassy^1,2; Kurt D. Christensen²; Melody J. Slashinski³; Denise Lautenbach²; Jill Oliver Robinson³; Jennifer A. Blumenthal-Barby³; Lindsay Feuerman³; Robert C. Green²; Amy L. McGuire³. ¹VA Boston Healthcare System, Boston, MA; ²Brigham and Women’s Hospital, Boston, MA; ³Baylor College of Medicine, Houston, TX. (Tracking ID #1934488)

BACKGROUND: There is excitement that medicine is entering a genomic era, in which knowledge of one’s genetic makeup can guide medical decision-making. Whole genome sequencing (WGS) can identify 3–5 million genetic variants for an individual. However, the clinical significance of most of these variants is currently unknown, and the specific clinical contexts in which WGS might add value to current standards of care have not been identified. The uptake of WGS into medical practice will largely depend on whether and how physicians perceive it to improve patient care.

METHODS: We have enrolled primary care physicians (PCPs) and cardiologists to participate with their patients in the MedSeq Project: a randomized trial of WGS in the clinical care of healthy adult primary care patients and cardiomyopathy patients. We used mixed methods to describe the clinical utility that physicians perceive WGS to have. In a baseline survey on enrollment, we asked physicians to rate on a scale of 1–10 (“Not at all useful” to “Extremely useful”) how useful they thought WGS would be for “managing [their] patients’ health” at two times: now and in the future. We categorized responses as not useful (1–5) or useful (6–10) for analyses. We then conducted in-depth semi-structured interviews to explore whether and in what circumstances physicians might perceive WGS to have clinical utility. Transcribed interviews were consensus-coded with thematic content analysis to identify emergent themes.

RESULTS: Of 17 physicians (mean age 52years), nine were women, and five were of non-white race. Nine were PCPs, and 11 reported no genetics training “beyond the typical medical school curriculum.” Only 4 (24%) said that WGS had utility now, but 15 (88%) predicted it would have future utility for their patients’ care. We identified three major themes influencing physician perceptions of utility. 1) Inevitability: Physicians expected WGS to be widely integrated into medical practice in the near future. When asked his opinion about the routine use of WGS in clinical care, one responded, “That’s like saying, ‘What do I think about getting cholesterol tests on people?’ I mean, it’s a reality, and I think it’s going to happen.” 2) Uncertainty: Physicians perceived WGS results to have too little analytic and clinical validity at present to be useful for patient care. One commented, “It’s like doing a CT scan and finding a nodule and not knowing what it means. And the next thing you know, you have a biopsy, and then you have a complication from the biopsy.” 3) Contexts of utility: In particular, physicians anticipated that WGS may 1day be useful for nudging patient behavior change, tailoring drug therapy (pharmacogenomics), and efficiently targeting preventive health maintenance. One said, “If it turns out that we don’t have to do colonoscopies every 10years on 100% of the people, maybe that will save a lot of money. If we know who we really should be doing PSA’s or mammograms on, instead of doing them on everybody, we can save a lot of money, stop creating a lot of harm and disability.” Physicians often used analogies to current medical testing to describe their perceptions of WGS.

CONCLUSIONS: Physicians are skeptical about the current clinical utility of WGS for general patient care but anticipate a near future when the clinical integration of WGS will be widespread. Evidence that WGS improves health outcomes will likely be necessary to persuade physicians of its utility in patient care.

“WE TORTURE THEM BEFORE THEY DIE, EVEN THOUGH WE KNOW THAT THEY ARE GOING TO DIE:” THE INFLUENCE OF HOSPITAL POLICIES AND CULTURE ON ETHICAL DNR DECISION MAKING IN THE US AND UKElizabeth Dzeng^1,2; Michael P. Kelly²; Sydney M. Dy¹; Thomas J. Smith¹; Martin Roland²; Stephen Barclay²; David M. Levine¹. ¹Johns Hopkins School of Medicine, Baltimore, MD; ²University of Cambridge, Cambridge, United Kingdom. (Tracking ID #1939234)

BACKGROUND: Physicians face ethical challenges when there are conflicts between autonomy and beneficence, which is defined as acting in the best interest of the patient. Shared decision-making, where the patient and doctor form a partnership to decide on treatment plan, has emerged as a model on the spectrum in between medical paternalism and patient autonomy. In the United States, there are significant variations in hospital policies regarding Do Not Resuscitate (DNR) decision-making. In the United Kingdom, it is standard practice for doctors to make unilateral decisions to institute a DNR order in situations where survival is unlikely. The aim of this study is to explore how doctors in the US and UK balance autonomy and beneficence in decisions regarding DNR status at the end of life and how their ethical values and clinical approach are shaped by their institution’s culture and legal frameworks.

METHODS: Semi-structured in-depth interviews were conducted with 58 internal medicine doctors across four sites (New York, Baltimore, Seattle and Cambridge, England), who were routinely involved in DNR conversations with end of life patients. Participants were purposively sampled by stage of training and medical subspecialty to provide a wide range of perspectives and contribute to understanding emerging patterns and themes. Interviews lasted on average 60min and were audio-taped and transcribed. Transcripts were analyzed and double coded using thematic analysis with an interpretive approach.

RESULTS: 13–16 doctors participated at each site. Approximately equal numbers of senior faculty, junior faculty, fellows and residents were interviewed. Experienced doctors at all sites generally felt comfortable engaging in shared decision-making and when clinically appropriate, making more paternalistic recommendations against resuscitation. However, there was greater variation amongst residents and fellows that was site specific. At each hospital, there appeared to be a dominant mode of decision-making that impacted trainees’ perspectives on DNR decision-making. Although hospital policies primarily dictate whether patient preferences should be honored, trainees at hospitals with policies that prioritize patient autonomy often interpreted them to mean that they should not provide clinical recommendations. They felt compelled to offer choice even if the chance of successful resuscitation was negligible. In contrast, trainees at hospitals that had policies that encouraged paternalism or shared decision-making, which included the hospital in the UK and one in the US, felt more comfortable expressing clinical judgment against resuscitation when appropriate. Trainees at hospitals that prioritized autonomy more frequently perceived conflict between honoring autonomy and acting with beneficence and recalled more conflicts with patients regarding preferred decision.

CONCLUSIONS: Institutional norms and policies influence how doctors develop their professional attitudes and practices regarding DNR decision-making. Doctors at the American hospital that encouraged shared decision-making, had attitudes and beliefs about DNR decision-making that were more similar to British doctors than doctors at the other American sites. Hospitals characterized by higher levels of shared decision-making and paternalism provided an environment where trainees felt more comfortable withholding resuscitation when success was unlikely and in doing so, experienced fewer patient conflicts.

“WHAT DOES A GOOD LIFESTYLE MEAN TO YOU?” A QUALITATIVE ANALYSIS OF FIRST YEAR MEDICAL STUDENTS’ VIEWSKimberly Clinite¹; Alex Garnett¹; Shalini Reddy¹; Kent DeZee²; Stephanie Kazantsev¹; Steven Durning²; Jennifer R. Kogan³. ¹University of Chicago, Pritzker School of Medicine, Chicago, IL; ²Uniformed Services University, Bethseda, MD; ³Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA. (Tracking ID #1938451)

BACKGROUND: Medical students consider a specialty’s lifestyle when selecting a career; meeting the physician workforce shortage requires an understanding of how students define a “good” lifestyle. Understanding the lifestyle expectations of entering students may help medical educators guide students towards careers that will fulfill their ambitions and serve the health needs of our society. The purpose of this study was to determine how first year medical students define a good lifestyle.

METHODS: First year medical students at 11 schools were surveyed at matriculation in 2012. The survey included the open-ended question, “When someone says ‘That specialty has a good lifestyle,’ what does that mean to you?” Two investigators performed a directed content analysis of responses. Descriptive statistics of the themes were calculated.

RESULTS: 60% (1020/1704) of students completed the survey; 934 answered the open-ended question. Responses encompassed four major themes: schedule control (66% of students), off-time (54.5% of students), financial aspects (35% of students), and work-life (21% of students). Schedule control was described with multiple subthemes (reasonable hours, flexible hours, predictable hours, and limited call). Nine student responses were categorized as disparaging.

CONCLUSIONS: This study has helped establish how matriculating medical students define lifestyle, a known factor in career selection. Matriculating medical students’ definition of lifestyle is largely defined by schedule control, though students’ concept of schedule control varies. Salary associated with a good lifestyle ranged from $175,000–$400,000. Understanding early perceptions of career decision-making may provide a context for medical educators to begin conversations with students about career choice. Additionally, specialties experiencing physician shortages should consider how their specialty will evolve to meet the needs of society while addressing the concerns of trainees. Policies that hope to successfully address the medical workforce shortages must account for students’ desire to have control of their work schedule.

“WHEN YOU ARE OLD, HAVE CANCER AND A STORM IS APPROACHING” -THE EFFECTS OF HURRICANE SANDY ON CANCER PATIENTS AND PROPOSALS ON POTENTIAL INTERVENTIONSMisako Nagasaka¹; Hassan Alsabbak¹; Zaid Aljuboori²; Koji Sasaki³; Alfred Burger¹; Benjamin Levy¹. ¹Beth Israel Medical Center, New York, NY; ²Bronx Lebanon Hospital, New York, NY; ³MD Anderson Cancer Center, Houston, TX. (Tracking ID #1939110)

BACKGROUND: The estimated loss to the nation’s economy from Hurricane Sandy has climbed to as much as $50 billion, making it one of the nation’s most costly disasters. While data on the economic impact of natural disasters are often reported in a timely manner, very little data exists on the impact of natural disasters to health care, especially in cancer care.

METHODS: A retrospective chart review was performed to examine the impact of Hurricane Sandy and the subsequent blackout in southern Manhattan on emergency room (ER) visits of patients with a diagnosis of cancer. The records of cancer patients who had visited ER during the week of Hurricane Sandy (10/29/2012 to 11/4/2012) were reviewed. For control, ER visits of cancer patients of the same dates 1year prior (10/29/2011 to 11/4/2011) were reviewed and compared. All data were extracted from the ER and/or inpatient electronic medical records. Collected data included age group, sex, cancer type, chief complaint, ER diagnosis and admission status. The diagnosis of cancer was extracted from the “History of Present Illness” or “Past Medical History” and included solid as well as hematologic malignancies. For patients who were admitted, additional data including metastatic disease, comorbidities, length of stay and mortality were extracted.

RESULTS: During the week of Sandy, 144 ER visits of cancer patients were identified. One hundred eighteen were identified the prior year. In patients over age 70, 84 visits were observed during Sandy, while only 41 visited the previous year (p value: 0.000143). The chief complaint of “shortness of breath (SOB) or asthma” was significant, with 21 ER visits in 2012 and only eight in 2011 (p value: 0.045). Out of the 21 visits of SOB, 10 required continuation of their current therapy (8 requested home oxygen therapy and two requested dialysis), while these requests were unobserved the prior year. During the Sandy week, 77 ER visits resulted in admission, while 64 were admitted the year prior (p value: 0.902). Out of 77, seven expired and from 64, four expired (p value: 0.759). The ER admitting diagnoses of the seven who expired during Sandy included a septic patient with fungating head and neck cancer, a brain cancer patient with respiratory failure, a lung cancer patient with pleural effusion, a sarcoma patient with pneumonia, an ovarian cancer patient with anemia, a melanoma patient with acute coronary syndrome and a lung cancer patient with SOB requesting home oxygen therapy. Corresponding data from the year prior included a lung cancer patient with pancytopenia, a CLL patient with pneumonia, a septic patient with breast cancer and a lung cancer patient with suicidal ideation.

CONCLUSIONS: Although there were no significant increase in the total ER visits of cancer patients during Sandy, there were significantly more ER visits during Sandy in cancer patients over age 70, suggesting increased vulnerability to this cohort. To our knowledge, this is one of the first few studies to evaluate cancer care related to natural disasters. In retrospect, the admission and death of a lung cancer patient with SOB requesting continuation of home oxygen therapy may have been the one potentially preventable. Cancer patients requiring additional medical assistance such as home oxygen therapy and dialysis may benefit from a natural disaster back-up plan. House calls, early and prioritized evacuation of these patient groups may be some options to minimize the effects of an approaching natural disaster.

Top 10 Chief Complaints

30 DAY READMISSION RATE FOR PATIENTS DISCHARGED WITH CHRONIC OBSTRUCTIVE PULMONARY DISEASE (COPD): ANALYSIS OF 1,858,618 ADMISSIONS. WHICH PATIENTS ARE MOST AT RISK?Devin B. Malik; Sourabh Aggarwal. Western Michigan University School of Medicine, Kalamazoo, MI. (Tracking ID #1939043)

BACKGROUND: COPD is a major cause of morbidity and is associated with a high 30-day re-admission rate (RR) and economic burden on health care. This study was done to determine demographic parameters associated with high 30-day RR secondary to COPD

METHODS: Nationwide Inpatient Sample data was used to extract data of patients discharged with DM with complications for years 2009–2011 using clinical classification software (CCS). NIS represents 20% of all hospital data in US. All the patients who were discharged with primary diagnosis of COPD and readmitted within 30days were identified and categorized based on admitting diagnosis. Patients were classified as readmissions secondary to COPD as a primary cause, readmissions with COPD as a secondary cause and non-COPD associated readmissions. Statistical analysis was done using SPSS.

RESULTS: We identified a total of 1,858,618 admissions for COPD nationwide during the study period with total 30-day RR of 21.06% and RR of 7.22% secondary to COPD as a primary cause (Table 1). Age group (45–64), males, patients under Medicaid, patients in low median income for zip code and in metropolitan areas had higher 30-day re-admissions secondary to COPD as a primary cause (P < 0.001). Young patients (18–44years of age), females, patients covered under Medicaid and living in Metropolitan areas had higher 30-day RR secondary to non-COPD related causes.

CONCLUSIONS: Strategies to reduce 30-day readmissions secondary to COPD should be focused on more susceptible population including males, aged 45–64years old, under Medicaid, having low median income for zip code and staying in metropolitan areas.

A CLUSTER RANDOMIZED CONTROLLED TRIAL OF INTERVENTIONS TO IMPROVE WORK CONDITIONS AND CLINICIAN STRESS: RESULTS FROM THE HEALTHY WORK PLACE (HWP) STUDYMark Linzer¹; Sara Poplau²; Ellie Grossman³; Steven H. Yale⁴; Anita B. Varkey⁵; Eric Williams⁶; Lanis Hicks⁷; Roger Brown⁸; Jill Wallock⁵; Diane Kohnhorst⁴; Michael Barbouche⁹. ¹Hennepin County Medical Center, Minneapolis, MN; ²Hennepin County Medical Center, Minneapolis, MN; ³New York University Medical Center, New York, NY; ⁴Marshfield Clinic, Marshfield, WI; ⁵Loyola University Medical Center, Chicago, IL; ⁶University of Alabama, Tuscaloosa, AL; ⁷University of Missouri, Columbia, MO; ⁸University of Wisconsin, Madison, WI; ⁹Forward Health Group, Inc,, Madison, WI. (Tracking ID #1927618)

BACKGROUND: Work conditions in primary care are associated with stress, burnout, intent to leave the practice and quality of care. Time pressured office visits, lack of control of the workplace, a chaotic work pace, and lack of values alignment between clinicians and leaders contribute to stress and burnout. To determine the best ways to organize clinical practice, we measured the effects of work condition interventions on clinician and patient outcomes. Our goal was to assess if a “healthy workplace” could decrease clinician stress and improve care quality. This abstract reports on clinician outcomes.

METHODS: We performed a cluster randomized controlled trial in 34 rural and urban clinics in the upper Midwest and New York City. Primary care clinicians (general internists, family physicians, nurse practitioners and physician assistants) were studied. Seventeen clinics were randomized to participate in quality improvement interventions (QIIs) to improve worklife. Time pressure, work control, pace of work (chaos) and values alignment with leaders were measured at baseline and approximately 1year later, along with clinician stress, burnout, satisfaction and intent to leave the practice in 2years. The 17 clinics randomized to the intervention arm received their baseline data in an Office and Work Life (OWL) summary measure. QIIs were chosen by clinics based upon baseline OWL data, and focused predominantly in three areas: 1) improved communication, 2) changes in workflow and 3) clinical QI projects (e.g., a new prescription-refill telephone line, or a medication reconciliation project). Two level multilevel regression analysis was performed to assess the impact on clinician outcomes of being in the intervention arm. When these analyses did not disclose an overall intervention effect, we then explored whether clinicians who showed improvement in burnout, satisfaction or intent to leave the practice were more likely to be found in intervention clinics. Multivariate modeling determined which types of interventions were associated with improvement in these clinician outcomes.

RESULTS: Within 34 clinics, 166 clinicians were initially studied, and 139 subjects completed the study. Data were analyzable from 32 clinics and 135 clinicians. Burnout was present at baseline in 41% of clinicians in intervention clinics and in 33% of controls. While there was no overall effect of all interventions considered together, there were more clinicians with improvement in burnout in intervention vs. control clinics (21.8% versus 7.1%, Odds Ratio (OR) of burnout improvement 4.0, p = 0.014). Burnout was almost six times more likely to improve within clinics performing workflow interventions than in control sites (OR 5.9, p = 0.015), and was also more likely to improve in clinics performing clinical QI projects (OR 4.8; p = 0.021). Clinician satisfaction improved more often in the 10 clinics with interventions in the areas of communication and workflow than in control sites (OR 3.1; p = 0.042).

CONCLUSIONS: While these results should be confirmed in future studies, they suggest that organizations can address high rates of burnout by 1) measuring clinician outcomes (satisfaction, stress and burnout), 2) providing feedback on these outcomes to clinicians, clinic managers and staff, and 3) intervening by improving communication and workflow and initiating clinical QI projects.

A COMMUNITY PARTNERED APPROACH TO ENGAGE DIVERSE COMMUNITIES IN STROKE DISPARITIES RESEARCHNazleen Bharmal¹; Aziza L. Wright^3,2; Stefanie D. Vassar¹; Felica U. Jones³; Loretta Jones^3,2; Rebekah Wells⁴; Jason Cienega²; Arleen Brown¹. ¹University of California, Los Angeles, Los Angeles, CA; ²Charles R. Drew University of Medicine & Science, Los Angeles, CA; ³Healthy African American Families II, Los Angeles, CA; ⁴University of North Carolina, Greensboro, NC. (Tracking ID #1929958)

BACKGROUND: Racial/ethnic minorities have a higher burden of stroke and worse post-stroke outcomes. Contributors to this disparity include lower awareness of stroke signs and treatment, differences in access to care, mistrust in the healthcare system, and low participation in stroke clinical research studies. Despite emerging evidence that community input into interventions may enhance feasibility and sustainability, few stroke studies have incorporated the community perspective. The Los Angeles Stroke Prevention & Intervention Research Program to Reduce Health Disparities (LA SPIRP) is a center to reduce stroke disparities. The Community Engagement Core of LA SPIRP consists of community and academic co-leaders that review projects and work with investigators to understand community research priorities. To enhance public engagement in stroke research, our community-academic collaborative hosted a one-day symposium in South Los Angeles. The objectives of the symposium were to increase awareness of stroke, share information about and receive feedback on the LA SPRIP research projects, and promote collaboration between researchers and community partners to address stroke disparities.

METHODS: We used a community-partnered participatory research process to conceptualize the symposium. The symposium started with talks that provided information on stroke risk factors, stroke disparities, and the impact of stroke from the perspective of patients and caregivers. Simultaneous translation was available for monolingual Spanish speakers, and printed materials were available in English, Spanish, Korean, and Mandarin. We surveyed participants about stroke knowledge, stroke disparities, and trust in medical researchers. In the afternoon, attendees participated in six facilitated, small group breakout sessions (one conducted in Spanish) that addressed two questions: 1) What can we do as a community to prevent stroke?, and 2) How can we recruit more diverse populations in stroke research studies? Breakout session notes were analyzed using qualitative data analysis: content analysis used to code notes from group and pile-sorting by community and academic attendees to identify themes.

RESULTS: The symposium was attended by 236 community residents; staff from local clinical, social service, and faith-based organizations; and researchers. The survey response rate was 54%. Over 90% of the attendees were African American, Latino, or Asian; 40% had hypertension. The majority of respondents had limited prior knowledge of stroke, and most had learned about stroke from family members, friends, or media. The average increase in knowledge about stroke pre- and post- didactic sessions was 15%. On average, 21% of respondents did not trust medical researchers to provide full explanations of research studies or protect participants from harm. Community recommendations for community-based strategies to prevent stroke focused on types of educational efforts (n = 128 quotes), educational content and cultural considerations (79), venues to share information (65), healthy eating efforts (50), partnerships with trusted stakeholders and institutions (34), types of audiences to receive stroke awareness information (30), physical activity efforts (26), and blood pressure screening (14). Community recommendations to enhance minority recruitment into clinical research studies included types of outreach methods (116), communication strategies (72) and partnerships with trusted stakeholders (51) to gain trust, types of incentives for participation (48), and considerations about the relevance and location of the research study for minorities (11).

CONCLUSIONS: Community-academic partnered events are effective ways to obtain community input into stroke research and may build trust and foster collaborations with community members for stroke research. Recommendations from the community stroke symposium will be used to frame future annual symposia, inform current LA SPIRP research, and develop smaller, culturally targeted stroke disparities programs in African-American, Latino, Korean, and Chinese-American communities in Los Angeles.

Recommendations from Breakout Sessions

A COMPARISON OF ACTIVE ADVERSE EVENT SURVEILLANCE SYSTEMS WORLDWIDEJodi B. Segal^1,2; Yulin Huang¹; Jinhee Moon¹. ¹Johns Hopkins Bloomberg School of Public Health, Baltimroe, MD; ²Johns Hopkins University School of Medicine, Baltimore, MD. (Tracking ID #1930394)

BACKGROUND: Typically, fewer than 5,000 people have been exposed to a drug before marketing, so adverse drug events (ADEs) at low frequency will not have been observed. Drug regulatory agencies and industry are vigorously developing active surveillance systems for drugs, vaccines, and medical devices in order to quickly identify therapies causing harm to exposed patients. We conducted an environmental scan in order to describe the status of active surveillance systems for the detection of ADEs, worldwide.

METHODS: We developed a working definition of active surveillance systems. The key components of the systems that we sought to capture were: Structure—what is the structure, partnership relationships, and funding of existing active surveillance systems? Data—what are the data used in existing surveillance systems, what patient populations are included, and what is the content of these data? Function—what are examples of adverse drug events that have been identified with the active surveillance systems? Paired researchers reviewed published literature, as well as online sources, to identify existing systems. Data was extracted into tables designed to capture the key components of the active surveillance systems. We reviewed and synthesized the information across systems to identify common, and presumably, essential features of active surveillance systems.

RESULTS: We identified 12 active surveillance systems meeting our criteria, meaning that they were designed for post-marketing surveillance of pharmaceutical products (including vaccines), had a goal of generating post-marketing drug safety information, do not require personnel to initiate safety reports (describing individual cases), use real-world data which are generated from routine practice, and involve data from more than a single institution. Three systems are U.S. based - the FDA Sentinel Initiative, the Federal Partners Collaboration, and the Vaccine Safety Datalink; two are Canadian systems—The Canadian Network for Observational Drug Effect Studies (CNODES) and the Vaccine and Immunization Surveillance in Ontario (VISION); and two are European—the EU-Adverse Drug Reaction (ADR) Alliance and the Vaccine Adverse Event Surveillance and Communication (VAESCO). Additionally, there is a Japanese surveillance system called the Medical Information for Risk Assessment Initiative (MIHARI), a trans-Asian system called the Asian Pharmacoepidemiology Network (AsPEN), and one in Shanghai called the Shanghai Drug Monitoring and Evaluative System (SDMES). Two additional systems in the U.K. met our criteria: the Tayside Medicines Monitoring Unit (MEMO) and the Drug Safety Research Unit (DSRU), an independent academic unit that uses diverse European data for pharmacovigilance. The surveillance systems all use administrative claims or electronic medical records for data; some conduct pharmacovigilance explicitly on behalf of a regulatory agency, while others function more independently.

CONCLUSIONS: Active surveillance systems are rare. North America and Europe have the most developed systems and most coverage of the population; although some Asian countries are making good advances. There is no coverage of the patients exposed to medications in India, the Eastern European countries, or the South American and African continents. With the ever increasing availability of electronic data, active ADE surveillance should be increasing feasible. This should translate to more rapid signal detection, timely validation, and then steps to assure that drugs and vaccines are used safely.

A DESCRIPTION OF NEEDS FOR CARE COORDINATION AT A COMMUNITY-BASED GENERAL HOSPITAL IN JAPAN’S SUPER-AGED SOCIETYYukio Tsugihashi. ¹Tenri Hospital, Tenri, Japan; ²Tenri Hospital, Tenri, Japan. (Tracking ID #1920518)

BACKGROUND: Japan’s population is aging rapidly because of long life expectancy and a low birth rate. In 2025, it is estimated that the percentage of aged population will increase to 25%. Furthermore, the standout features of the Japanese healthcare system are “universal insurance scheme” and “unrestricted access to any doctor” which add strain to an already overburdened healthcare system. Therefore, a well-designed collaborative system is essential, one that supports the community healthcare and enables functional differentiation and the establishment of referral/transferal networks among a variety of institutions. The purpose of this study is to investigate 1) a medical service area in a community-based general hospital in Japan, 2) the places where the patients were discharged with professional discharge planning and 3) to estimate the future demographic changes in the number of inpatients.

METHODS: Participants: Inpatients discharged from one community-based general hospital in 2012. Setting: Tenri hospital, Tenri city, Nara, Japan (a total of 1001 beds, 26 departments and over 230 medical doctors) Study design: Retrospective cohort study Outcomes: 1) Mapping the medical service area 2) Specifying the places where the participants were discharged from the hospital that provided professional discharge planning 3) Estimating the changes of inpatient numbers and their age groups in 2012, 2025 and 2035. Analysis: First, we described the medical service area by connecting the participants’ address code and Geographical Information System (GIS). Second, a focus group interview lead by professional physicians was performed to find out which diseases frequently required professional discharge planning. Then, we evaluated the places where these patients were discharged from the hospital. Lastly, we projected the annual change in the number of inpatients and their respective age groups in the major medical service area for 2025 and 2035 respectively.

RESULTS: A total of 17,196 patients were included in this study. The medical service area was wide and included some cities outside Nara prefecture, the established government-designated administrative medical area (Figure). In addition, we highlighted five disease categories that needed professional discharge planning: cerebral stroke, fracture, cancer, infectious disease and heart failure categories. The results of professional discharge planning showed that the patients’ homes were the main place to be discharged to. There were some notable trends found between the places where patients were discharged to and the disease categories (Table). Furthermore, we estimated the change in the number of inpatients living in the top 10 areas to be 12,242 in 2025 (41.2% of patients being 75years old or over) and 11,337 in 2035 (42.2%) compared with 12,088 (30.9%) in 2012.

CONCLUSIONS: We investigated the medical service area of a community-based hospital in Japan highly reflective of its super-aged society and the needs for care coordination based on the actual practical performance. GIS analysis would help the general hospital to foresee the medical service area more concisely while preparing for the collaborative healthcare network. Additionally, to prepare for the rapid aging of inpatients, general hospitals should develop a collaborative healthcare system and a system for discharge planning appropriate to the medical service area and disease-specific needs for care coordination.

Table. Places where the participants were discharged from that provided professional discharge planning

*mean ± SD, **n (%), †the most frequent place in each disease category

Figure. The medical service area of Tenri Hospital. Each dot represents the density of inpatients’ number living in each specified address code. Thick lines indicate the borders of cities in Nara Prefecture. *Tenri City.

A FAILURE TO COMMUNICATE: CHALLENGES AND SOLUTIONS TO CARE COORDINATION BETWEEN HOSPITALISTS AND PRIMARY CARE PROVIDERS AROUND HOSPITALIZATIONSChristine D. Jones¹; Maihan B. Vu²; Christopher O’Donnell³; Mary Anderson¹; Snehal Patel³; Eric A. Coleman⁴; Darren A. DeWalt⁵. ¹University of Colorado, Denver, Aurora, CO; ²University of North Carolina at Chapel Hill, Chapel Hill, NC; ³Duke University School of Medicine, Durham, NC; ⁴University of Colorado, Denver, Aurora, CO; ⁵University of North Carolina at Chapel Hill, Chapel Hill, NC. (Tracking ID #1933333)

BACKGROUND: Care coordination (e.g., information exchange, direct communication) between hospitalists and primary care providers (PCPs) is critical to successful care transitions from hospital to home, yet such coordination is frequently lacking. We sought to understand care coordination challenges and potential solutions from the perspective of both hospitalists and PCPs in North Carolina.

METHODS: We conducted a qualitative study with eight focus groups (4 hospitalist groups and four primary care groups, 59 total participants). The team recruited practices from diverse settings; six of the groups shared patients between inpatient and outpatient settings. Interview guides were based on the AHRQ Care Coordination Measurement Framework and included questions about care coordination, information exchange, follow up care, accountability, and medication management. Focus groups were digitally recorded, transcribed verbatim, and analyzed in ATLAS.ti. Two team members independently coded transcripts and resolved differences through discussion. The constant comparison method was used to evaluate differences between hospitalists and PCPs.

RESULTS: Overall, we found that hospitalists and PCPs encounter many similar challenges to care coordination, including: 1) lack of time, 2) difficulty reaching the other provider by telephone, 3) lack of personal relationships with the other provider, 4) lack of clarity about accountability for pending tests and home health orders, and 5) discrepancies between what medications patients are taking and should be taking at home. Specific concerns for hospitalists include: 1) uncertainty about whether PCPs receive information about patient hospitalizations, and 2) difficulty obtaining timely follow up appointments for patients, particularly for after-hours or weekend discharges. Specific concerns for PCPs include: 1) not knowing when their patients are hospitalized, 2) not having records available for post-hospitalization appointments, 3) feeling disrespected by hospitalists when medication changes are made for unclear reasons without their involvement, and 4) difficulty locating important information (e.g., pending tests) in discharge summaries. Hospitalists and PCPs identified common themes to achieving successful care coordination: 1) making extra efforts to coordinate care for “high risk” patient populations (e.g., patients on Warfarin needing INR checks), 2) improving direct telephone access to each other, 3) improved information exchange through sharing EMRs with improved integration of care transitions workflow into EMRs (e.g., enhanced messaging capabilities, scheduling capabilities for follow up), and 4) enhancing personal relationships with clinicians in the other setting.

CONCLUSIONS: Hospitalists and PCPs encounter many similar challenges to care coordination, yet have important experiential differences related to sending and receiving roles for hospital discharges. Challenges that were cited in one setting were often not experienced by clinicians in the other setting. Efforts to improve care coordination between hospitalists and PCPs should aim to understand perspectives of clinicians in each setting. Such solutions include: 1) enhancing direct communication between providers in different settings, 2) clearly defining accountability for tests and home health orders, and 3) building personal relationships between clinicians in each setting.

A HEALTH LITERACY INTERVENTION TO IMPROVE MEDICATION ADHERENCE USING MEDUCATION® TECHNOLOGYLeah L. Zullig; Felicia McCant; S. Dee Melnyk; Susanne Danus; Hayden bOSWORTH. Durham Veterans Affairs Medical Center, Durham, NC. (Tracking ID #1926044)

BACKGROUND: Many Americans are reliant upon medication management for chronic conditions. Medication non-adherence is rampant, leading to increased healthcare cost and worsened health outcomes. Low health literacy is a contributing factor to medication non-adherence. Our objective was to determine whether antihypertensive medication adherence could improve using a Meducation® technology health literacy intervention.

METHODS: We conducted a six-month feasibility study among patients with cardiovascular disease risk factors who received care from hospital-based primary care clinics associated with Durham Veterans Affairs Medical Center (VAMC). To enhance potential benefit and future generalizability, the intervention was integrated into clinics’ practice tailored to address needs of high risk, potentially non-adherent patients. All patients received the intervention, which included a personalized Meducation® calendar. We evaluated changes in medication adherence and clinical outcomes at 6months. Potential participants were initially identified through an electronic health record data pull. To be eligible for study inclusion, patients must have met all of the following criteria: enrolled in one of three primary care clinics affiliated with the Durham VAMC for at least 1year; had at least one visit to an affiliated primary care provider in the previous 12months; an outpatient diagnostic code for hypertension (ICD9 401.1, 401.9) diabetes (ICD9 250.x), and/or hypercholesterolemia (ICD9 272.x); prescribed metoprolol tartrate (beta blocker) and have poor medication adherence for the 12months prior to baseline (medication possession ratio <80%); and be prescribed more than nine active medications.

RESULTS: There was a 42% enrollment rate (n = 23). The sample was predominantly married (57%), African American (61%), and male (91%). Few participants (9%) completed less than 12years of school. Forty percent had low health literacy, defined as less than 9th grade reading level. The majority of participants were diagnosed with hypertension (91%), hyperlipidemia (87%), or diabetes (61%). At 3months self-reported medication adherence improved. There was a 45% decrease in participants who agreed or strongly agreed that they sometimes forget their heart medications and 58% fewer participants reported that they agreed or strongly agreed that they were ‘sometimes careless with how they take their medications’. At 6months, medication possession ratio improved nearly 13% and there were decreases in patients’ average systolic blood pressure (0.5mmHg), diastolic blood pressure (1.5mmHg), and body weight (3.6lb) (p > 0.05).

CONCLUSIONS: A health literacy intervention is a feasible mechanism to improve cardiovascular-related medication adherence and outcomes. In this sample, 9% of participants attained less than a high school level of education, yet 40% had health literacy levels below the 9th grade reading level. This is problematic as much health promotion and medication adherence information is targeted to a tenth grade reading level, which may be inaccessible to many patients. Interventions to improve adherence through health literacy have the potential to require relatively few resources to implement, making them particularly important tools to combat medication non-adherence. Future interventions may target and/or oversample patients at greatest risk for health literacy-related medication non-adherence.

A LONGITUDINAL ANALYSIS OF THE EFFECT OF TOTAL OPIOID ANALGESIC DOSE ON SUBSEQUENT USE AND DURATION OF HOSPITALIZATIONBarbara J. Turner^1,3; Benjamin R. Ehler²; Yuanyuan Liang^2,3. ¹University of Texas Health Science Center at San Antonio, San Antonio, TX; ²University of Texas Health Science Center at San Antonio, San Antonio, TX; ³University of Texas Health Science Center at San Antonio, San Antonio, TX. (Tracking ID #1937908)

BACKGROUND: High opioid analgesic (OA) doses for chronic non-cancer pain (CNCP) increase the risk of adverse events such as drug overdose and car accidents but it is not known whether these drugs are associated with an increased risk of future all cause hospitalization. Objective: To conduct a population-based study of the association of total dose of filled OA prescriptions (Rxes) in a baseline 6month interval and patient age with any hospitalization and duration of hospitalization in the next 6months.

METHODS: For persons enrolled >1year from 1/2009 to 7/2012 in Aetna’s national HMO (2.1 million), 278,975 persons aged 18 to 64years were selected if they filled ≥2 Rxes for non-injectable Schedule II or III OAs. From claims and enrollment files for these subjects, a database was developed including demographics, encounters, ICD-9-CM codes, and filled Rxes. For this analysis, we selected persons aged 45–64, no cancer diagnosis (except basal cell) within 6months of an OA Rx, and no OAs for opioid dependence. Follow-up after the first OA Rx was divided into 6-month intervals until 1/2012 (excluding incomplete intervals). The total morphine equivalent dose (TMED) was calculated from all filled OAs Rxes for each 6-month interval and categorized for analysis as none and approximate quartiles: NO risk (0days); LOW risk (1–190 TMED); MEDIUM risk (191–450 TMED); HIGH risk (451–1830 TMED); and VERY HIGH risk (>1830 TMED). Days covered by filled Rxes for other drugs used for CNCP—benzodiazepines, zolpidem, and antidepressants—were analyzed per 6-month interval. From ICD-9 codes, we created indicators for five categories of CNCP, mental health conditions, and substance abuse. Repeated measures logistic regression was used for hospitalization (Y/N) and repeated measures Poisson regression was used for inpatient days. All analyses adjust for patient demographic, clinical, and treatment factors. No significant interaction was observed for OA risk and age; therefore, we only report results for OA risk.

RESULTS: The cohort totaled 109,292 persons with a mean age of 53.7years (SD = 5.5), 54.7% female, and residence in South (46.9%), West (16.8%), Midwest (6%) and Northeast (30.3%). In the cohort’s 253,531 six-month follow-up intervals, the proportion hospitalized was 6% and median hospital days was 4 (IQR = 3–6). For hospitalization, the adjusted odds ratio [95% CI] for OA risk categories vs. NO risk were: VERY HIGH: 1.45 [1.38, 1.54]; HIGH: 1.15 [1.08, 1.21]; MEDIUM: 0.98 [0.93, 1.04]; LOW: 1.04 [0.98, 1.10]. For inpatient days, the incident rate ratio [95% CI] for OA risk vs. NO risk were: VERY HIGH: 1.47 [1.34, 1.62]; HIGH: 1.10 [1.01, 1.20]; MEDIUM: 0.87 [0.80, 0.95]; and LOW: 0.95 [0.87, 1.03].

CONCLUSIONS: High or very high total dose of filled OA Rxes in a baseline 6-month interval is associated with an increased risk of subsequent hospitalization and more inpatient days independently of age, pain-related comorbidities, and other prescribed medications commonly used for pain. Reasons for these hospitalizations need to be examined to inform interventions to address this high use of costly inpatient care.

A MIXED METHOD STUDY OF THE QUALITY OF WRITTEN FEEDBACK BY ATTENDINGS OF INTERNAL MEDICINE RESIDENTSCynthia Kay¹; Jeffrey L. Jackson^2,1; Wilkins Jackson³. ¹Medical College of Wisconsin, Milwaukee, WI; ²Zablocki VAMC, Milwaukee, WI; ³University of Wisconsin-Milwaukee, Milwaukee, WI. (Tracking ID #1920218)

BACKGROUND: Residents uniformly criticize the quality of feedback they receive. The purpose of this study was to evaluate the quality of feedback in an internal medicine residency program.

METHODS: Subjects for this retrospective analysis were Medical College of Wisconsin (MCW) internal medicine residents who completed residency training between 2004 and 2012. Over the 8years, there were 6603 evaluations by 334 attendings. These evaluations followed ACGME guidelines and assessed residents in six domains and included a comment box. From these 6603 written comments, we randomly selected 500 and coded them using grounded theory methods. An initial coding tree was developed by two raters (CK, JLJ). Once the tree was saturated, 25% of utterances were double coded with good inter-rater reliability (Spearman: 0.94). Utterances were further classified as high, moderate or low quality, also with good inter-rater reliability (Quadratic Kappa: 0.83). Focus groups were also conducted with medicine faculty and housestaff. First written comments were placed on notecards and the groups sorted the cards into high, moderate and low quality feedback. As they did this, they discussed the characteristics that merited this rating. The last few minutes of each focus group was spent reflecting and developing criteria for high, moderate and low quality feedback.

RESULTS: The majority of the attending’s 500 written comments were rated as moderate quality (65%); 22% were rated as high and 13% low. The most common type of feedback (29%) was nonspecific comments, often about the resident’s personality, 16% about patient care, 14% interpersonal communication, 7% medical knowledge, 6% professionalism, 4% each commented on practice based learning or systems based practice. The majority of written feedback comments (88%) were positive, 8% were negative and 4% were neutral; three ABIM domains were more likely to include negative comments (medical knowledge: OR: 3.5, 95% CI: 2.2–5.6; Practice Based Learning: OR: 2.5, 95% ci: 1.3–4.8, Systems Based Practice: OR: 4.6, 95% CI: 2.5–8.3). Nonspecific comments and comments on resident’s attitude or personality were much less likely to include negative comments than the other domains (nonspecific: OR: 0.22, 95% CI: 0.13–0.39, attitude/personality OR: 0.53, 95% CI: 0.34–0.82). Focus groups uniformly sorted quality into low, moderate and high quality feedback based on three criteria: specificity, quantitative ratings, actionable, Low quality feedback was nonspecific, moderate feedback had at least some specificity and high quality feedback included specificity plus either some aspect of quantification or was actionable. Despite the uniformity of the criteria espoused by focus groups, the inter-rater reliability between the focus groups was low, (quad kappa: 0.24).

CONCLUSIONS: Written feedback suffers from lack of balance, lack of specificity and lack of actionability. Much of it is focused on personality characteristics of the resident. While the focus groups developed nearly identical criteria for quality, agreement between the groups was low because the groups disagreed as to what constituted specific and actionable feedback. Residents are justified in expressing chagrin at the quality of feedback they are receiving.

A MIXED-METHODS EVALUATION OF A VA PATIENT-CENTERED MEDICAL HOME FOR SPECIALTY CAREMichael Ho²; Catherine Battaglia¹; Leah M. Haverhals²; Katherine M. Fagan²; Jackie G. Szarka³; Rebecca Alexander³; George Sayre³; Christian Helfrich³; Susan Kirsh⁵; David Au³; David Aron⁶; Julie Lowery⁴. ¹Eastern Colorado Health Care System, Denver, CO; ²Eastern Colorado Health Care System, Denver, CO; ³Puget Sound Health Care System, Seattle, WA; ⁴Ann Arbor Health Care System, Ann Arbor, MI; ⁵Case Western Reserve University, Cleveland, OH; ⁶Louis Stokes VA Medical Center, Cleveland, OH. (Tracking ID #1938016)

BACKGROUND: The Veterans Health Administration (VA) implemented a pilot program called Specialty Care Neighborhood (SCN) to align with the VA’s Patient Centered Medical Home to deliver comprehensive patient-centered care. SCN is a team-based specialty care model, in which teams of specialists engage in proactive and long term collaboration with primary care in order to provide coordinated, optimal care for Veterans with the greatest specialty care needs. SCNs were implemented at 25 VAs in 20 specialties including Heart Failure, Diabetes, Cancer, Pain, Orthopedic Surgery, and HIV. We completed an evaluation of the SCN initiative for the VA Office of Specialty Care (OSC).

METHODS: Findings are presented here from semi-structured interviews and staff surveys. A goal of the evaluation was to identify factors that differentiated four high from two low performing sites as determined by OSC from regular site reports on implementation progress. The evaluation consisted of 27 semi-structured interviews with Primary Care Providers, specialists, and key stakeholders (administrators, chiefs of staff, and facility directors) at the participating sites. Additionally, SCN staff from the six sites returned 39 surveys about their experience with SCN. Interview and survey data were coded using the Consolidated Framework for Implementation Research.

RESULTS: SCNs were viewed positively. Participants from all sites described improved professional relationships and communication as well as improved quality and quantity of specialty services. Interviewees felt improved relationships led to increased job satisfaction and engagement. Participants expressed frustration as to lack of clarity regarding how to implement SCN, and despite this barrier, pulled together to creatively implement SCNs best adapted to meet local needs. In structured ratings of implementation facilitators and barriers, High performing sites, in contrast to low performing sites, engaged leadership, utilized a thoughtful planning process that allowed for ample implementation time, and used systems redesign techniques to streamline processes. Additionally, there were factors from interview data that facilitated implementation at both high and low sites: 1) positive knowledge and beliefs about the program; 2) obtaining patient and provider feedback; 3) recognizing patient needs; 4) active champions; and 5) ability to adapt the program easily. Common barriers included 1) problems with hiring staff, 2) lack of commitment to long-term staff funding, 3) lack of administrative support, 4) insufficient materials for facilitating implementation, and 5) coordination and communication challenges among different providers and services. Interview data and staff survey data were largely congruent. However, survey participants reported high leadership engagement despite interview data suggesting this was less positive among low performing sites.

CONCLUSIONS: The findings were used to identify areas on which to focus attention for improving dissemination of the program. Recommendations were based on factors that distinguished high from low performing sites and included recommendations to overcome barriers: 1) Provide opportunities for sharing best practices across sites; 2) Encourage innovation through site-based flexibility and individualization while limiting micromanaging; 3) Facilitate access to systems redesign expertise; 4) Minimize complexity by utilizing programs in place; and 5) Begin planning process early and involve all key stakeholders.

A MODEL TO PREDICT RISK OF CERVICAL ABNORMALITIES IN PRIMARY CAREMichael B. Rothberg; Bo Hu; Sarah Schramm; Kelly Nottingham; Glen Taksler; Xian W. Jin; Laura Lipold; Andrea L. Sikon. Cleveland Clinic, Cleveland, OH. (Tracking ID #1937479)

BACKGROUND: Cervical cancer screening tests are highly sensitive but have low specificity, resulting in high false positive rates among low-risk women. Although there are known risk factors for cervical cancer, there are no validated tools for assessing risk in primary care. Using data from electronic health records, we derived and validated a risk prediction model for cervical intraepithelial neoplasia grade 2 or higher (CIN2+) lesions.

METHODS: Our dataset included information on all women aged ≥30years who underwent cervical cancer screening with cytology and human papilloma virus (HPV) testing at 33 physician practices associated with the Cleveland Clinic between 2004 and 2010. Patient risk factors at the time of screening (age, race, marital status, insurance type, smoking history, median income based on zip code and physician practice) were determined by reviewing the electronic health record. Pathology results were obtained from the Cleveland Clinic’s pathology database (for patients who underwent colposcopy with biopsy) or by imputation (for patients with abnormal cytology who did not undergo colposcopy). Imputation was based on observed associations between patient risk factors, HPV and cytology results and CIN2+ status among women who underwent colposcopy. Using CIN2+ as our outcome, we then derived a hierarchical generalized linear model with a logit link that included all patient risk factors available before screening (i.e. not HPV or cytology results) as independent variables. We then stratified patients into tertiles of low, medium and high predicted risk of CIN2+. We then compared the actual rate of CIN2+ (either observed or imputed) in the low-risk tertile to an implied threshold value, which was based on combinations of HPV and cytology results for which colposcopy is typically not recommended. The threshold was 0.20%, which is equivalent to the rate of CIN2+ among patients with low risk ASCUS and negative HPV. The final model was validated using 100 bootstrap iterations for the entire population and noting the proportion of low risk patients who had CIN2+.

RESULTS: The dataset included 99,388 women. Of these, 574 had pathology findings of CIN2+ and 175 had CIN2+ assigned by imputation (overall rate of CIN2+=0.75%). The final multivariable model had a c-statistic of 0.73. All factors but race were associated with CIN2+: older age (OR 0.69 per 5years, 95% CI 0.66, 0.73), Medicare vs. private insurance (OR 0.51, 95% CI 0.34, 0.75) and higher income (OR 0.97 per $5K, CI 0.94, 0.99) were negatively associated, while being single vs. married (OR 1.50, 95% CI 1.24, 1.82) and smoking (OR 1.66, 95% CI 1.35, 2.04) were positively associated. Tertiles of predicted risk were 0.20%, 0.53%, and 1.51%, with observed rates of CIN2+ over the same tertiles of 0.19%, 0.52%, and 1.53%. Therefore, before cervical cancer screening, women in the lowest risk tertile had observed CIN2+ risk slightly below that of women with ASCUS and HPV negative (0.20%). In 100 bootstrap iterations, the observed risk in the lowest tertile ranged from 0.13 to 0.25% (IQR 0.18% to 0.22%).

CONCLUSIONS: A multivariable model based on data derived from the electronic health record was able to risk stratify women across an 8-fold gradient of risk for cervical abnormalities and to identify a large group of women who are at very low risk for cervical cancer and may not require screening. After further validation, use of a similar model could enable more targeted cervical cancer screening.

A NATIONAL SURVEY OF PATIENT CENTERED MEDICAL HOME INITIATIVESSamuel T. Edwards^1,2; Asaf Bitton^3,4; Johan S. Hong⁴; Bruce E. Landon^4,5. ¹Boston VA Healthcare System, Boston, MA; ²Harvard Medical School, Boston, MA; ³Brigham and Women’s Hospital, Boston, MA; ⁴Harvard Medical School, Boston, MA; ⁵Beth Israel Deaconess Medical Center, Boston, MA. (Tracking ID #1932105)

BACKGROUND: The reorganization and improvement of primary care through the implementation of the patient-centered medical home (PCMH) is central to healthcare delivery system reform efforts. PCMH initiatives, often organized by health plans, states or multi-stakeholder groups, seek to create the appropriate context to promote transformation of participating primary care practices into medical homes by building learning collaboratives, providing practice coaching, and reforming physician payment to support medical home services. Although there has been expansion of PCMH activity across the country, the current national scope of PCMH initiatives remains undefined. To address this gap, we conducted a nationwide survey of PCMH initiatives. We describe their size, scope, organization and payment methodology and assess their approach to practice transformation and program evaluation.

METHODS: We performed a cross-sectional survey of all PCMH initiatives nationally that included payment reform as an operational feature. We identified initiatives for the survey from existing databases supplemented by formal literature review, internet searches, and by contacting known experts, state Medicaid directors, and representatives from all major national health plans. We included initiatives that were active as of February 1st, 2013 but excluded programs directed solely toward specific medical conditions or serious mental illness. The survey included descriptive questions including initiative start and end dates, numbers of participating practices, physicians and patients, and identity of key stakeholders. We asked about the selection process for practices to enter the initiative, the use of formal PCMH recognition, and payment methods. Finally, we asked about how initiatives facilitated practice transformation and plans for program evaluation. The survey was fielded from February 1st 2013–October 31st 2013 via web or telephone. Respondents were physicians, senior administrators, or senior health policy experts from each initiative. We present descriptive statistics to compare initiatives sponsored by commercial health plans, state Medicaid programs, and multiple payers.

RESULTS: Of 119 identified initiatives, we received 114 responses (96% response rate). Forty-one programs were small commercial single payer initiatives, 29 were large single payer initiatives, 23 were Medicaid only and 21 were multi-payer. In total, the programs included 63,011 providers caring for 20,389,327 patients nationally. Small commercial programs had a median of 3,896 patients and typically included a small number of practices (median = 4). Large commercial single payer initiatives included a median of 160,000 patients, often covering most or all practices in a specific region or an entire state (median 105 practices). Medicaid initiatives were typically large statewide programs (median 224,040 patients) run directly by state Medicaid agencies. Multi-payer initiatives included a median of six health plans (IQR 4–7) with a median of 187,343 patients, and 68 practices. These initiatives included commercial health plans, Medicare and Medicaid. Formal PCMH recognition was required by 69% of initiatives. Nearly half of initiatives use practice consultants (46%) or learning collaboratives (48.7%), but this was more common in multi-payer initiatives (90.5% and 95.2%, respectively). Nearly all initiatives had a formal evaluation plan (92%). A combination of fee for service payments (FFS) for office visits, a per-member-per month care management fee (PMPM), and pay for performance bonuses was the most common payment structure (55% of programs). Median PMPM was $4.90 (range $0.25–$60.00), but was higher in multi-payer initiatives (median $7.00). Since 2009, fewer initiatives are time-limited pilots, shared savings programs are more commonly used, and more initiatives are performing formal evaluations (table).

CONCLUSIONS: In this national survey of PCMH initiatives, we demonstrate there has been a large expansion of PCMH activity over the last several years, with 114 initiatives representing over 60,000 providers and nearly 21 million patients in 2013. Initiatives are on average larger, feature more robust transformation efforts, and have higher care management fees. This expansion of PCMH initiatives suggests that the PCMH will continue to play an important role in national delivery reform efforts.

Comparison of PCMH initiatives in 2009 and 2013

A PROFILE OF HOSPITALIZED SMOKERS AND THE CARE THEY RECEIVEEllie Grossman¹; Alissa R. Link²; Erin Rogers^2,3; Binhuan Wang²; Scott Sherman^2,3. ¹NYU School of Medicine, New York, NY; ²NYU School of Medicine, New York, NY; ³VA New York Harbor Healthcare System, New York, NY. (Tracking ID #1939163)

BACKGROUND: Identification and treatment of patients who smoke is a priority treatment area for hospitals. Hospitalization can be a ‘teachable moment’ for patients contemplating behavior change, and they may be newly motivated by health consequences. Evidence suggests that appropriate treatment should begin during an inpatient stay and continue for at least four weeks post-discharge. To optimize success, treatment should include both counseling and medications. Patients with mental illness tend to smoke at higher rates than those without, and tend to die earlier of cardiovascular disease, but there is a paucity of evidence about the state of treatment offered to these patients. We aimed to describe the care offered to hospitalized smokers at two urban safety-net hospitals.

METHODS: As part of a randomized controlled trial of smoking cessation interventions focused on the post-discharge time period, we assessed the care for smokers hospitalized at two large public hospitals in New York City. Adult patients were eligible for study inclusion if they: smoked at least one puff of a cigarette in the past 30days; spoke English, Spanish, or Mandarin; were not incarcerated or in police custody; were not pregnant or breastfeeding, and had a U.S. phone number. At the time of discharge, study staff reviewed participants’ electronic medical records (EMR) for information about type of inpatient unit, history of medical or psychiatric illnesses, and tobacco treatment received during hospitalization and on discharge.

RESULTS: From July 2011 to December 2013, we enrolled 1486 study participants. 511 (35%) were on psychiatric inpatient units and 934 (65%) were on medical/surgical units (n = 41 missing). Mean age was 47.9years, 79% were male. Of these patients, 17% had EMR documentation of coronary artery disease, 12% chronic obstructive pulmonary disease, 14% asthma, 5% cerebrovascular disease, 4% congestive heart failure, 55% significant alcohol or drug use history, and 10% schizophrenia or schizoaffective disorder. We found EMR documentation that smoking status was assessed at admission in 1444 (97%) and bedside counseling was provided in 1294 (87%). Of those that received counseling, the EMR indicated that 1015 received it from nurses and 145 received it from physicians (for the rest the personnel was not specified). For 647 patients (44%), physicians ordered smoking cessation pharmacotherapy during hospitalization (36% NRT patch, 3% NRT gum, 2% NRT lozenge, 1% other). On discharge, 274 patients (18%) were prescribed cessation medication. Patients on inpatient psychiatric units were more likely to be counseled by their physician regarding smoking cessation (odds ratio (OR) 1.93, 95% CI 1.20–3.09) and were more likely to be prescribed cessation medication in hospital (OR 3.21, 95% CI 2.51–4.11) than were patients on non-psychiatric units. There was no association between type of inpatient service (psychiatric vs. non-psychiatric) and likelihood of prescription of smoking cessation medication at discharge.

CONCLUSIONS: At urban safety-net hospitals, patients who smoke have high rates of mental illness and substance use disorders. Although assessment and counseling are provided at relatively high rates, there is much room for improvement in prescribing smoking cessation pharmacotherapy in the hospital and on discharge.

A QUALITATIVE ANALYSIS OF HOSPITAL LEADERS OPINIONS ABOUT PUBLICLY REPORTED MEASURES OF HEALTH CARE QUALITYSarah L. Goff^1,2; Tara Lagu^2,1; Amy S. Shatz²; Penelope S. Pekow²; Peter K. Lindenauer^2,1. ¹Baystate Medical Center/Tufts University School of Medicine, Springfield, MA; ²Baystate Medical Center, Springfield, MA. (Tracking ID #1937659)

BACKGROUND: Public reports of hospitals’ performance on quality measures are intended to both encourage hospital leaders to stimulate improvement activities and to enable consumers to make more informed decisions about where to seek care. How effective public reporting is at fostering quality improvement may be influenced by whether and to what extent hospital leaders find the measures valid and meaningful. We sought to qualitatively describe hospital leaders’ views regarding quality measures reported on the Centers for Medicare and Medicaid Services’ Hospital Compare website.

METHODS: We mailed or e-mailed a 21-item questionnaire to the chief executive officer and the senior executive responsible for quality at a stratified random sample of U.S. hospitals. Stratification was based upon performance on selected quality measures to achieve a balance of high, average and low performers. While the quantitative portion of the study assessed attitudes regarding publicly reported quality measures, opinions regarding quality measures were elicited using an open-ended prompt at the end of the survey, “Please share your additional thoughts about publicly reported measures of healthcare quality, including strengths or weaknesses of current measures, ideas for new measures, etc.” We qualitatively analyzed responses to this prompt, identifying the presence and frequency of pertinent themes in an iterative process.

RESULTS: Leaders (n = 480) from 380 of the 630 hospitals surveyed (60%) completed the questionnaire. Of those who completed the questionnaire, 146 leaders from 131 hospitals responded to the open-ended prompt. The majority of respondents to the open-ended prompt were chief executive officers (27%), chief quality officers or vice presidents of quality (17%) or chief medical officers (19%). Half of the responses to the prompt came from leaders at urban hospitals and 68% were from leaders at non-teaching hospitals. Most respondents were from hospitals that had fewer than 200 beds (41%), with 23% from hospitals with more than 400 beds. Forty-one percent of the respondents’ hospitals had performance scores higher than the national average and 27% scored lower than the national average. Major themes identified included: 1) Positive views of the value of quality measures (12.5%, e.g., measuring quality can stimulate change and enhance patient engagement); 2) Concerns regarding the validity and utility quality measures (63%, e.g., measures fail to measure what is important, inadequate risk-adjustment); 3) Challenges to improving quality scores (e.g., inability to impact factors affecting care outside of the hospital, such as patient adherence to prescribed medication); and 4) Recommendations for improving measures (e.g., compare similar hospitals to each other). Many of the positive views expressed were general support for the concept of measuring quality: “Oversight and comparison with peers/standards is necessary and beneficial.” Criticisms of quality measures included concerns about whether doctors can actually influence change in the entities measured: “…the scores for which the doctors are held accountable are subject to profound influence by others besides the doctor to whom the score is attributed.” Challenges described included lack of resources: “Resources are stretched to the limit with meaningful use and data reporting!” Recommendations for improvement in quality measures included an emphasis on the continuous learning cycle, suggesting new targets for quality improvement should be identified as older measures reach acceptable levels of compliance: “…care measures should evolve as compliance becomes ingrained thus continuing the improvement cycle.”

CONCLUSIONS: Among hospital leaders who responded to an open-ended prompt eliciting their views about quality measures some expressed support for the concept of measuring quality, while the majority offered criticisms concerning the validity and utility of current quality measures or described challenges to improving quality. As current quality measures are reassessed and new measures developed, the concerns of those responsible for implementing system changes to address quality should be taken into account.

A QUALITATIVE ANALYSIS OF PATIENT FACTORS ASSOCIATED WITH HOSPITAL READMISSIONColleen M. Rafferty; Tyler Sechrist; Heather Stuckey; Nicole Swallow; Cynthia H. Chuang. Penn State Hershey Medical Center, Hershey, PA. (Tracking ID #1936483)

BACKGROUND: Given the focus on hospital readmissions as a quality indicator, health care organizations nation-wide are interested in finding ways to reduce hospital readmissions. Existing readmission prediction models have relied mainly on clinical and administrative data. A weakness in these methods is that they are not informed by the patient perspective. Exploring patient experiences with readmission may broaden our understanding of what leads to readmission, thereby facilitating a more patient-focused approach to identifying high-risk patients, and enabling institutions to allocate resources to the highest risk patients and to the most cost-effective and efficacious interventions. We conducted qualitative interviews with readmitted patients to identify more precisely the patient-centered factors associated with readmission.

METHODS: We conducted semi-structured interviews with a convenience sample of hospitalized patients readmitted to the Internal Medicine service within 30days of discharge to home from a previous hospitalization at our institution, a 500-bed tertiary care academic medical center. Patients were eligible if they received their primary medical care within our medical system and were readmitted from June–August 2011 (n = 19). Open-ended interview questions addressed the patients’ perspectives on the medical care they received during the previous admission, access to outpatient care, the patient’s knowledge of his/her medical conditions and treatments, ability to provide self- care, and social support available to them. We also encouraged the patient to discuss from his/her perspective what could have been done differently to help prevent the readmission. The contents of the interviews were transcribed and a thematic analysis was conducted. We also conducted a medical chart review to determine if the patient had a scheduled follow-up appointment that they attended.

RESULTS: We found that the majority of patients had a limited understanding of their medical conditions and their medications. Many patients relied heavily on an advocate (family member or friend) to assist them in their disease management. The vast majority of patients had identified advocates, but we found that not all advocates were effective in providing the needed assistance. Also, most patients did not view their health as under their control, but identified an external locus of control (i.e. the doctors, the disease process, God). All patients identified a primary care physician, and all had been scheduled for follow up appointments within 5–24day after discharge, but 9 of 19 patients were readmitted prior to their scheduled appointment (8/19 readmitted within 7day, 10/19 within 10day). The majority of patients did not feel their readmission was preventable, but attributed it to the disease process, or an issue that was not viewed by the patient as avoidable or predictable.

CONCLUSIONS: These findings suggest potential patient-centered factors that may be used to help identify patients at high risk for readmission: insufficient understanding of medical condition or medications, lack of an effective advocate, a belief that one’s health is not under the one’s own control, and lack of timely access to outpatient follow up care. Early assessment of these risk factors may help target appropriate interventions to high risk patients to help prevent hospital readmission.

A QUALITATIVE ASSESSMENT OF CPOE AND THE EFFECTS OF VARIATION IN DRUG NAME DISPLAYAlexandra Robertson¹; Arbor J. Quist¹; Thu-Trang Thach¹; Lynn A. Volk²; Adam Wright^1,3; Shobha Phansalkar^1,4; Sarah P. Slight^1,5; David W. Bates^1,3; Gordon D. Schiff^1,3. ¹Brigham and Women’s Hospital, Boston, MA; ²Partners HealthCare Systems, Inc., Wellesley, MA; ³Harvard Medical School, Boston, MA; ⁴Wolters Kluwer Health, Indianapolis, IN; ⁵The University of Durham, Durham, United Kingdom. (Tracking ID #1938546)

BACKGROUND: Previous work suggests that prescribing errors account for the highest proportion of medication errors (Lisby, 2010) and that one in every four reported in the U.S. can be linked to drug name confusion (Lambert, 1999). Computerized provider order entry (CPOE) has been widely proposed, implemented, and shown to reduce errors associated with traditional, handwritten, medication prescribing. However, CPOE has not fully lived up to its predicted benefits and many attribute these shortfalls to suboptimal design of systems that are “neither interoperable nor easy to use” (Kellerman, 2013, Health Affairs). Given recent calls for CPOE performance testing post-implementation (Thompson, 2010 & Bates, 2013) and the proportion of medication errors which occur during the prescribing phase, we undertook a study to assess drug name design and display issues in CPOE including: look-alike-sound-alike pairs, similar name adjacency errors, font, and visual display to better understand their potential to contribute to medication errors.

METHODS: As part of an FDA Task Order, this qualitative study assessed 10 leading vendor and home-grown CPOE systems in six major U.S. academic medical centers via remote walk-throughs and on-site visits. To standardize assessment, the team developed and iteratively refined a CPOE Assessment Tool (CAT) relating to 18 elements defined by the FDA to systematically test the potential for drug name/identification errors across various stages of the medication ordering process. This tool incorporated specific medication ordering, review, and deletion scenarios used to guide system assessments with a series of “regular” and “expert” users of these systems. Outputs from these evaluations were recorded from each system using screen shots, video and audio transcripts of user and designer comments. Recordings were transcribed, analyzed, and used to inform the content of onsite visits to each medical center to obtain additional system information from medical leadership, IT experts, users, and pharmacists. Inter- and intra-system characteristics were analyzed descriptively for patterns, themes, and variations.

RESULTS: We identified a variety of ways drug names are displayed in CPOE systems that could potentially contribute to medication errors. Recurrent issues and themes that emerged included: (a) Brand versus generic name: Presentation of brand and generic names varied both across and within systems depending on task functionality. In nine of the ten CPOEs assessed, for example, one brand name search (Plavix) yielded results displaying the drug in eight different ways. Some systems used all caps, some only initial caps, some used both. Some were inconsistent in the drug name searched and drug name returned. One system displayed two columns which listed both the brand and generic names, others included only one, and others presented one or the other parenthetically, based on the search term. One system allowed ordering by generic name only. If one searched by brand name, the medication did not appear; we observed end user frustrations resulting from inability to search for and find desired drugs searched non-generically. Systems also used varied methods to display brand and generic options to steer prescribing towards preferred products (typically the generic). (b) Combination products: In most systems, searching for a combination drug often yielded the brand name with the ingredients listed parenthetically. We rarely saw systems provide explanations of ingredient strengths. In some systems, searching the brand name yielded only the brand name, without its generic ingredients. (c) Drug-name modifiers: Across systems, drug modifiers seemed to have some consistency in both representation and display. Drug modifiers appeared primarily at the end of the drug name, in all caps, and were abbreviated (e.g. XR, DS, etc.). They were rarely truncated. (d) Extraneous additions to drug name fields: Across systems, we observed several instances of non-systematically populating (jerry rigging) name fields with additional information ranging from drug indication to which facility the drug could be ordered from. (e) Font, text size, capitalization: These all varied widely with no standardized approaches seen both across and often within systems.

CONCLUSIONS: There is little consistency in the ways that leading CPOE systems search for and present drug names. This lack of standardization was seen both across systems and often within a single system or institution’s CPOE systems. Brand vs. generic, combination products and drug name modifiers were particularly inconsistent, each presenting the potential for user frustration, confusion, and erroneous orders.

A QUALITATIVE STUDY EXPLORING THE VULNERABILITIES OF COMPUTERIZED PHYSICIAN ORDER ENTRY SYSTEMSSarah P. Slight^1,2; Tewodros Eguale^1,3; Mary Amato^1,4; Andrew C. Seger⁴; Diana L. Whitney⁵; David W. Bates^1,6; Gordon D. Schiff^1,6. ¹Brigham and Women’s Hospital, Boston, MA; ²Durham University, Stockton on Tees, United Kingdom; ³McGill University, Montreal, QC, Canada; ⁴MCPHS University, Boston, MA; ⁵Baylor College of Medicine, Houston, TX; ⁶Harvard Medical School, Boston, MA. (Tracking ID #1935926)

BACKGROUND: Computerized Physician Order Entry (CPOE) systems can prevent medication errors in both inpatient and outpatient settings. Depending on how they are designed, however, they can fail to optimally prevent various prescribing errors or introduce new errors. The Institute of Medicine report Health IT and Patient Safety: Building Safer Systems for Better Care recommended that specific examples of potentially unsafe processes and risk-enhancing interfaces be identified and shared amongst the health IT community. This study aims to test the vulnerabilities of a wide range of CPOE systems to different types of medication errors, and to develop a more comprehensive understanding of how CPOE human factors design could be improved.

METHODS: As part of a National Patient Safety Foundation-funded project, we examined a range of leading vendor and home grown CPOE systems (e.g., Cerner; Epic; Medi-tech; LMR; BICS; GE Centricity) in diverse organizations in United States and Canada. Typical users at each of 16 sites were asked to enter 13 different orders on test patients based on scenarios of previously reported CPOE errors. Users were encouraged to use both usual practice and, where-needed, workarounds to enter the erroneous orders, as well as reflect on their overall knowledge and experience of using their system. A research pharmacist and research assistant independently observed test users enter each order and rated the ease or difficulty of these entries using standardized operational definitions.

RESULTS: Ease of entry of erroneous orders and the generation of alert warnings in different CPOE systems was highly variable and appeared to depend on a number of factors including how the order information was entered (i.e., in a structured or unstructured way); whether a specific alert functionality (e.g., duplicate-drug checking) was operational in the system; and which drugs or drug combinations were included in the clinical decision support algorithms. Test users found the wording of many of the alert warnings confusing and expressed frustrations with the way irrelevant warnings appeared on the same screen as those more relevant to the current order. The timing of alert warnings also differed across CPOE systems, with dangerous drug-drug interaction warnings displayed, for example, only after both Imdur® (isosorbide mononitrate) and Revatio® (sildenafil) had been entered and the order signed off in two CPOE systems. Alert warnings also varied in their level of severity in different systems even within the same institution (e.g., test user was presented with a hard stop alert warning in the inpatient system after entering Lovenox® (enoxaparin sodium injection) 40mg subcutaneous daily followed by a second order for Lovenox® 100mg subcutaneous twice daily, whereas the equivalent outpatient system produced an alert for the same order that was easily overridden). Test users commented on how there may be situations where they needed to prescribe the same drug twice, requiring workarounds such as entering the brand name of the drug for the morning dose and the generic name for the evening dose, to avoid getting duplicate drug alert warnings.

CONCLUSIONS: We found a high degree of variability of vulnerability in ordering and alerting between different CPOE systems. Detailed qualitative analysis of both the observed CPOE functionality and users think-aloud comments provided rich insights into the ways both systems and users were susceptible to ordering errors. System developers and users need to be aware of these vulnerabilities and build in protection strategies to re-engineer their systems to ensure safer prescribing.

A RANDOMIZED TRIAL OF A COMMUNITY HEALTH WORKER LED INTERVENTION TO IMPROVE DIABETES INTERMEDIATE OUTCOMES AMONG LATINOS PATIENTS WITH POORLY CONTROLLED DIABETESOlveen Carrasquillo; Yisel Alonzo; Cynthia Lebron; Natalie Ferras; Ernesto Reyes-Arrechea; Hua Li; Sonjia Kenya. University of Miami, Miami, FL. (Tracking ID #1936786)

BACKGROUND: Diabetes disproportionately impacts Latinos. Community Health Workers (CHWSs) are one approach that may improve health outcomes in this population. However, evidence from rigorous randomized studies is limited.

METHODS: The Miami Health Heart Initiative examined the impact of a comprehensive community health worker (CHW) intervention on diabetes intermediate outcomes of systolic blood pressure (SBP), lipids (LDL) and Hemoglobin A1C (HbA1c). We recruited 300 Latino patients with poorly controlled diabetes (HbA1c>=8.0%) ages 35–65 from the primary care clinics of Miami-Dade county’s public hospital. Subjects were randomized to usual health care or a comprehensive structured CHW 1year intervention consisting of home visits, phone calls and group education sessions. The intervention included patient navigation as well as assistance with social and non-medical needs. An RA blinded to group assignment conducted initial and follow-up evaluations at 12months. We used linear mixed models to statistically test for the impact of our intervention on outcomes using intention to treat analyses.

RESULTS: The mean age of our patients was 56 + 7years, 55% were female and mean BMI was 31.6 + 7.4kg/m2. Over half (55%) were sedentary (IPAQ) and median daily fruit and vegetable was two (BRFSS). Cubans made up 29% of our sample with no other Hispanic ethnic subgroup representing over 15% of the sample; 47% of respondents scored the lowest possible acculturation score (Marin-Marin). Intervention patients received a median of five home CHW visits and 25 phone calls; 84% of intervention group participants had at least 12 CHW contacts over the course of the year. Participation at group session was more skewed with 52% of participants not attending any sessions. Of those that attended at least one session, average attendance was four sessions per participant. At 1year the retention rate was 72% and similar in both arms. Data on outcomes by group is shown in Table 1.

CONCLUSIONS: In a heterogeneous Latino population with poorly controlled diabetes, a rigorous CHW intervention resulted in statistically and clinically meaningful changes in HbAIC but not in blood pressure or lipids. The impact of the intervention on the latter two outcomes may have been limited by the fact nearly half of patients were at the SBP and LDL targets prior to the intervention.

Figure 1: Baseline and One Year Outcomes in MHHI

*p = 0.01 for HbA1C NS for SBP and LDL

A REDUCTION IN TELEMETRY ORDER DURATION DOES NOT CHANGE TELEMETRY ORDERING FREQUENCY AND REDUCES UTILIZATIONJoel Boggan^1,2; Ann Marie Navar-Boggan⁴; Ryan D. Schulteis^3,2; David Simel^6,5. ¹Durham Veterans Affairs Medical Center, Durham, NC; ²Duke University Health System, Durham, NC; ³Durham Veterans Affairs Medical Center, Durham, NC; ⁴Duke University Health System, Durham, NC; ⁵Duke University Health System, Durham, NC; ⁶Durham Veterans Affairs Medical Center, Durham, NC. (Tracking ID #1940176)

BACKGROUND: American Heart Association guidelines for appropriate use of telemetry have not decreased inappropriate use of in-hospital cardiac monitoring. We evaluated how changing the default order set for telemetry affected overall telemetry utilization.

METHODS: The computerized ordering system at the Durham VA Medical Center was changed so that the maximum number of hours of cardiac monitoring was reduced by 33%, from 72 to 48h, before requiring a renewal order. No other intervention or general education about telemetry appropriateness was done. We compared the frequency of orders/week for telemetry and duration of cardiac monitoring for 12weeks before through 6weeks after this intervention. To assess the safety of the intervention, we measured the number of rapid response team (RRT) or code events occurring at any time during the hospital stay of general medical or general surgical patients. We compared event rates on telemetry before and after the intervention.

RESULTS: Over a 4-month period, there were 1051 telemetry orders placed for general medical or surgical patients before the intervention and 450 after. The average number of orders/week did not differ significantly before vs after the reduction in telemetry order duration (88 +/− 10 vs. 75 +/− 22 orders/week, p = 0.11). The mean duration of telemetry was reduced by 32%, which was the same percent reduction required by the change in policy (mean reduction in telemetry time decreased from 64 +/− 3.1h/patient to 44 +/− 1.2h/patient, p < 0.01). The rate of code events and rapid response team events in hospitalized general medical or surgical patients decreased from 2.5 events/week to 2.2 events/week. Only seven RRT/code events occurred in patients on telemetry prior to the intervention, while one RRT/code event occurred in a patient on telemetry after the intervention (p = 0.28).

CONCLUSIONS: A simple change in the default maximum number of hours a provider can prescribe cardiac monitoring before requiring a renewal order led to a dramatic decrease in telemetry utilization. Although the number of orders for patients to receive cardiac monitoring was unchanged, the duration of cardiac monitoring dropped substantially. Even with less monitoring, we found no increase in the rate of adverse events. Similar reductions at other medical centers may reduce telemetry utilization, technologist and nurse staffing requirements, and cost while preserving patient safety.

A RESIDENT INITIATIVE TO INCREASE USE OF NALTREXONE FOR ALCOHOL USE DISORDER AT A SAFETY NET HOSPITALMaria E. Otto¹; Jeffrey K. Hom¹; Shilpa M. Shah¹; Natalie C. Young¹; Weston S. Fisher²; Edgar Pierluissi¹. ¹University of California San Francisco, San Francisco, CA; ²University of California San Francisco, San Francisco, CA. (Tracking ID #1939058)

BACKGROUND: Alcohol use disorder (AUD) is extremely common at San Francisco General Hospital (SFGH), with nearly one in four inpatients on the Medicine service affected. These patients frequently present to the Emergency Department (ED) with alcohol-related falls, seizures, and symptoms of withdrawal. Their 30-day readmission rate is 16%, above the 13% for the remainder of the hospital. Hospitalization provides a window of opportunity to help this vulnerable population, as patients are often motivated to make changes following an alcohol-related illness but are frequently without primary care doctors. Despite the effectiveness of naltrexone, an inexpensive, FDA-approved medication for treatment of AUD, it has been underutilized. A pilot intervention in 2011 at SFGH increased prescriptions of naltrexone to eligible patients at discharge, but was not sustained.

METHODS: A resident-led, quality improvement (QI) project was implemented in February 2013 to 1) sustainably increase the prescribing of naltrexone to eligible inpatients with AUD upon discharge and 2) to measure the effect of the intervention on ED visits and readmissions within 30days. The intervention 1) identified inpatients with AUD via the electronic handoff system, 2) educated providers about naltrexone, 3) prompted providers to prescribe naltrexone to eligible patients at discharge planning rounds, and 4) utilized principles of QI to continuously evaluate and improve the intervention. Data were collected via retrospective chart review to identify patients with AUD and determine eligibility for naltrexone. Among patients with AUD, number of ED visits and readmissions within 30-days of discharge were also obtained.

RESULTS: Among 118 patients with AUD identified between March 1 and April 15, 2013, 43% (n = 51) were eligible for naltrexone. Following the initial intervention, naltrexone offering increased from 2 to 58% of eligible patients. This effect was sustained through continuous cycles of improvement, with 67% of patients offered it most recently. Compared to patients not prescribed naltrexone (n = 31), patients prescribed the medication (n = 20) had 30-day readmission rates and hospital visits that were 74% and 88% lower, respectively.

CONCLUSIONS: These results are promising and suggest that an intervention to systematically identify and offer treatment to patients with AUD is important in caring for our patient population and decreasing hospital utilization. Additionally, this QI initiative demonstrates that a deep-rooted problem can be addressed with the commitment of an interdisciplinary team, which can include a rotating group of trainees. By assessing the effect of this intervention on hospital utilization, the project shows that measuring true outcomes, rather than just processes, can help build a business case for sustainability. Next steps include 1) expanding the intervention to other hospital services, 2) ensuring the safe transition of patients on naltrexone from the hospital setting to their outpatient providers, and 3) collaborating with hospital leadership to improve systems to identify and offer treatment to all patients with AUD.

A ROLLING, INTEGRATED EVIDENCED-BASED MEDICINE CURRICULUM’S IMPACT ON EVIDENCE-BASED PRACTICE SKILLS, KNOWLEDGE, BEHAVIOR AND ATTITUDES ON INTERNAL MEDICINE STUDENTS AND RESIDENTS AT THE WASHINGTON, DC VETERANS AFFAIRS MEDICAL CENTERMatthew Tuck; Samantha Mcintosh; David Maron. Department of Veterans Affairs Medical Center, Washington, DC. (Tracking ID #1941446)

BACKGROUND: Teaching evidence-based medicine (EBM) skills to students and residents has become an imperative for medical education. The AAMC made recommendations to incorporate problem-based clinical skills in medical schools’ curricula. Similarly, the Next Accreditation System of the ACGME requires that residency programs train and evaluate residents in practice-based learning and improvement. Developing effective techniques for integrating these practices across the years of medical education remains a challenge for medical schools and residency programs.

METHODS: We developed a rolling EBM curriculum for medical students and residents rotating through the Medical Service of the Washington, DC Veterans Affairs Medical Center (VAMC) that provides repeated opportunities for students and residents to learn, practice, and evaluate their skills within the context of real-time patient care. The curriculum is integrated into morning report over the course of a month-long rotation. Using a patient case on the wards, learners are guided through the process of formulating a PICO question, searching the literature to find an article that answers the question, appraising the validity of the article and applying the literature to the case. The aim of the study was to evaluate the curriculum on learners’ self-reported patient care practices, attitudes and knowledge using validated rubrics. We hypothesized that exposure to the curriculum would improve learners’ EBM-related patient care practices, attitudes and knowledge. Using a before-and-after design, consented learners were given questionnaires. We used validated rubrics on attitudes and behavior as well as a rubric to assess EBM knowledge (the so-called Berlin Questionnaire) at the beginning and end of their rotation at the VAMC. Incentives were provided for completion of the questionnaires. Means, standard deviations, percentages and univariate paired t-tests were calculated for pre- and post-exposure scores. The research was funded by a Medical Education Research Grant awarded by George Washington University.

RESULTS: 231 residents and students were exposed to the curriculum. Sixty-three learners completed the pre-curriculum questionnaire and 21 completed both the pre- and post-curriculum questionnaires. Up to 74% of respondents had never received formal education on EBM prior to their rotation at the VAMC. Mean scores on attitudinal and belief questions improved after exposure to the curriculum, but to a non-significant level. Similarly, knowledge of EBM, as assessed by the Berlin Questionnaire, improved to a non-significant level (pre-curriculum mean score 4.5, post-curriculum mean score 5.3, p = 0.10). A regression was conducted to determine if the number of days between the pre and post-curriculum assessment predicted improved performance on the Berlin Questionnaire and was non-significant (p = 0.13).

CONCLUSIONS: We successfully implemented a rolling, integrated EBM curriculum for internal medicine students and residents. Attitudinal, behavioral and knowledge scores improved to a non-significant level after the month-long exposure to the curriculum. The lack of significant findings may be in part due to small sample size.

A SALARY EQUITY STUDY: DIFFERENCES BY GENDER AND RACE, DEPARTMENT OF MEDICINE, UNIVERSITY OF MINNESOTAAnne Joseph; Wesley Miller; Jeremiah Menk; Stewart McMullan. University of Minnesota, Minneapolis, MN. (Tracking ID #1938954)

BACKGROUND: Salary inequities for female faculty have been documented in many academic disciplines.

METHODS: The University of Minnesota Department of Medicine (DOM) conducted cross-sectional studies in 2011, 2012, and 2013 with the objective of examining gender equity issues related to salary (n = 165, n = 196, and n = 213 respectively). A pre-specified multivariable linear regression was used to adjust the salary comparison between men and women and Whites and non-Whites. The analysis included the following variables: gender, race, academic rank (and time in rank), leadership roles, degree(s), number of ABIM certifications, time from highest degree to current position, percent clinical effort, and subspecialty.

RESULTS: In 2011, 8% of women were full Professors compared to 35% of men. Leadership roles were held by 7% of women compared to 19% of men; similar differences in career characteristics were observed across all 3years. The unadjusted mean salary for women was $69,717–$76,009 lower compared to men depending on the year. In 2011, 2012, and 2013 the adjusted mean comparisons for women vs. men were $-10,513, −$9,377, and −$16,132 respectively. For non-Whites vs. Whites the comparisons were −$9,018, −$741, and −$13,002. R-squared values for the models were between 0.757 and 0.767. As predicted, the model confirmed that the following were associated with increased salary: a PhD or Masters in addition to MD degree, number of ABIM certifications, rank, time variables, and holding a leadership position. For clinicians, procedural subspecialties (cardiology, gastroenterology, pulmonary, renal) conferred higher salaries than hematology-oncology-transplant, endocrinology, rheumatology, infectious disease, general internal medicine (in that order).

CONCLUSIONS: Gender differences persisted over the 3-year period despite preferential salary adjustments for women. Professional accomplishment characteristics do not fully explain faculty salary differences by gender and race. In response, the DOM has formed a Salary Equity Committee to systematically address these differences abiding by principles of transparency, simplicity, value of clinical, educational and research missions, validated benchmarks of productivity, equal access to leadership positions and supplemental income, and sustainability.

Adjusted Mean Comparisons

A SIMPLE SCREENING METHOD TO DETECT MALIGNANCY USING RED CELL DISTRIBUTION WIDTH (RDW)Hiroaki Wakakuri; Toshihiko Ohara; Makoto Suzaki; Sonoko Kirinoki; Naoko Onodera; Takashi Araki; Hideya Hyodo; Makoto Kawai; Masahiro Yasutake. Nippon medical school, Bunkyoku-sendagi, Japan. (Tracking ID #1928358)

BACKGROUND: RDW is rapidly calculated parameter using an automatic cell counter. Recent studies showed that RDW is strongly associated with prognosis in disorders such as chronic heart failure, idiopathic pulmonary fibrosis, sepsis, various cancer, and chronic kidney disease. However, it is still unclear that the relationship of RDW and diagnosis of malignancy. The aim of this study is to evaluate the usefulness of RDW to screening malignancy at the first visit outpatient clinic.

METHODS: Fifty one consecutive patients diagnosed as malignancy at ER and general internal medicine of our hospital were reviewed in the period between September 2012 and September 2013. Fifty one patients, matching age and sex, with common diseases, including hypertension and dyslipidemia were assigned to control group. The patients complicated severe anemia (Hb < 9.0), chronic heart failure, chronic inflammation disease, with a dosing history of medication that have effect on RDW were excluded from control group. RDW was calculated using an automatic cell counter (Sysmex XE-5000). We also measured Hb, BUN, Cre, TC, albumin and CRP at the same time. The receiver operating characteristic (ROC) curve showed the best cut-off value of RDW for predicting malignancy.

RESULTS: Malignancy group (age 67.8 [24–88], man 27, woman 24) was classified into the digestive organs n 26 (gastric ca, colon ca, cecal ca, vater ca, HCC, GIST), hemopathy n 10 (malignant lymphoma, multiple myeloma, AML), respiratory organs n 6 (lung ca), gynecologic disease n 3 (ovarian ca, uterine ca), urinary organs n 3 (prostate ca, renal pelvic ca), breast cancer n 2, and liposarcoma n 1. Control group (age67.4 [28–86]) included 27 men and 24 women. Malignancy group had lower Hb and alb values than control group, and higher CRP values than control group (Hb: 12.0 [5.0–17.3] vs 13.6 [10.5–16.9], albumin: 3.64 [2.6–4.9] vs 4.21 [3.2–4.8], CRP: 3.33 [0.03–21.3] vs 0.37 [0.01–2.88], P < 0.05). There was no strong correlation between the RDW and other parameters (vs BUN r 0.096, vs Cre r 0.071, vs TC r −0.284, vs CRP r 0.182, vs albumin r −0.437, vs Hb r = −0.537). RDW of malignancy group (14.6 [12.6–25]) was significantly higher than RDW of control group (13.3 [12.2–15.6]), (P < 0.05). Using ROC analysis (AUC = 0.744), RDW of 13.9% was the best cut-off value for predicting malignancy with a sensitivity of 58.9% and a specificity of 78.5%.

CONCLUSIONS: Our result shows that RDW of malignancy group is significantly higher than control group. It is not enough to discriminate patients with malignancy diseases between healthy persons by just RDW value, but an elevated RDW > 13.9% value is a reliable additional predictor for screening malignancy at the first visit outpatient clinic.

A STRUCTURED REVIEW OF BENEFIT AND HARM DATA IN CANCER SCREENING GUIDELINES AND POINT-OF-CARE CLINICAL RESOURCES: DO CLINICIANS HAVE THE INFORMATION THEY NEED?Tanner Caverly; Elyse Reamer; Rodney A. Hayward; Michele Heisler; Angela Fagerlin. University of Michigan Medical School, Ann Arbor, MI. (Tracking ID #1939040)

BACKGROUND: Data on the absolute benefits and harms of preventive interventions represent essential information for clinicians. Clinicians cannot practice informed or shared decision making if they lack ready access to this information, which is needed to help weigh trade-offs between potential benefits and harms. We might expect that transparent quantification of potential benefits and harms would be provided in prevention guidelines and point-of-care resources (such as UpToDate©), but to date, a systematic assessment of whether these clinical resources provide this important clinical information has not been conducted. We sought to address this gap by conducting a review of current cancer prevention guidelines and point-of-service resources to examine what information they provide on the chance of benefit and the potential for harms.

METHODS: We reviewed all recommendation statements related to cancer prevention and screening from the United States Preventative Services Task Force (USPSTF), excluding statements rated I for insufficient evidence, pediatric and adolescent guidelines, and those only discussing counseling interventions. To capture other non-USPSTF guidelines in the same topic areas, we examined within the National Guidelines Clearing House (NGC) all discrete guidelines developed by US organizations. We also searched two popular point-of-care clinical resources (UpToDate© and Dynamed). A standardized abstraction form was iteratively created. We abstracted basic guideline demographic data and information about how benefits and harms are presented.

RESULTS: We reviewed 11 cancer prevention guidelines from the USPSTF, 19 guidelines from other organizations within the same cancer-prevention areas, 15 related articles from Dynamed and 14 from Up-to-Date. These resources contained 114 separate recommendations, 77 of which described recommendations for the use of the preventive service in at least some sub-populations (positive recommendations). Among the positive recommendations, 31% (24 out of 77) simply made a recommendation without mentioning any estimate of the chance of benefit. Twenty-seven percent (21 out of 77) presented the benefit as an absolute risk reduction. Regarding the potential for harm inherent in any intervention, 48% (37 out of 77) of the recommendation statements did not mention any harms and an additional 18% (14 out of 77) mentioned but did not quantify those harms. Of the 27% (21 out of 77) that quantified both potential benefits and potential harms, 19% (4 out of 21) presented benefit as a relative risk reduction (larger number) while presenting harm as an absolute risk increase (smaller number)—a format known as mismatched framing that exaggerates the relative magnitude of the benefit compared with the harm. Only 21% (n = 16; 10 from guidelines and 6 from point-of-care resources) of the 77 positive recommendations quantified both the potential benefits and harms as absolute risks so that they could be directly compared.

CONCLUSIONS: Guidelines and point-of-care clinical resources for cancer prevention interventions frequently do not provide needed information on the chance of the intervention resulting in patient benefit or the potential of being harmed. The resources we reviewed may not represent all of the resources that clinicians use in daily practice. However, other resources are unlikely to present this information if the resources we reviewed did not. After undertaking the effort to summarize and rate the quality of evidence supporting use of a preventive service, evidence review bodies should also seek to quantify and present the potential benefits and harms of the proposed intervention. A satisfactory way to present this data would be to depict estimates of absolute benefits and harms together in a single table within the guideline—as is promoted by the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach.

ACCEPTABILITY AND FEASIBILITY OF HPV SELF-SAMPLING FOR CERVICAL CANCER SCREENING AMONG PATIENTS AND PROVIDERS IN TWO SAFETY-NET INSTITUTIONS IN MIAMIKumar Ilangovan¹; Erin N. Kobetz-Kerman^2,4; Tulay Koru-Sengul^2,4; Erin N. Marcus¹; Brendaly Rodriguez³; Yisel Alonzo³; Orieta Fontan²; Valentine Cesar²; Olveen Carrasquillo^1,4. ¹University of Miami Miller School of Medicine, Miami, FL; ²Sylvester Comprehensive Cancer Center, Miami, FL; ³University of Miami Miller School of Medicine, Miami, FL; ⁴University of Miami Miller School of Medicine, Miami, FL. (Tracking ID #1931851)

BACKGROUND: Cervical cancer disproportionately impacts minority women. While screening substantially reduces cervical cancer mortality, many women in safety-net settings remain unscreened. System level barriers include limited resources and staffing. Patient barriers include financial (lack of insurance and co-payments), health literacy, and cultural norms. Provider barriers such as visit time constraints also exist. Human Papilloma Virus (HPV) self-sampling, in which women use a cotton swab device to self-collect material for an HPV assay, may be an attractive option in such settings.

METHODS: 1) To measure the acceptability and feasibility of HPV self-sampling among patients, we had two Community Health Workers (CHWs) recruit women from two clinics in Miami-Dade County. One site was a public hospital based clinic primarily serving Hispanics; the other a community based facility serving Haitians. CHWs identified women aged 30–65years with no Pap smear in the past 3years and gave a brief description of the study. We offered all eligible women the option of HPV self-sampling or discussing screening through Pap smears with their provider. If HPV self-sampling was selected, instructions on sampler use and assistance with collection was provided by the CHW. Patient attitudes towards HPV self-sampling were assessed using a 12-item survey modeled after existing surveys. Among women in whom the HPV self-sampler option was not chosen, the electronic medical record or paper chart were reviewed to assess if a Pap smear was performed. 2) To measure acceptability and feasibility among clinic staff (medical assistant, nurse, nurse practitioner, and physician), we developed a 9-item survey based on prior studies of provider screening acceptability. Through emails and personal invitations, staff were encouraged to participate in an anonymous online questionnaire.

RESULTS: Of 1725 women assessed, 62% were not eligible due to having a Pap smear in the last 3years. Of the 184 study eligible women, 91% agreed to participate (n = 167). Latinas represented 73% and Haitians 27% of the sample; 99% were uninsured. The HPV self-sampling approach was selected by 67% of participants. The majority of participants agreed that HPV self-sampling was faster than the Pap smear (86%), more private (93%), and easy to use (97%). Nearly all participants agreed they would use the self-sampler again (97%) and recommend it to a friend or family member (96%). Overall, 76% of participants preferred the self-sampler to the Pap smear. High-risk HPV was detected in 11% of samples. The CHW provided navigation to Pap smear for all women having high-risk HPV strains. Of those not selecting the self-sampler, 34% had not yet had a Pap smear at 1month post-recruitment. Of 178 clinic staff invited to participate in our survey, 16% (n = 29) completed the survey. Of those completing the survey, 69% agreed it was difficult to perform a Pap smear because of time constraints and 79% agreed they would be willing to incorporate HPV self-sampling into their practice.

CONCLUSIONS: HPV self-sampling was both feasible and highly acceptable in the safety-net clinics studied. A low response rate by clinic staff may limit the generalizability of the provider survey results. HPV self-sampling may be an important tool to address barriers to cervical cancer screening in resource limited settings.

ACCESS TO AND PREFERENCES FOR TEXT MESSAGING FOR MEDICAL AND INSURANCE REMINDERS IN A SAFETY NET POPULATIONLeah Zallman^1,2; Alex Harsha^1,3; Danny McCormick^1,3. ¹Cambridge Health Alliance, Cambridge, MA; ²Institute for Community Health, Cambridge, MA; ³Harvard Medical School, Boston, MA. (Tracking ID #1934972)

BACKGROUND: Text messaging has been used successfully to deliver health information interventions to underserved communities. Evidence suggests that text message reminders can increase patient appointment and treatment completion, and is a reliable method of communication even in communities with low literacy. Text messaging may offer similar opportunities to reach patients in health care safety net settings who are disproportionately impacted by homelessness or unstable housing, low health literacy and may be lost to follow up. The immediate, direct-to-patient nature of text messaging and the ability to send automated reminders at relatively low cost may help overcome communication barriers between healthcare providers and underserved communities in the US. While the opportunity to improve this communication may exist through text messaging, preferences for text messaging and patient characteristics associated with preferring text messages have not been elucidated

METHODS: Between August and December 2013, we conducted a face-to-face verbally delivered survey regarding preferences for text messaging with 415 patients presenting to three emergency departments at a large integrated safety net health care system in the greater Boston area. We included patients aged 18–64, who had an Emergency Severity Index (ESI) of 2–5 (excluded most severely ill), had only one insurance type and spoke English, Portuguese, Spanish or Haitian Creole. We asked participants about their access to mobile phones with text messaging capability, preferences for receiving specific types of reminders by text (appointment, medication, and expiring insurance), and preferences for text messaging over other forms of communication. We used logistic regression models that controlled for age, gender, race/ethnicity, health insurance type and medical comorbidities to identify predictors of preferring text messages for medical reminders over other forms of communication with medical providers.

RESULTS: Overall, 96.5% of participants reported having access to mobile phones with text messaging capability. Half reported a preference for text messaging over other forms of communication such as phone, mail or email. Respondents expressed preferences for appointment reminders (77%), reminders regarding expiring insurance (55%), and medication reminders (32%). A minority (17%) indicated that they preferred not to be text messaged by their healthcare provider. Reasons for not preferring text messaging included: cost of receiving text messages (11%), privacy (17%), not liking or understanding text messaging (40%) and other (31%). In logistic regression analyses, Hispanic ethnicity (1.68 [1.03–2.75]) and younger age (0.98 [0.96–1.00]) were associated with preferences for medication reminders. Similarly, Hispanic ethnicity (1.93 [1.09–3.43]) and younger age (0.98 [0.95–1.00)) were associated with preferences for expiring insurance reminders.

CONCLUSIONS: In a safety net population, the vast majority of patients have access to text messaging and a high proportion express preferences for text message reminders over other forms of communication. Reminders regarding appointments and expiring insurance were most highly desired. Text messaging may be a promising method to improve healthcare provider communication with underserved communities in the US.

ACCESS TO DEPRESSION CARE SERVICES: A NEEDS ASSESSMENT IN AN URBAN ACADEMIC PRIMARY CARE CLINICLauren Peccoralo; Susan Truong. Mount Sinai Medical Center, New York, NY. (Tracking ID #1937528)

BACKGROUND: Major depressive and anxiety disorders are common and these and other psychiatric conditions are often encountered by primary care practitioners. Urban academic training centers serve patients at high risk for mental health disorders who often have little access to care, yet little is known about the perceived need for mental health services amongst primary care providers in these environments. The US Preventive Services Task Force endorses routine screening of all adults for depression in settings with adequate staff-assisted depression care support to provide appropriate care and follow up. The purpose of this survey was to assess the needs and barriers to depression care services amongst physician primary care providers in an urban academic training practice in the United States prior to the implementation of universal screening and a depression care program.

METHODS: From April through June 2013, we conducted an online survey about the perceptions of barriers to depression care and unmet psychiatric needs in the primary care practice in our clinic. Survey participants included primary care providers (all internal medicine residents and faculty providers) at the resident training clinic practice at the Mount Sinai School of Medicine in New York, New York. The survey was analyzed as descriptive data.

RESULTS: Ninety-two participants responded to the survey (74 resident trainees and 18 attending physicians, 62% and 78% response rate, respectively) during the study period. Eighty-three percent of residents and 88% of attendings are currently caring for a patient(s) with depression. Eighty-six percent of residents and 100% of attendings believe there are barriers to providing optimal care for patients with depression. The most commonly identified barriers to care are lack of psychiatrists (85% residents, 94% attendings), lack of coordination between primary care and psychiatry (83% residents, 100% attendings). and lack of patient education (54% residents, 78% attendings). Ninety-six percent of residents and 94% of attendings have at least one patient with an unmet psychiatric need and 18% of residents and 35% of attendings have >10 patients with such a need. The most commonly cited psychiatric needs are depression (93% residents, 100% attendings), anxiety (74% residents, 94% attendings) and substance abuse (45% residents, 47% attendings).

CONCLUSIONS: Primary care providers at all levels perceive high need for and multiple barriers to depression and other mental health care. These data support the recommendation that appropriate mental health services should be available in tandem with the initiation of universal depression screening. Further research is needed to determine the impact of implementation of universal screening and depression care program in a primary care training environment.

ACCULTURATION AND MEASURES OF CARDIOVASCULAR RISK AMONG FILIPINA IMMIGRANTSJane Jih¹; Alka M. Kanaya¹; Maria Rosario G. Araneta². ¹UCSF, San Francisco, CA; ²UCSD, San Diego, CA. (Tracking ID #1938915)

BACKGROUND: Acculturation is a dynamic process that encompasses a broad spectrum of cultural changes, interactions and adaptations that affect health. In immigrant populations, acculturation has been associated with increased prevalence of obesity, diabetes and coronary heart disease (CHD). CHD and its risk factors are highly prevalent among Filipino Americans. However, limited studies have examined acculturation and CHD specifically among Filipino Americans. We used the University of California San Diego (UCSD) Filipino Women’s Health Study to examine the relationship of acculturation and CHD and related risk factors among Filipina immigrants.

METHODS: The UCSD Filipino Women’s Health Study is a longitudinal study of 453 Filipino American women ages 39–86years recruited from 1995 to 1999 in which the prevalence of and risk factors for CHD, hypertension (HTN), diabetes and obesity was measured. In addition to collecting immigration data, an adapted acculturation scale with three domains [a) functional integration via language use; b) values on preserving culture; c) attitudes on family structure and sex-gender roles] was administered. Outcome variables included prevalence of CHD (defined by ECG, Rose questionnaire and history of myocardial infarction), HTN, diabetes and obesity. We examined several measures of acculturation including length of residence in the United States (US), proportion of life years in the US (years residing in US/age at enrollment) and each domain of the acculturation scale. We used logistic regression, adjusting for age, education and exercise, smoking, alcohol use and body mass index (BMI) for the CHD outcome, to estimate the association of each measure of acculturation to each outcome.

RESULTS: The mean age of Filipina immigrants (n = 446) was 58 ± 9years old. The mean length of US residence was 16years (range 0–48). The prevalence of CHD was 21%, HTN was 51%, and type two diabetes was 33%. Mean BMI was 25 ± 3kg/m² and 23% were obese as defined by the World Health Organization (WHO) Asian BMI scale (BMI ≥ 27.5kg/m²). Measures of acculturation were not associated with CHD, HTN and diabetes outcomes. Length of residence in the US but not proportion of life years or the acculturation scale was associated with obesity defined by the WHO Asian BMI scale. For each additional year of living in the US, Filipina immigrants had a 21% increase in the odds of obesity (adjusted odds ratio 1.21 (95% CI 1.01–1.47)).

CONCLUSIONS: Filipina immigrants have a substantial burden of CHD and risk factors. Unlike other immigrant populations, acculturation to life in the US was not associated with CHD, HTN and diabetes prevalence in Filipino women. However, longer duration of residence in the US was associated with increased odds of obesity. Exposure to a westernized lifestyle during the Spanish and US colonization in the Philippines may mitigate the impact of migration on cardiovascular health among Filipina immigrants.

ACCURACY AND GAMING OF THE ACUTE BRONCHITIS ANTIBIOTIC PRESCRIBING PERFORMANCE MEASUREPatrick P. Dempsey¹; Harry Reyes Nieva¹; Alexandra C. Businger¹; Jeffrey A. Linder^1,2. ¹Brigham and Womens Hospital, Boston, MA; ²Harvard Medical School, Boston, MA. (Tracking ID #1938252)

BACKGROUND: Acute bronchitis is a cough-predominant illness of less than 3weeks duration in a patient without chronic lung disease. A national performance measure states that antibiotics are not indicated for acute bronchitis. However, concerns have been raised about the measure’s accuracy and susceptibility to gaming. We compared the performance measure to chart review to assess the accuracy of the performance measure and detect gaming.

METHODS: We assessed 57,766 visits to an academic primary care practice between May 2011 and September 2012. For the chart review, we included patients aged 18–64 and excluded patients with chronic lung disease on the problem list or who made a study clinic visit in the prior 30days (n = 26,539). Two trained chart abstractors reviewed these notes and coded whether the treating clinician diagnosed the patient with acute bronchitis. We calculated a modified Healthcare Effectiveness Data and Information Set (HEDIS) performance measure by including patients aged 18–64 with ICD-9 diagnosis codes of 466 or 490, excluding patients with chronic lung disease, immunosuppression, and cancer. Using chart review as the gold standard, we calculated the sensitivity, specificity, and positive predictive value of the acute bronchitis performance measure.

RESULTS: There were 31,174 visits which neither the performance measure nor chart review identified as an acute bronchitis visit. According to the performance measure, there were 90 acute bronchitis visits (88 with a diagnosis code of 466 and 2 with a diagnosis code of 490). There were 143 visits identified as acute bronchitis visits by both the performance measure and chart review (n = 41; “true positives”), the performance measure only (n = 49; “false positives”), or chart review only (n = 53; “false negatives”). These patients had a mean age of 48years old; 77% women; and 44% white, 28% Black, and 17% Latino. Compared to chart review, the performance measure had a positive predictive value of 46%, a sensitivity of 44% and a specificity of >99%. Of the 49 “false positives,” according to the chart, 10 visits were for other ARIs, seven visits were to other practices, seven the patient had been coughing for more than 3weeks, five the patient did not have cough, five the patient had chronic lung disease, and 15 the patient had a visit in the prior 30days. Of the 53 “false negatives”—for which the clinician diagnosed the patient with acute bronchitis in their note—52 did not have an acute bronchitis diagnosis code (most common diagnosis codes were cough, hypertension, upper respiratory infection, and viral infection) and one had a diagnosis code of asthma that was not mentioned in the chart. The antibiotic prescribing rate for true positives was 22%, for false positives was 4%, and false negatives was 30%.

CONCLUSIONS: Compared to chart review, the acute bronchitis performance measure has a sensitivity and positive predictive value of less than 50%. The performance measure omitted visits with antibiotic prescribing and falsely included visits without antibiotic prescribing, suggesting gaming.

ADAPTIVE RESERVE AT COMMUNITY HEALTH CENTERS: THE CANCER PREVENTION AND CONTROL RESEARCH NETWORK (CPCRN) MULTI-STATE SURVEYShin-Ping Tu^1,2; Alan Kuniyuki²; Vicki M. Young³; Maria Fernandez⁴; Rebecca S. Williams⁵; Amanda Kimura². ¹Virginia Commonwealth University, Richmond, VA; ²University of Washington, Seattle, WA; ³South Carolina Primary Healthcare Association, Columbia, SC; ⁴University of Texas at Houston, Houston, TX; ⁵UNC Chapel Hill, Chapel Hill, NC. (Tracking ID #1936531)

BACKGROUND: Adaptive Reserve comprises intangible elements of a practice’s capacity (i.e., human relationships, teamwork, a learning culture) that provides flexibility, resilience, and enhance positive change. Our objective is to determine the Adaptive Reserve at Community Health Centers (CHCs) in seven states (California, Colorado, Georgia, Missouri, South Carolina, Texas, and Washington).

METHODS: We selected the Patient Centered Medical Home (PCMH) and the Consolidated Framework for Implementation Research to guide this research. Our survey included questions on colorectal cancer screening PCMH best practices and items from the Clinician Staff Questionnaire to examine the practice Adaptive Reserve (PAR) of participating CHC clinics. PAR scores were scaled from 0.00 to 1.00, with higher scores representing greater Adaptive Reserve. In collaboration with Primary Care Associations and CHCs, we recruited a convenience sample of 76 CHC clinics and invited their clinical staff (providers, nurses, and medical assistants) to participate in our CPCRN Community Health Center Survey.

RESULTS: Survey respondents (N = 246) are mostly female (82%) and non-Hispanic (61%). Forty-two percent reported their clinic had 7–8 colorectal cancer (CRC) screening PCMH best practices (maximum 8). Preliminary PAR scores in the states ranged from 0.61 to 0.7. In a study on PCMH implementation, 36 practices expected to have significant capability for change had a mean baseline PAR score of 0.69. Our survey indicates higher levels of PAR to be positively associated with CRC screening best practices in the clinics (adjusted OR = 3.62; 95% CI 1.31, 10.02) and how frequently staff performed the best practices (p = 0.005).

CONCLUSIONS: Evidence from primary care transformation indicates that practices with strong Adaptive Reserve are more successful at incorporating change. With the Affordable Care Act, understanding the Adaptive Reserve of CHC clinics will be timely for the many changes needed to successfully expand primary care at CHCs.

ADVANCE DIRECTIVE INTERVENTION FOR CHRONICALLY HOMELESS PEOPLE: ACCEPTABILITY, FACTORS ASSOCIATED WITH COMPLETION, AND TREATMENT PREFERENCESAlexander Leung¹; Dhruv Nayyar¹; Manisha Sachdeva¹; John Song²; Stephen Hwang¹. ¹St. Michael, Toronto, ON, Canada; ²University of Minnesota, Minneapolis, MN. (Tracking ID #1933803)

BACKGROUND: End-of-life care planning is relevant for homeless individuals because they experience high morbidity and mortality. A previous study has shown that homeless persons are more likely to complete advance directives with a one-on-one counselor-guided intervention than if the advance directive is self-completed. However, the acceptability of a counselor-guided intervention for the completion of advance directives among chronically homeless individuals has not been investigated. The aim of this study was to determine the acceptability of a counselor-guided intervention for the completion of advance directives among chronically homeless individuals. Secondary objectives were to identify characteristics associated with willingness to complete an advance directive and to characterize treatment preferences in this population.

METHODS: In a single-group intervention study, homeless men residing at a shelter in Toronto, Ontario were recruited from April to June 2013. Recruitment took place among programs that served chronically homeless individuals who were residents in a managed alcohol harm reduction program, a medical respite program that treats individuals with acute or uncontrolled chronic illnesses, and a long-term program for individuals who have been homeless for extended periods of time. Two hundred five homeless men (89.1% of those approached) participated in the study. Participants completed a survey which obtained information on potential predictors of advance directive completion, including socio-demographic characteristics, health status, health care use, and attitudes towards end-of-life care and death. After completing this survey, all participants were offered an opportunity to complete an advance directive with a trained counselor. The primary outcome was the proportion of participants who completed an advance directive. The characteristics associated with completion of an advance directive were examined.

RESULTS: The duration of homelessness was 12months or longer in 72.8% of participants. A total of 103 participants (50.2%) completed an advance directive. Socio-demographic characteristics, health status, and health care use were not associated with completion of an advance directive. Participants were significantly more likely to complete an advance directive if they reported thinking about death on a daily basis, believed that thinking about their friends and family was important, or reported knowing their wishes for end-of-life care but not having told anyone about these wishes. Of 103 individuals who completed an advance directive, the proportion expressing a preference to receive cardiopulmonary resuscitation in the event of a cardiorespiratory arrest was 94.1%, 81.3%, 64.7%, and 36.3% if there was a chance of returning to their current state of health or hypothetical health states of moderate dementia, severe dementia, and permanent coma, respectively.

CONCLUSIONS: As the homeless population continues to age, health care professionals will encounter an increase in chronically homeless persons facing complex end-of-life care needs. A single-encounter, one-on-one counselor-guided intervention can achieve a high rate of advance directive completion among chronically homeless individuals. Health care providers should initiate advance care planning discussions with their patients who are homeless. Future research aims to ascertain whether the completion of an advance directive has an effect on subsequent care in homeless individuals.

ADVANCED DIRECTIVE USE IN INTERNAL MEDICINE RESIDENCY CLINICS: A RETROSPECTIVE REVIEWBriana N. Ketterer; Corinne Self; Guadalupe Martinez; Jennifer Gabbard; Macy Whitley; Ellyn Lee. University of Arizona, Tucson, AZ. (Tracking ID #1923316)

BACKGROUND: Advanced directives (AD) allow patients to voice preferences about life sustaining therapies and end-of-life (EOL) care. ADs improve satisfaction, reduce stress, anxiety and depression at EOL. However, this information needs to be discussed and documented to be effective. Resident physicians play a key role as one of the first providers to encounter patients thus, making them an integral hub in the health delivery system. We sought to evaluate Internal Medicine (IM) Residents’ practice habits and assess participation in EOL/AD planning, and documentation of those preferences. This information will serve to address potential weaknesses in the Residency curricula and areas for improvement.

METHODS: Retrospective chart review of patients seen in IM Residency clinics at The University of Arizona Medical Center (UAMC) from July 2011–2013. Inclusion criteria: ≥65years of age with at least one chronic, debilitating diagnosis associated with morbidity or mortality (dementia, AIDS, cancer, neurodegenerative disease, chronic obstructive pulmonary disease, chronic kidney disease, cerebral vascular accident or transient ischemic attack, liver cirrhosis, coronary artery disease or myocardial infarction, post-transplant) documented in the electronic medical record, AllScripts. Documentation was reviewed and ADs were recorded along with demographics (age, gender), number and type of diagnoses.

RESULTS: Reviewed 1075 charts, 165 met inclusion criteria (M 47%, F 53%; average age, 75; average number of problems, 7) with most frequent diagnoses of: cardiovascular disease (52%), cancer (34%), CKD (17%). AD documented in 17 of 165 charts (10%) which included: AD discussion (2), no specific qualifiers (1), Medical Power Of Attorney (8), DNR/DNI (9), Living Will (4), Hospice (1), and documentation for first responders (4).

CONCLUSIONS: IM Residents at UAMC clinics documented ADs in a mere 10% of seemingly tenuous patients ≥65years of age. This represents a cohort who would benefit from AD conversations to preserve their wishes at EOL. Although residents have some training in EOL care and AD, the curricula appears insufficient as the rate of AD documentation is very low and residents feel uncomfortable with those discussions as unpublished data from our group similarly suggests. One way to improve upon this is to implement an educational intervention to enhance training, improve AD discussions, and address all AD components with patients. Improvement in this area is crucial as it would not only fulfill residency competency requirements but also lead to more complete ADs allowing patients greater autonomy in their EOL care.

AGING WITH HIV: SYMPTOM BURDEN AND IMPACT ON QUALITY OF LIFEMeredith Greene^1,2; Christine Ritchie¹; Jeff Martin⁴; Joy A. Madamba³; Monica Mattes⁵; Yinghui Miao²; Steven Deeks³; Victor Valcour^1,6. ¹University of California San Francisco, San Francisco, CA; ²San Francisco VA Medical Center, San Francisco, CA; ³University of California San Francisco, San Francisco, CA; ⁴University of California San Francisco, San Francisco, CA; ⁵University of Central Florida College of Medicine, Orlando, FL; ⁶University of California San Franscico, San Francisco, CA. (Tracking ID #1932204)

BACKGROUND: As HIV infection has shifted from a life-threatening disease to a chronic illness, the HIV-infected population is aging and often facing multiple co-morbid conditions. Symptom assessment is an important clinical tool in the management of patients with complex chronic disease, yet limited data exists about symptom burden in older HIV-infected adults especially in the modern treatment era. The primary aim of this study was to describe the symptom burden (frequency and severity of symptoms) among HIV-infected adults over age 50 on combination antiretroviral therapy. We also examined factors associated with symptom burden in this population and the association of symptom burden on health related quality of life.

METHODS: We conducted a cross-sectional survey in 2012–2013 of HIV-infected adults over age 50 that had an undetectable viral load (HIV RNA) on antiretroviral therapy and were enrolled in a clinic based cohort at a county medical center in San Francisco. Sixteen symptoms were measured by self-report in a two part question based on the 12 item HIV symptom index:1) yes/no for presence of each symptom in the past week and 2) degree of severity of each symptom (not a problem, mild, moderate, severe, very severe) as judged by symptom frequency, intensity, and duration. Symptom severity was analyzed by examining the number of symptoms rated as moderate, severe, or very severe. Health related quality of life was assessed by a single-item self-report of health as excellent, very good, good, fair or poor. Demographics, HIV related factors, co-morbidities and medications were measured by self-report and verified by chart review. Descriptive statistics were used for data analysis.

RESULTS: 142 participants were enrolled, of which 94% were male, 63% were Caucasian, with a median age of 57 (range 50–74). The median CD4 count was 576 (IQR 393–715) and the median length of HIV infection was 22 (IQR 18–25) years. Participants had a median of 4 (IQR 3–6) co-morbid conditions, most commonly hyperlipidemia (61.3%) and hypertension (52.8%) and were taking a median of 9 (IQR 5–12) non-antiretroviral medications. Sixty-nine percent of participants reported two or more symptoms, with a median of 3 (IQR 1–5) symptoms reported. Fatigue (n = 72, 51%), “feeling sad or depressed” (n = 60, 42%), “pain, tingling, loss of feeling in feet or hands” (n = 54, 38%) and insomnia (n = 53, 37%) were the most commonly reported symptoms. Fifty-five percent of participants reported at least one moderate to very severe symptom with a median of 1 (IQR 0–2) moderate to very severe symptom reported. Fatigue (n = 31, 22%), “feeling sad” (n = 31, 22%) and insomnia (n = 25, 18%) were the most frequent severe symptoms. The total number of co-morbid conditions was associated with both the total number of symptoms reported (rs = 0.21, p = 0.01) and symptom severity (rs = 0.27, p < 0.001) and the total number of non-antiretroviral medications was associated with symptom severity (rs = 0.21, p = 0.01). HIV related factors such as CD4 count and length of HIV infection did not have statistically significant correlations with symptom frequency or severity. Both the total number of symptoms and symptom severity were inversely correlated with overall health related quality of life (rs −0.42, p < 0.001; rs −0.47, p < 0.001).

CONCLUSIONS: Among adults age 50 and older with well-controlled HIV-infection, symptom burden was substantial and associated at least in part with co-morbid conditions. Addressing symptom burden during routine clinical encounters merits attention and may improve quality of life as the HIV-infected population ages.

ALLOWING PATIENT CHOICE BETWEEN TWO DIVERSE DIETS DID NOT IMPROVE WEIGHT LOSSWilliam S. Yancy^1,2; Cynthia Coffman^1,3; Valerie Smith¹; Stephanie B. Mayer⁴; Ronette L. Kolotkin⁶; Paula J. Geiselman⁵; Megan McVay¹; Eugene Z. Oddone^1,2; Corrine Voils^1,2. ¹VAMC, Durham, NC; ²Duke University, Durham, NC; ³Duke University, Durham, NC; ⁴Virginia Commonwealth University, Richmond, VA; ⁵Louisiana State University, Baton Rouge, LA; ⁶Quality of Life Consulting, Durham, NC. (Tracking ID #1941417)

BACKGROUND: A variety of diet approaches lead to modest weight loss, and dietary adherence is a strong predictor of success. Allowing individuals to choose diet composition after considering their food preferences may improve adherence by increasing autonomy and self-efficacy. In a randomized trial, we examined whether participants who chose a diet experienced greater weight loss than those randomly assigned a diet.

METHODS: Participants were Durham Veterans Affairs Medical Center outpatients aged <75years with body mass index (BMI) ≥30kg/m2 and no unstable health problems. They were randomized to the Control or Choice arm, stratified by sex, BMI < or ≥40kg/m2, and presence of type two diabetes. Control participants were randomly assigned to follow a low-carbohydrate (LCD) or a low-fat/reduced-calorie (LFD) diet. Choice participants were presented individual results from the Geiselman Food Preference Questionnaire, indicating which diet fit their food preferences best, as well as information about the two diets; 1week later, they were called to elicit their diet choice. All participants then received diet counseling led by a dietitian in groups of 8–19 participants separated by arm and diet approach. In-person group sessions occurred every 2weeks for 24weeks, then every 4weeks for 24weeks interspersed with one-on-one telephone sessions mid-month from the dietitian focusing on behavioral goal-setting. All participants received literature specific to their diet; a pocket calorie/carbohydrate/fat counting guide; and a recommendation to exercise 150min per week. At week 12, Choice participants could switch to the other diet if dissatisfied with their initial diet choice. The primary outcome, weight, was measured on a digital scale at the 19 visits in light clothing and no shoes. Secondary outcomes measured every 12weeks were dietary intake by Block food frequency questionnaire and the Impact of Weight on Quality of Life-Lite (IWQOL-Lite) questionnaire (scale 0–100; higher score indicates higher quality of life). Linear mixed models were fit for these continuous outcomes, and included a common intercept, time effect, time*treatment interaction, randomization stratification variables and random effect to account for group clustering.

RESULTS: A total of 207 participants were enrolled. Mean age was 54years, 27% were women, 51% were White, 43% were Black, 22% had diabetes. Mean weight at baseline was 109kg and mean BMI 36kg/m2. In the Choice arm, 61 (58%) chose the LCD and 44 (42%) chose the LFD; only three LCD and two LFD participants switched diets at 12weeks. Final measurements were collected in 88 of 102 (86%) in the Control arm and 87 of 105 (83%) in the Choice arm. Mean weight loss at 48weeks was −5.3% in Control participants and −4.2% in Choice participants (difference 1.1%, 95% CI 0.0 to 2.2, p = 0.06). Dietary adherence was similar in the Control and Choice arms, with mean percent deviation from macronutrient goal differing by 0.6% (10.1% vs 10.7%, respectively) at week 48. Mean IWQOL-Lite total scores were similar at week 48 (Control 82.9, Choice 81.7, difference −1.3, 95% CI −4.5 to 2.0).

CONCLUSIONS: Educating patients on diet approaches and their food preferences, and then allowing them to choose their diet composition, did not result in greater weight loss compared to random diet assignment. Future research could consider targeting diet to individuals based on metabolic, genetic or other characteristics that might predict improved outcomes.

AN ACHILLES HEEL OF DISPARITIES RESEARCH: MISSING RACIAL DATA IN A STUDY OF MORTALITY AFTER CRITICAL ILLNESSMallika Mundkur^1,2; Swapna Abhyankar¹; Fiona M. Callaghan¹; Clement J. McDonald¹. ¹National Institutes of Health, Bethesda, MD; ²George Washington University, Washington, DC. (Tracking ID #1936298)

BACKGROUND: Research has consistently demonstrated the presence of racial disparities in care and outcomes across inpatient and outpatient settings. However, studies focused on disparities in the intensive care unit (ICU) have yielded conflicting results, albeit with considerable variation in design and case-mix. Some analyses of ICU outcomes have detected greater mortality for minorities as compared to non-minority populations for conditions such as severe sepsis or trauma (Mayr et al., 2010; Haider et al., 2007), while others have found lower mortality rates for minorities with conditions such as community-acquired pneumonia (Frei et al., 2010). Recently, a retrospective study of 9518 patients in California concluded that in-hospital all-cause mortality for ICU patients was equivalent across race categories after adjusting for socioeconomic status, resuscitation preferences and severity of illness (Erickson et al., 2011). In the study by Erickson et al., patients without known race, equivalent to 10% of overall study population, were excluded from the analysis—potentially biasing results, but noted to be common practice among other studies of racial disparities (Long, Bamba, Ling, Shea, 2006). We designed our study not only to contribute to the unresolved discourse on disparities in the ICU but also to ensure that patients with missing race would be examined.

METHODS: We performed a retrospective cohort study utilizing the Multiparameter Intelligence Monitoring in Intensive Care (MIMIC-II) research database (Saeed et al., 2011; Lee et al., 2011). MIMIC-II contains de-identified clinical information on 32,425 adult and neonatal ICU patients admitted to Beth Israel Deaconess Hospital in Boston, MA between 2001 and 2008. Our cohort was comprised of adult patients (age >18) admitted to a medical or surgical ICU for at least 4h for a non-elective admission. For patients with multiple ICU admissions, only the first stay was included. Our primary outcome was 30-day all-cause mortality. Covariates were: race, age, gender, marital status, type of insurance coverage (indicator of socioeconomic status), code status within 2h of ICU admission, Simplified Acute Physiology Score (SAPS) (severity of illness indicator), admission source (e.g., emergency room vs. transfer) and comorbidities including congestive heart failure, chronic renal failure/end-stage renal disease, chronic liver disease, and cancer. We used demographic data to group patients into the following race categories: White, Black, Hispanic, Asian/Pacific Islander, Other and included an Unknown category for patients without documented race. We used the software package R for all statistical analysis. We used Kruskal-Wallis, Pearson’s chi-squared, and Fisher’s exact tests to assess the unadjusted associations between race and the covariates, log-rank test to detect differences in survival curves, and multivariable logistic regression for our main analysis.

RESULTS: 13, 491 patients met our inclusion criteria and crude mortality rates for the race categories (p < 0.0001) were as follows: 17.6% for the total population, 17.6% for the White group, 13.6% for the Asian group, 6.7% for the Hispanic group, 14.7% for the Other category, and 28% for the Unknown group. Odds ratios for our primary outcome of 30-day mortality are summarized in Table 1. For the Unknown group, 40.2% of the patients were transfers as compared to 14.6% for the overall study population (p < 0.0001).

CONCLUSIONS: We were surprised by the finding that all-cause mortality at 30days was substantially higher for patients with unknown race than for any of the other race categories. Aside from the White group, the unknown group was larger than any of the other race categories. Thus, while the White group appeared to have the greatest mortality among the known race groups, if the unknown group were predominantly Black, Hispanic or another race, the results could have been different. For the Unknown category, we observed a disproportionate number of transfers from other hospitals. Our findings generate questions for those researching healthcare disparities, including why race is unknown for many patients and whether this group is at higher risk of other adverse outcomes. Efforts should be made to include these patients in analyses and reduce the number of patients without documented race to improve our ability to accurately examine and eliminate healthcare disparities.

Table 1. Risk for 30-day Mortality by race based on adjusted multivariable logistic regression

AN EVALUATION OF “CLINICALLY PROVEN” DIETARY SUPPLEMENTSAlysha Melnyk; Bimal Ashar; April S. Fitzgerald. Johns Hopkins University, Baltimore, MD. (Tracking ID #1939374)

BACKGROUND: The Dietary Supplement Health and Education Act (DSHEA) permits manufacturers to sell products without providing pre-market evidence of safety or efficacy. One fundamental reason for the passage of the DSHEA was to empower consumers to make their own choices. Yet, data suggests that the public is unaware of the limited regulation on product marketing and advertising. Dietary supplements are frequently advertised as being “clinically proven.” Yet, there are not standards in place to determine acceptable use of this term. This study was undertaken to evaluate products that have been advertised as being “clinically proven” in order to determine the availability of data used to make such an assertion.

METHODS: An internet survey for products was performed by using the search the search tems “clinically proven dietary supplements”, “dietary supplements”, “best dietary supplements” in common search engines. Advertisements were then reviewed to see if they contained terms that suggested definitive efficacy. Products whose advertisements contained terms such as “clinically proven”, “scientifically proven”, “clinically supported” were included in the study. Products returned via the search were then categorized for what they were purported to be proven to do. Additionally, the supplement advertisements were analyzed to determine if evidence was readily available to consumers that would support the “clinically proven” label. If evidence supporting the statement was not readily available on their website, companies were contacted to attempt to obtain any information supporting their claim. Descriptive statistics were used to categorize products and determine what percentage of such products had available references to studies supporting their claims.

RESULTS: A total of 99 dietary supplements were found that referenced definitive scientific efficacy. Most of these products (63%) were advertised as being useful for helping consumers lose weight and/or build muscle. Despite the assertion for efficacy, only 28 made direct reference on their website to any scientific studies. After being contacted, an additional eight companies provided studies and/or links to evidence. The remaining companies (64%) either had no testing to report (21%), claimed that they had testing but would not provide it (4%), claimed that their evidence was all on their website (13%), did not respond to inquiries (14%), or had no listed contact information on their internet advertisement (11%)

CONCLUSIONS: A number of dietary supplements are advertised as being clinically or scientifically proven, yet provide little to no evidence to consumers to substantiate their claims. Truth-in-advertising standards for dietary supplements should define parameters for using such terminology in order to protect the public from potentially misleading claims.

AN EVALUATION OF COMPUTERIZED DRUG-DRUG INTERACTION ALERT OVERRIDES IN AMBULATORY CARESarah P. Slight^1,2; Diane L. Seger³; Karen C. Nanji^4,5; Insook Cho^1,6; Nivethietha Maniam³; Patricia C. Dkyes^1,4; David W. Bates^1,4. ¹Brigham and Women’s Hospital, Boston, MA; ²Durham University, Stockton on Tees, United Kingdom; ³Partners Healthcare Systems, Inc.,, Boston, MA; ⁴Harvard Medical School, Boston, MA; ⁵Massachusetts General Hospital, Boston, MA; ⁶Inha University, Incheon, MA. (Tracking ID #1935807)

BACKGROUND: Computerized physician order entry with clinical decision support (CDS) represents a valuable tool for improving medication safety, quality, and efficiency. CDS systems can provide health care providers with real-time, relevant, patient-specific information and alert them to potential prescribing errors. Providers may choose to override these alerts however, selecting a coded reason in order to proceed with the medication order. It is often unclear whether these particular actions, e.g., will monitor drug levels as recommended, are carried out by the provider or just simply ignored. We evaluated why providers overrode drug-drug interaction (DDI) alerts and what actions they took as a consequence of the alert.

METHODS: This study was a cross-sectional, observational study of DDI alerts generated over a three-year period between January 2009 and December 2011. All Level 2 DDI alert overrides were downloaded with IRB approval from 36 primary care practices affiliated with two Harvard teaching hospitals (total 14,966 overrides, 60.2% of alerts generated). Any duplicates were removed and replaced. A physician and pharmacist expert panel assessed the appropriateness of these DDI alert overrides. A detailed chart review was conducted for each of the appropriate overrides to ascertain whether the provider had carried out their intended action(s).

RESULTS: Providers appropriately overrode just over two-thirds (338/496) of the DDI alerts and carried out the intended action in less than two-thirds (214/338) of these cases. Eight drugs were responsible for generating approximately three quarters of these alerts. The most common coded reasons for overriding DDI alerts were ‘will monitor as recommended’ (43.9%, n = 218), ‘will adjust dose as recommended’ (16.9%, n = 84), and ‘patient has already tolerated combination’ (15.7%, n = 78). Of the 121 appropriate alert overrides where the provider said they would ‘monitor as recommended’, only 35.5% (n = 43) actually completed the monitoring. Our study also revealed how only 60% (n = 21) of providers who indicated that they would ‘adjust the dose as recommended’ actually did. Furthermore, some providers selected the coded reason ‘patient has already tolerated combination’, yet no information was found to suggest that the patient had been taking both drugs together previously.

CONCLUSIONS: We found that providers continue to override important and useful alerts, some of which likely cause serious patient injuries, although we did not measure injuries in this study. A key difference between our study and previous studies1 is that we also reviewed the charts to see whether the provider actually took action as a consequence of the alert. We found that the intended action was carried out in only two-thirds of cases. More research is needed to explore the human factors elements that influence provider behaviour, such as patient demands, workload, and time constraints. References: 1. Shah NR, Seger AC, Seger DL et al. Improving Acceptance of Computerized Prescribing Alerts in Ambulatory Care. JAMIA 2006;13:5–11.

AN EVALUATION OF COMPUTERIZED MEDICATION ALERT OVERRIDE BEHAVIOR IN AMBULATORY CARENivethietha Maniam¹; Sarah P. Slight^1,3; Diane L. Seger¹; Mary Amato^1,7; Dustin McEvoy¹; Karen C. Nanji^4,5; Insook Cho^1,6; Patricia C. Dkyes^2,4; David W. Bates^1,2. ¹Partners HealthCare, Cambridge, MA; ²Brigham and Women’s Hospital, Boston, MA; ³The University of Durham School of Medicine, Stockton on Tees, United Kingdom; ⁴Harvard Medical School, Boston, MA; ⁵Massachusetts General Hospital, Boston, MA; ⁶Inha University, Incheon, Republic of Korea; ⁷Massachusetts College of Pharmacy and Health Sciences University, Boston, MA. (Tracking ID #1939988)

BACKGROUND: While evidence suggests that computerized decision support (CDS) increases safety and quality of care, understanding how physicians respond to CDS alerts is a critical factor in achieving meaningful use of electronic health records. We continue to observe a high level of medication alert overrides for many prescription domains. While some overrides are justified clinically, others are not, and it is important to be able to reach out to those providers who are not prescribing optimally and understand their reasons for overriding alerts.

METHODS: All Level 2 alert overrides that required a coded reason to be provided at the time of prescribing were downloaded between January 2009 and December 2011 in the outpatient primary care setting. We limited our sample to providers who had received 20 or more alerts in each of the prescribing domains (drug-drug interaction, drug-allergy interaction, renal suggestion, age-based, duplicate-drug, and formulary substitution alerts) and the number of times each provider overrode these alerts was calculated. Of the 725 providers eligible, those with a high inappropriate override rate (above 75% within a specific domain or overall) were targeted for interviews. The sessions were tailored to each provider’s overrides and conducted by a research pharmacist or physician trained in counter-detailing techniques. Graphical material, including performance level data, a list of their inappropriate overrides and supporting evidence-based summaries, was presented to each provider during the discussion. An analysis of the data was carried out and prevalent concepts related to general views on alert functionality and specific prescribing behavior identified.

RESULTS: We conducted 23 interview sessions across primary care clinics affiliated with Brigham and Women’s Hospital and Massachusetts General Hospital. We identified seven high level content categories: clinical satisfaction, clinical utility and relevance, variant user knowledge, impact on clinician reviewing process, patient preferences, current alerting tool challenges, and considerations for the future. Overall, providers were favorable towards the alerts and found them helpful in identifying possible adverse interactions. Many providers, however, found that the clinical relevance of the alerts could be improved by including magnitude of risk, suggested alternate treatments, recent laboratory values, and more up-to-date reference material. Additionally, many participants said they failed to provide a valid override reason because they found the alerts time-consuming and disruptive. Limitations of the existing EHR infrastructure such as inaccurate medication and allergy lists were identified as contributing to unnecessary alerts. Creating a more patient specific alerting tool was recognized as a future area of focus.

CONCLUSIONS: Many providers were unaware of their relatively high rate of overriding and this study allowed providers the unique opportunity of objectively assessing their prescribing behavior. Key issues that emerged from the sessions included the perceived risk to physician autonomy in decision-making, the increase of clinically irrelevant alerts leading to alert fatigue, and the lack of supplementary clinical information. By incorporating provider preferences, customizing alerts to the context of the visit, and offering additional clinical data, providers felt that CDS alerts would be less likely to be overridden providing more effective, efficient, and equitable care.

AN INTERNATIONAL EVALUATION OF DRUG-DRUG INTERACTION ALERTS THAT SHOULD BE NON-INTERRUPTIVE IN U.K. AND U.S. SETTINGSAlexandra Robertson¹; Sarah P. Slight^1,2; Diane L. Seger³; Sarah K. Thomas⁴; Jamie Coleman⁴; David W. Bates^1,5; Shobha Phansalkar^1,6. ¹Brigham and Women’s Hospital, Boston, MA; ²The University of Durham, Durham, United Kingdom; ³Partners HealthCare Systems, Inc., Wellesley, MA; ⁴The University of Birmingham, Birmingham, United Kingdom; ⁵Harvard Medical School, Boston, MA; ⁶Wolters Kluwer Health, Indianapolis, IN. (Tracking ID #1937914)

BACKGROUND: Clinical Decision Support (CDS) has the potential to improve patient safety by providing knowledge and support at the point of prescribing. CDS systems can, however, deliver an overdose of alerts that may result in “alert fatigue,” with providers overlooking clinically significant alerts as well as those considered less critical. In a previous study, Phansalkar et. al. (2012), identified 33 non-critical drug-class and class-class interaction alerts that could be safely made non-interruptive with a goal to reduce alert fatigue. This study aims to establish whether these non-critical drug-drug interaction (DDI) alerts were set as interruptive or non-interruptive in U.K. and U.S. systems, and the rate at which these alerts were generated and overridden.

METHODS: With the necessary ethical approvals, we downloaded all low priority drug-class and class-class DDI alerts generated from January 2009 to December 2011 in three home-grown electronic health records (EHRs) at two academic medical centers: one British inpatient system and one U.S. inpatient and one outpatient system. We defined interruptive alerts as those requiring a provider response or action when generated and non-interruptive alerts as those that did not require a provider response. In the U.K. system we downloaded all level 3 inpatient alerts, which require a box to be ticked to acknowledge receipt of the alert. No override reasons are collected by this system. In the U.S. systems we downloaded all level 2 alerts, which require a coded reason to be provided at the point of prescribing. The number of alerts generated and overridden were downloaded for all three systems. Using this data, we calculated the frequency of alerts generated and override rates for all low priority DDI alerts.

RESULTS: In the U.S. system, 2 of the 33 low-priority drug-class and class-class interactions were set as interruptive; these included niacin and statins, and proton pump inhibitors and imidazoles. These drug-drug combinations triggered a total of 301 alerts in the U.S. outpatient and inpatient systems, which were overridden in 77.7% of cases (n = 234) during the period assessed. The niacin and statin interaction was triggered most often in both the U.S. inpatient (n = 62) and outpatient (n = 215) systems, and overridden 90.3% of the time in the inpatient system (n = 56) and 70.2% in the outpatient system (n = 151). In the U.K. system, 8 of the 33 low priority drug-class and class-class interactions identified were set as interruptive and triggered a total of 2,354 interruptive alerts, which were overridden in 63.5% of cases (n = 1,495). The most common were: ACE inhibitors and angiotensin II receptor antagonists; anticoagulants and corticosteroids; NSAIDs and β-adrenergic blockers; thiazide-type diuretics and NSAIDs; ACE inhibitors and NSAIDs; and sulfonylureas and ACE inhibitors, with the latter two pairs overridden most often. Sulfonylureas and ACE inhibitors triggered 42.6% of all class-class alerts (n = 1,002) and were overridden in 63.4% of cases (n = 645). Of the sulfonylureas and ACE inhibitors, the gliclazide/gliclazide MR and perindopril alerts accounted for almost half of this class-class category (55.4%, n = 556) and were overridden in 69.9% (n = 387) of cases. ACE inhibitors and NSAIDs occurred second most often (22.3%, n = 525) in the U.K system, with an 87.4% override rate (n = 459). Within this class-class category, ramipril and ibuprofen accounted for more than half (52.4%, n = 275) and were overridden in 90.1% of cases (n = 248).

CONCLUSIONS: Our team has previously shown that certain non-critical alerts may be safely made non-interruptive in provider workflows. This study offers assessment of the non-critical alerts previously identified to better understand user response to these alerts. Given the high rates of override amongst those reviewed, we feel this work offers further validation that these alerts may be safely made non-interruptive to minimize alert fatigue and support the patient safety aims of CDS.

AN INTERNATIONAL EVALUATION OF USER PERCEPTIONS OF DRUG-DRUG AND DRUG-ALLERGY INTERACTION ALERTSAlexandra Robertson¹; Pamela M. Neri²; Elisabeth Burdick¹; Sarah P. Slight^1,3; David W. Bates^1,4; Shobha Phansalkar^1,5. ¹Brigham and Women’s Hospital, Boston, MA; ²Partners HealthCare Systems, Inc., Wellesley, MA; ³The University of Durham, Durham, United Kingdom; ⁴Harvard Medical School, Boston, MA; ⁵Wolters Kluwer Health, Indianapolis, IN. (Tracking ID #1938373)

BACKGROUND: Electronic medical records (EMR) with integrated clinical decision support (CDS) have the potential to improve patient safety. Increasing adoption, refining the delivery and content of existing CDS, and improving user-centered design are key to meeting core measures for Meaningful Use of EMRs. Override rates of CDS alerts are estimated to be between 49 and 96%. Limited evidence supports a direct link between the availability of CDS to measurable improvements in patient safety. Among several reasons, poor user acceptance of alerts is widely recognized as a major barrier to realizing the potential impacts of CDS systems on patient safety. Poor alert design and lack of contextual specificity are causes for negative user perceptions about the utility of CDS. In this study, our objective was to evaluate the relationships between user perceptions of Drug-Drug Interaction (DDI) and Drug-Allergy Interaction (DAI) alert volume, content, and user acceptance.

METHODS: We conducted a survey of user perceptions of DDI and DAI alerts using a validated survey tool, developed by Zheng, et al. The purpose of this survey was to understand the relationship between the acceptance of alerts to users’ perceptions of the number and content of these alerts. To increase the generalizability of our evaluation we included a multi-national sample of EMRs, with users of both home-grown and commercially available products, surveying physicians at seven healthcare institutions in two countries. The EMRs evaluated were two commercially available products and four home-grown products. A total of 1,423 internal medicine physicians were invited to participate in the study. The survey questions evaluated providers’ perceptions regarding the volume of alerts they received. Specifically, providers were asked to quantify the DDI/DAI alerts they receive in an average week, read thoroughly, found clinically relevant, changed their prescribing behavior, and overrode. We performed descriptive statistics on the survey responses to establish correlations between perception of alert frequency, alert relevancy, and alert override. We ran the means in three groups based on the number of alerts reported per week; 1–10 alerts per week (group 1), 11–50 (group 2), and greater than 50 (group 3).

RESULTS: Of the 1,423 physicians invited to participate, 342 consented to participate for an overall response rate of 24%. For both DAI and DDI alerts across all three groups, we found that as the number of perceived alerts increases, the percentage of providers who report reading, finding these alerts relevant, or changing prescribing behaviors based on the information provided decreases while the number of alerts overridden increases (Table 1, 2). Overall, participants estimated receiving a greater number of DDI (22.9) than DAI (13.8) alerts per week, but were more likely to override DAI than DDI alerts, with reported override rates of 83.22% and 78.5%, respectively.

CONCLUSIONS: This is the first study to establish an empirical correlation between the perceptions of physicians on alerts to alert acceptance. Physicians who perceive receiving a high number of alerts are less likely to read them, find them clinically relevant, allow them to affect their prescribing behavior, and more likely to override them. Future research should focus on how providers’ perceptions of alert volume can be improved. Decreasing the volume of interruptive alerts may foster a more positive attitude of physicians towards CDS alerting in EMRs.

AN INTERNET-BASED STUDY CHARACTERIZING THE ASSOCIATION BETWEEN PAIN SEVERITY AND THE BURDEN OF FIBROMYALGIA IN THE USElizabeth T. Masters²; Caroline Schaefer¹; Rachael Mann³; Joseph C. Cappelleri⁴; Shoshana Daniel⁵; Edgar Adams¹; Gergana Zlateva²; Heather McElroy⁶; Arthi Chandran²; Annlouise Assaf⁴; Michael McNett⁷; Philip Mease^8,9; Stuart Silverman¹⁰; Roland Staud¹¹. ¹Covance Market Access Services Inc, Gaithersburg, MD; ²Pfizer Inc, New York, NY; ³Covance Market Access Services Inc, San Diego, CA; ⁴Pfizer Inc, Groton, CT; ⁵Covance Market Access Services Inc, Conshohocken, PA; ⁶Covance Market Access Services Pty Ltd, Sydney, NSW, Australia; ⁷Aurora Pain Program, Milwaukee, WI; ⁸Swedish Medical Center, Seattle, WA; ⁹University of Washington, Seattle, WA; ¹⁰Cedars-Sinai Medical Center, Beverly Hills, CA; ¹¹University of Florida, Gainesville, FL. (Tracking ID #1922702)

BACKGROUND: Although patients with fibromyalgia (FM) report a variety of symptoms that impact function and quality of life, chronic widespread pain is the hallmark symptom. Pain management in patients with FM is challenging, and the presence of pain contributes to the overall burden. This study characterizes the burden of FM by pain severity in the US.

METHODS: Subjects ≥18years old with FM were identified during a study to determine the prevalence and burden of FM and chronic widespread pain in a sample of US adults recruited from a national online panel. Given the gender distribution of FM, females were oversampled (80:20). FM confirmation was made by physician evaluation during a site visit among 350 with and 125 subjects without chronic widespread pain. A total of 171 FM subjects subsequently completed an online questionnaire to assess clinical characteristics, patient-reported outcomes (including but not limited to the Revised Fibromyalgia Impact Questionnaire [FIQ-R], Brief Pain Inventory-Short Form [BPI-SF], Medical Outcomes Study Sleep Scale [MOS-SS], and Short Form 12-Item Heath Survey, version 2 [SF-12]), and direct and indirect costs during the past 3months. Subjects were categorized by pain severity using cutoff scores on the BPI-SF (mild, 0–3; moderate, 4–6; severe, 7–10); statistical significance was evaluated at the 0.05 level across severity levels with ANOVA (continuous variables) and chi-square or Fisher’s exact test (categorical variables).

RESULTS: Mild, moderate, and severe pain were reported by 27 (15.8%), 57 (33.3%), and 87 (50.9%) FM subjects, respectively. Across pain severity levels, age, gender, and race were similar. However, the percentage of subjects employed for pay was lower with increasing pain severity: 70.4%, 45.6%, and 23.0% for mild, moderate, and severe, respectively (P < 0.0001). Among those reporting at least one comorbid condition, greater pain severity was associated with a higher mean ± standard deviation (SD) number of comorbid conditions, from 3.0 ± 2.0 (mild) to 4.5 ± 3.1 (moderate) and 5.8 ± 2.9 (severe) (P < 0.0001). The FIQ-R and its subscales (Function, Overall Impact, and Symptom Intensity) showed progressive worsening with increasing pain severity (all P < 0.0001). The MOS-SS Overall Sleep Problems Index score indicated poorer sleep with increasing pain (P < 0.0001). Pain severity was significantly associated with poorer health status, as indicated on the SF-12 domains (all P < 0.05) as well as the Physical Component Summary (P < 0.0001). Higher total costs during the past 3months were associated with increasing pain (P = 0.0053), with the highest costs among those with severe pain (mean ± SD:$9,553 ± 9,601) relative to moderate ($6,701 ± 6,477) and mild ($4,143 ± 4,657). Across all severity levels, indirect costs were the primary driver of total costs, and increased as pain increased (P = 0.0063).

CONCLUSIONS: Severe pain was common in a broadly representative sample of FM patients, underscoring the challenges of pain management. Pain severity was significantly associated with sleep, function, and health status, indicating that the burden of FM was greater among subjects with more severe pain. The economic burden of FM was higher with greater pain severity; the highest costs were observed among subjects with severe pain. These findings highlight the importance of comprehensive treatment and the need for effective pain management strategies, especially among those experiencing greater pain levels.

AN OBJECTIVE STRUCTURED TEACHING EXAM DEMONSTRATES THAT CLINICAL FACULTY PERFORM BETTER AT FEEDBACK GIVING WHEN INSTRUCTED TO FOCUS ON FEEDBACK SKILLSSheira Schlair¹; Sharon Parish¹; Mooseong Heo²; Felise Milan¹. ¹Albert Einstein College of Medicine/Montefiore Medical Center, Bronx, NY; ²Albert Einstein College of Medicine, Bronx, NY. (Tracking ID #1937665)

BACKGROUND: The Objective Structured Teaching Exam (OSTE) is an emerging faculty development and teaching assessment tool. Often the only data available to assess clinical teaching quality are learner evaluations, which are subject to bias. The objectives of this evaluation were to assess faculty’s areas of effectiveness and those needing improvement on simulated clinical teaching encounters.

METHODS: 18 faculty participants (FP), six expert observers (EO) and 10 standardized students (SS) participated in a three-station OSTE at AECOM’s Clinical Skills Center. Cases and checklists were developed by a team of experts and further modified with input from SSs and EOs after a 3-hour training session. FPs participated in three stations: (1) Attending rounds involving a medical error with a team of an intern, resident and sub-intern (“attending rounds”); (2) Bedside teaching of a focused cardiac exam in a patient with chronic atrial fibrillation (“bedside teaching”); and (3) Learner feedback to a medical student with professionalism issues in his clinical work (“feedback”). FPs were assessed by EOs who completed station-specific checklists consisting of core teaching skills, case specific skills, and two global items (communication and teaching skills) using a behaviorally-anchored 5-point scale (1 = poor to 5 = excellent). Inter-station comparisons were performed using a mixed effects linear regression model followed by a Bonferroni correction, and paired t-tests were used to assess the impact of specific teaching domains on overall station performance.

RESULTS: EO assessment of FP performance showed that core teaching skills were significantly different across stations: (1) FPs’ instructional skill scores were higher in the bedside teaching and attending rounds stations as compared to the feedback station (4.11, 3.83, 2.85, p = 0.0340); (2) FPs’ feedback scores were higher in the feedback station (4.00 vs. 2.97 for attending rounds, 3.81 for bedside teaching; p = 0.0358) and the highest rated teaching skills in the entire OSTE; (3) Case specific skills were significantly worse in the feedback as compared with attending rounds and bedside teaching stations (3.12, 4.03 and 4.33, respectively; p = 0.0014). FPs’ feedback skills had a negative impact on bedside teaching and attending rounds’ overall station performance, while it had a significantly positive impact on the feedback station performance.

CONCLUSIONS: Faculty case specific skill performance was stronger in the attending rounds and bedside teaching cases as compared to the professionalism station. However, feedback skills were superior and most highly rated in the feedback station in which feedback was the specific teaching task being evaluated. These findings raise the question of whether feedback can be effectively delivered while concentrating on a clinical task. One possible explanation for is that that when clinical teachers have multiple tasks to juggle, feedback may not be given priority and perhaps may be performed less skillfully. Implications are that feedback to learners may need to be explicitly separated from other tasks in the daily clinical workflow. In addition, these findings might suggest the need for systematic faculty development in feedback-giving in complex clinical environments.

AN OCCULT FINDING IN HEPARIN DRIP ORDER SETDaniel A. Kim²; Ileana L. Ponor^1,2; Rajanigandhi Hanumanthu³; Scott Wright². ¹Johns Hopkins University School of Medicine, Baltimore, MD; ²Johns Hopkins University School of Medicine, Baltimore, MD; ³Johns Hopkins Bayview Medical Center, Baltimore, MD. (Tracking ID #1939349)

BACKGROUND: Order sets in electronic medical records (EMR) have been shown to reduce variation in clinical practice, to increase quality of care, and to reduce costs. However, it is important to continually evaluate the benefit and appropriateness of existing order sets. Some order sets may not be evidence-based, and all need to be periodically reviewed and updated. There have been efforts to reduce, or even eliminate, fecal occult blood test (FOBT) in hospitalized patients. Friedman’s 2010 study asserts that there is no place for FOBT in an acute hospital setting as the test leads to inappropriate clinical decisions with considerable costs. One example of using FOBT in an ill-chosen and ill-timed manner, is the inclusion of FOBT in the heparin drip order set, which occurs at our hospital and others. Our study was conducted to assess the frequency of FOBT ordered and sent, as well as the results, documentation, and change in management among hospitalized patients who are on heparin drip therapy.

METHODS: A retrospective study of 898 adult hospitalized patients, who had heparin drip ordered at a single academic medical center during a 6month period (from November 2012 to May 2013) was conducted. Demographics, hospital admission information, diagnosis, discharge disposition, laboratory data, and indication for heparin drip were collected. For patients with FOBT sent, further data was gathered and analyzed for FOBT results, documentation, and influence over clinical decision making and management.

RESULTS: During the 6month period studied, 898 patients had heparin drip ordered through the available order set, which includes a nurse collect lab order for FOBT. A total of 128 (14.3%) patients had FOBT sent, of which 33 (25%) came back positive. About half (51.5%) of these results were documented by a provider either in a progress note (36%) or discharge summary (36%). An occult blood test was sent despite the fact that seven patients (21%) had obvious signs of bleeding. In only one instance, was the heparin drips were stopped as a direct result of the positive test. From all positive tests, 7 (21%) resulted in a gastroenterology consult, with four of these patients (57.1%) getting an inpatient endoscopy. These did not change anticoagulation management in a single patient. The endoscopies were unremarkable in three patients (no cause of bleed found) and in one patient adenocarcinoma was diagnosed via a polyp biopsy. Among the 14 patients that were discharged home (n = 14), approximately 43% were readmitted within 90days with a diagnosis of GI bleed.

CONCLUSIONS: Regular review of order sets in our EMRs, with careful consideration of the consequences, is essential. While the automatic FOBT order in the heparin drip order set at our hospital seems ineffective, we were surprised to learn that the order is ignored by our nurses recurrently. Because the FOBT results changed management very infrequently, it is probably not indicated to be included in the heparin order set. That said, any positive FOBT that is noted in the hospital should be passed on to the patient and the primary care physician, particularly given the very high 90day readmission rate for a GI bleed.

ANALYSIS OF 256 CASES OF CLASSIC FEVER OF UNKNOWN ORIGINMasashi Yamanouchi; Toshio Naito; Yuki Uehara; Hirohide Yokokawa; Tomohiro Hosoda; Yukiko Watanabe; Takayasu Shiga; Akihiro Inui; Yukiko Otsuki; Kazutoshi Fujibayashi; Hiroshi Isonuma; Teruhiko Hisaoka. Juntendo University School of Medicine, Tokyo, Japan. (Tracking ID #1937281)

BACKGROUND: Fever of unknown origin (FUO) can be caused by many diseases that vary depending on region and time period. Studies on the causes of FUO in Japan has been small number of cases, often representing regional studies limited to a single hospital. The aim of the present study were <1> Age group, sex, and causes of FUO (%) in all patients. <2> Differences in causes of FUO (%) between patients 65years old and older, and <65years old. <3> Differences in causes of FUO (%) between the periods 1994–2003 and 2004–2012. <4> Number of cases requiring 150days and longer from the time of fever onset for diagnosis.

METHODS: For the present single-institution retrospective observational study, data were extracted from the medical records of patients ≥18years old who met the criteria for classic FUO and were hospitalized between August 1, 1994, when our department was established, and December 31, 2012. Durack and Street’s criteria for classic FUO were used in this study: fever ≥38°C at least twice over a ≥3-week period, without a diagnosis after three outpatient visits or 3days of hospitalization, and no diagnosis of HIV or immunodeficiency prior to the fever.

RESULTS: A total of 256 patients were enrolled in this study, including 131 men and 125 women, with a median age of 55years (range, 18–94years). The most frequent age group comprised patients in their 60s. Patients 65years old and older accounted for 33.2% (85/256) of all patients. The cause of FUO was infection in 27.7% of patients (n = 71), Non infectious inflammatory disease (NIID) in 18.4% (n = 47), malignancy in 10.2% (n = 26), other in 14.8% (n = 38), and unknown in 28.9% (n = 74). The most common single cause was HIV/acquired immunodeficiency syndrome (AIDS) in 17 patients, accounting for 25% of all infections. Among these, 16 patients were <65years old. The next leading causes were malignant lymphoma and infective endocarditis in 12 patients each; with these causes, four patients and 10 patients, respectively, were <65years old. The fourth leading cause was polymyalgia rheumatica in 10 patients; seven were 65years old and older. In patients 65years old and older, compared to those <65years old, percentages of infections (21.7% vs. 30.6%) and other causes (7.2% vs. 18.5%) were lower, whereas percentages of NIID (27.7% vs. 13.9%) and malignancy (14.5% vs. 8.1%) were higher. The percentage of unknown causes was the same (28.9%) in both patients <65years old and 65years old and older. The trend in 2004–2012, compared to 1994–2003, was similar to that in the 1990s in an overseas report. In eight of the 256 patients, 150days and longer were required from the time of fever onset for diagnosis. The longest time was 300days in a case of adult Still’s disease.

CONCLUSIONS: This study reviewed patients with classic FUO evaluated at our hospital department since it was established. Our study found that together with the increase in HIV patients in Japan, HIV/AIDS is now the most common cause of FUO. In addition, the percentage of unknown causes of FUO has increased, now exceeding 20% of cases. As the present study collated data from a long period and reviewed the largest number of FUO cases in Japan to date, these findings will serve as a useful reference for the future care of FUO patients.

ANALYSIS OF PREDICTIVE FACTORS OF BACTEREMIA AMONG INPATIENTS OF GENERAL MEDICINESayato Fukui¹; Yuki Uehara¹; Toshio Naito¹; Teruhiko Hisaoka¹; Osamu Takahashi². ¹Juntendo University School of Medicine, Tokyo, Japan; ²St. Luke’s Life Science Institute, Tokyo, Japan. (Tracking ID #1937325)

BACKGROUND: Criteria to take blood culture from patients with suspicion of severe infection has not been established, and decision depends on each physician’s opinion. Several report investigated predictive factors for positive blood culture results among ICU patients or aged patients, but analysis of predictive factors among patients hospitalized in department of general internal medicine has not been reported yet. This study was performed in inpatient wards of general medicine of Japanese university hospital to clarify the predictive factors of bacteremia and to identify patients from that blood cultures should be taken.

METHODS: This cross-sectional study was performed at the department of general medicine in Juntendo University Hospital, Tokyo, Japan, from January 2011 to December 2012. All inpatients from that blood cultures were taken were included in this study. The results of blood cultures of the patients were collected from the database of the clinical laboratory, and only the first culture results were used for analysis in each patient if blood cultures were taken repeatedly to check the treatment effects or to deny bloodstream infection. Other clinical information was also extracted by chart review; age, sex, height, weight, admission except from their own houses, type of devices used when blood cultures were taken, antimicrobial use within 2weeks, surgery within a month, hemodialysis, use of immunosuppressive agents, active malignant diseases and HIV infection. Body temperature, systolic and diastolic blood pressure, pulse rate, white blood cell count with percentages of neutrophils and lymphocytes, blood urea nitrogen, creatinine, glucose and CRP levels just at or before blood culture sampling were extracted from medical charts, as well. Predictive factors of bacteremia were analyzed using t-test or Chi-square test, followed by multivariate logistic regression model.

RESULTS: A total of 223 patients (M:F = 130:93, 64 ± 19years old) were included in this study and 60 patients (26.8%) had positive blood culture results. Multivariate logistic regression model showed that older age (OR = 1.5, 95%CI = 1.1–2.0, p = 0.017), higher temperature (OR = 2.7, 95%CI = 1.8–4.1, p < 0.01) and higher neutrophil percentage (OR = 1.1, 95%CI = 1.0–1.1, p < 0.01) were the independent predictors of bacteremia. In contrast, other factors were not predictors of bacteremia; for example, CRP was not associated with bacteremia (p = 0.56). Area under the ROC curve of this model was 0.84 (95%CI = 0.79–0.90) .

CONCLUSIONS: Existence of bacteremia can be highly predicted by fundamental clinical information such as age, body temperature and neutrophil percentage. Our results emphasize the importance of taking blood culture from elderly even though they are not febrile, and the importance of ordering differential count of white blood cell in addition to total cell count. To confirm the patient characteristics that have diagnostic and therapeutic benefits from blood culture, further prospective study should be performed with standard criteria and methods for taking blood culture.

ANNUAL SALARY OR HOURLY WAGE? HOW 4TH YEAR MEDICAL STUDENTS CONSIDER FINANCIAL COMPENSATION WHEN SELECTING A SPECIALTYKent DeZee^1,2; Elexis McBee^2,1; Nathalie Paolino^2,1. ¹Uniformed Services University, Bethesda, MD; ²Walter Reed National Military Medical Center, Bethesda, MD. (Tracking ID #1939766)

BACKGROUND: Financial compensation is a well known consideration in medical student selection of their medical discipline, particularly as it relates to primary care specialties. However, the manner in which students define financial compensation is unclear. The purpose of this study is to determine whether or not 4th year medical students view annual salary and hourly wage as equally important when selecting their specialty in order to help guide future research.

METHODS: The authors conducted a cross sectional survey of 4th year medical students from the University of Chicago, the University of Pennsylvania and all 4th year medical students with a military service obligation in the U.S. during the 2011–2012 academic year. Following the residency match, participants were invited to participate in the study via e-mail. The student’s self-reported specialty choice in the residency match was classified into one of four categories: 1) primary care (PC: internal medicine, family medicine, pediatrics, and internal medicine-pediatrics), 2) non-controllable lifestyle specialties (Non-CL: general surgery, neurosurgery, obstetrics and gynecology, orthopedics, and urology), 3) controllable lifestyle, considered PC (CL-considered PC: all other specialties in which the student listed PC as a secondary choice), and 4) controllable lifestyle, did not consider PC (all other specialties in which the student did not list PC as a secondary choice). Using a five-point Likert-type scale (1 = not important at all; 5 = extremely important), respondents rated the importance of two items related to attending salary when choosing their specialty: 1) annual salary earned by civilian (not military) physicians in their chosen specialty and 2) hourly wage of civilian (not military) physicians in their chosen specialty. In addition, the amount of self-reported educational debt and length of military service obligation were collected. One-way ANOVA was used to compare the mean of the importance of annual salary and hourly wage to the four categories of specialty choice, with a Bonferroni multiple comparison test. Effect sizes were calculated to measure the strength of statistically significant associations. T-tests and stratified analyses were conducted to test for confounding by military service and level of debt.

RESULTS: The response rate for the questionnaire was 46% (540/1184). Attending physician annual salary was slightly more important than hourly wage when selecting a specialty (mean annual salary 2.6, mean hourly wage 2.4, p < 0.0001, paired T-test). Students selecting PC rated annual salary as less important than all other students (PC mean 2.3, all other students 2.8, p < 0.0001, small effect size of 0.42). When rating attending hourly wage, the importance was different between all four specialty groups (PC mean: 2.2, Non-CL mean: 2.4, CL-considered PC mean: 2.6, CL-did not consider PC mean: 2.8), with a moderate effect size (0.57) between PC and CL-did not consider PC, and a small effect size (0.40) between both PC and CL-considered PC as well as Non-CL and CL-did not consider PC. The results were not affected by educational debt or military service obligation.

CONCLUSIONS: Fourth year medical students placed slightly more importance on annual salary versus hourly wage when selecting their specialty. However, these factors were equally important for students who chose a controllable lifestyle specialty and did not consider primary care. Future studies of specialty choice should select the type of financial compensation question or intervention (annual salary or hourly wage) based on the research goal and the intended specialty type of the student to be studied.

ANTIBIOTIC SUSCEPTIBILITIES OF URINARY ISOLATES IN OLDER ADULTSTheresa A. Rowe; Lee Lindquist; Abel Kho. Northwestern University Feinberg School of Medicine, Chicago, IL. (Tracking ID #1927517)

BACKGROUND: Background: Older adults are at an increased risk for development of multi-drug resistant organisms because of compounded exposure to antibiotics. Clinicians empirically treat suspected urinary tract infections (UTI) in older adults, selecting antibiotics based on guidelines intended for younger patients. No prior studies have examined whether there are age-related differences in UTI antibiotic susceptibility. We sought to identify antibiotic susceptibility patterns of urinary isolates in outpatient older adults and compare them to younger patients over a 3-year period.

METHODS: Retrospective analysis of all adult outpatients with a positive urine culture who received antibiotic treatment for UTI within 3days. The most common organisms Escherichia Coli (E. Coli) and Klebsiella spp. were tested for resistance to commonly prescribed antibiotics. We compared resistance patterns of adults over the age of 65 to younger adults under the age of 65 using descriptive statistics and chi square analysis.

RESULTS: Urine cultures from 8,659 patients were reviewed. For E. Coli the percentage of resistance for adults ≥65, compared to adults <65 were: Ciprofloxacin (31.97% in 65years and older patients vs. 16.85% in under 65year old patients; p < 0.01), Trimethoprim/Sulfamethoxazole (30.12% vs 25.19%; p < 0.01), Ampicillin/Sulbactam (19.47% vs 19.36%), Cefuroxime (5.27% vs 4.23%; p < 0.05), Nitrofurantoin (1.80% vs 1.47%). For the second most common organism, Klebsiella species: Ciprofloxacin (5.28% vs 4.99%), Trimethroprim/Sulfamethoxazole (12.61% vs 13.59%), Ampicillin/Sulbactam (6.88% vs 12.85%; p < 0.01), Cefuroxime (6.39% vs 7.91%), Nitrofurantoin (26.28% vs 24.36%)

CONCLUSIONS: Older adults with UTIs have different resistance patterns to microorganisms when compared to younger patients. Current clinical guidelines for treatment may not reflect these differences. Clinicians need to be aware of the resistance patterns unique to older adults so that they can be appropriately treated

ANTICOAGULATION-RELATED PROCESSES OF CARE AND SHORT-TERM OUTCOMES IN ELDERLY PATIENTS WITH ACUTE VENOUS THROMBOEMBOLISMCharlene Insam¹; Marie Méan¹; Andreas Limacher²; Nicolas Rodondi¹; Drahomir Aujesky¹. ¹Bern University Hospital, Bern, Switzerland; ²University of Bern, Bern, Switzerland. (Tracking ID #1937453)

BACKGROUND: Whether recommended anticoagulation-related processes of care are associated with improved clinical outcomes in elderly patients with acute venous thromboembolism (VTE) remains uncertain.

METHODS: We studied 991 in- and outpatients aged ≥65years with acute VTE in a prospective multicenter Swiss cohort study (09/2009–03/2012). We assessed the performance of the following processes of care recommended by the American College of Chest Physicians: use of low-molecular-weight heparin or fondaparinux rather than unfractionated heparin in patients with deep vein thrombosis or non-massive pulmonary embolism, parenteral anticoagulation (PAC) ≥5days, start of oral anticoagulation (OAC) within 24h of VTE diagnosis, and achievement of an international normalized ratio (INR) ≥2.0 for ≥24h before stopping PAC. Outcomes were overall mortality, VTE-recurrence, and major bleeding at 6months and the length of hospital stay. We assessed the association between processes of care and clinical outcomes using Cox regression models, and the association between processes of care and length of hospital stay using the lognormal survival model, adjusting for multiple patient baseline characteristics.

RESULTS: Overall, 9% of patients died, 2% had VTE- recurrence, and 7% had major bleeding at 6months after the index VTE. The median length of hospital stay was 8.0days (interquartile range 5.0; 11.0). Starting OAC within 24h of VTE diagnosis was associated with a lower risk of overall mortality (adjusted hazard ratio [HR] 0.44, 95% confidence interval [CI] 0.21–0.92) and major bleeding (adjusted HR 0.35, 95% CI 0.18–0.66) and a decreased length of stay (adjusted time ratio [TR] 0.77, 95% CI 0.69–0.85). While the use of low-molecular-weight heparin or fondaparinux significantly decreased the length of stay (adjusted TR 0.87, 95% CI 0.77–0.97), the achievement of an INR ≥2.0 for ≥24h before stopping PAC increased the length of stay (adjusted TR 1.20, 95% CI 1.08–1.34). None of the processes of care were associated with VTE recurrence. We found no association between PAC for ≥5days and outcomes.

CONCLUSIONS: In elderly patients with acute VTE, two out of four recommended processes of care were associated with improved clinical outcomes and/or a decrease in length of hospital stay. These processes should be implemented when treating elderly patients with acute VTE.

ANTIPSYCHOTIC MEDICATION USE AND THE RISK FOR HOSPITAL-ACQUIRED PNEUMONIAShoshana J. Herzig¹; Michael B. Rothberg³; Jamey R. Guess¹; Long H. Ngo¹; Jerry H. Gurwitz⁴; Edward R. Marcantonio^1,2. ¹Beth Israel Deaconess Medical Center, Boston, MA; ²Beth Israel Deaconess Medical Center, Boston, MA; ³Cleveland Clinic, Cleveland, OH; ⁴UMass Memorial Medical Center, Worcester, MA. (Tracking ID #1933568)

BACKGROUND: With the advent of atypical antipsychotic medications, use of antipsychotics has markedly increased in long-term care facilities, often in patients without a psychiatric diagnosis. Both typical and atypical antipsychotics have been associated with pneumonia in community and nursing home settings. However, rates of antipsychotic use in the acute hospital setting and the association between this use and hospital-acquired pneumonia (HAP) are unexplored.

METHODS: We studied an observational cohort of all patients admitted to a large academic medical center in Boston, Massachusetts from 1/2007 to 7/2013, at least 18years of age, and hospitalized for at least 2days. We excluded transfers from outside hospitals, patients admitted to the psychiatry service, and patients with a diagnosis of a psychotic disorder. Data were collected from electronic medical information databases at the medical center. We defined antipsychotic use as any pharmacy charge for an antipsychotic medication, subclassified as typical and atypical. The primary outcome was HAP, defined as any ICD-9 for bacterial pneumonia coded as not present on admission. We used a generalized estimating equation logistic regression model to control for repeated admissions and confounders. Our model included 40 variables, chosen based on hypothesized associations with antipsychotic use and/or HAP, including demographics, comorbidities as defined by Elixhauser et al., and concurrent medication use (sedatives, opioids, immunosuppressive agents, acid-suppressive medications). We assessed for effect modification by age using a dichotomous variable where we varied the cutpoint (< versus ≧ 60, 65, 70, 75years).

RESULTS: Our cohort included 146,680 admissions (median age = 56years; 39% male). Antipsychotics were used in 18,047 (12.3%) admissions (8.0% typical and 5.8% atypical, with 1.5% exposed to both), and HAP occurred in 2,992 admissions (2.0%). See table for rates of HAP and adjusted odds of HAP by exposure category. Use of each subclass of antipsychotic medication was associated with increased odds of hospital-acquired pneumonia, with highest risk seen in those admissions with exposure to both classes. There was no effect modification by age at any of the chosen cutpoints.

CONCLUSIONS: Antipsychotic use was common during hospitalization, occurring in 12% of non-psychiatric admissions. Both typical and atypical antipsychotics were associated with increased odds of hospital-acquired pneumonia after extensive adjustment for patient characteristics and concurrent medication exposures. Given their associated risks, further scrutiny is warranted regarding prescribing practices of antipsychotic medications in the acute hospital setting.

Rates of HAP and adjusted odds of HAP by exposure category

*Categories are mutually exclusive

ANTIRETROVIRAL DRUG EXPENDITURE, PRICING AND JUDICIAL DEMAND: AN ANALYSIS OF FEDERAL PROCUREMENT DATA IN BRAZIL FROM 2004–2011Jing Luo¹; Claudia Osorio-de-Castro²; Maria Olivera²; Mariana Ramos³; Aurelio Maia³. ¹Yale University School of Medicine, New Haven, CT; ²Sergio Arouca National School of Public Health, Rio de Janeiro, Brazil; ³Ministry of Health, Brasilia, Brazil. (Tracking ID #1942220)

BACKGROUND: Previous studies have described expenditures for antiretroviral (ARV) medicines in Brazil through 2005. While prior studies examined overall expenditures, they have not have analyzed drug procurement data in order to describe the role of court litigation on access and pricing.

METHODS: ARV drug procurement from the years 2004–2011 was obtained through the general procurement database of the Brazilian Federal Government (SIASG). Procurement was measured in Defined Daily Doses per 1000 persons-under-treatment. Expenditures and costs were calculated and expressed in US Dollars. Justifications for ARV purchases were examined in order to determine the relationship between health litigation and incorporation into Brazil’s national treatment guidelines.

RESULTS: Drug procurement of ARVs from purchased sources underwent marked expansion in 2005, peaked in 2009, and stabilized to 2008 levels by 2011. Expenditures followed procurement curves. Newer medications were on average four times more expensive than older medications. Judicial actions initially resulted in purchases of newer mediations for a select number of patients in Brazil but ultimately expanded availability to a larger population through incorporation into the national treatment guidelines. ARVs were purchased at the lowest prices through auction, intermediate prices through single suppliers, and highest prices through emergency procurement resulting from judicial action.

CONCLUSIONS: Drug procurement and expenditures for ARVs in Brazil varied between 2004 and 2011. Medications which once represented a large proportion of federal drug expenditures have been largely eliminated due to domestic production. Judicial demand has resulted in the procurement of large quantities of newer, more expensive medications. In order for the AIDS treatment program to remain sustainable, efforts should be pursued to reduce prices through price negotiation and other public health flexibilities.

Figure 1. Drug Procurement and total expenditures for antiretroviral medicines in Brazil from 2004 and 2011.

APPROPRIATENESS OF PREVENTIVE CARE FOR ANGIOPATHIC COMPLICATIONS IN OUTPATIENTS WITH DIABETESIdris Guessous^1,5; Sophie Excoffier Bottaro¹; Manuel R. Blum²; Nicolas Rodondi²; Jacques Cornuz³; Lukas Zimmerli⁴; Jean-Michel T. Gaspoz¹. ¹Geneva University Hospitals, Geneva 14, Switzerland; ²Inselspital, University of Bern, Bern, Switzerland; ³University of Lausanne, Lasuanne, Switzerland; ⁴University Hospital of Zurich, Zurich, Switzerland; ⁵Institute of Social and Preventive Medicine, Centre Hospitalier Universitaire Vaudois, University of Lausanne, Lausanne, Switzerland. (Tracking ID #1937359)

BACKGROUND: Preventive care of major diabetic angiopathic complications (foot ulcers and retinopathy) is essential. However, data on the prevalence of and factors associated with preventive care of angiopathic complications in outpatients with diabetes are very limited and non-existent within the Swiss context of universal health care coverage. We aimed to determine the prevalence of appropriate preventive care of angiopathic complications in outpatients with diabetes, and to identify factors associated with appropriate preventive care of angiopathic complications.

METHODS: We used data from a retrospective cohort study (the Corif study) that examined recommended preventive care in a random sample of patients aged 50–80years followed over 2years (2005–2006) in four Swiss university primary care settings using RAND’s Quality Assessment Tools indicators. We defined appropriateness of preventive care of angiopathic complications using the American Diabetes Association recommendations: annual foot (pallesthesia and visual exam) and ophthalmologic exam. Age-, sex-, center-adjusted regression models were conducted.

RESULTS: Among the 292 outpatients with diabetes included, 59.2% had appropriate preventive foot care and 55.8% had appropriate preventive eye care. Only 34.6% had appropriate preventive care of both foot and eye. Number of GP visits (Odds Ratio = 1.07, 95%CI 1.02–1.11) and Hb1Ac measured at least 2×/year (OR = 2.56, 1.34–4.90) were associated with appropriate foot preventive care. Obesity (OR = 0.37, 0.21–0.65) and hyperlipidemia (OR = 2.47, 1.38–4.43) were associated with appropriate eye preventive care. Hb1Ac measured at least 2×/year (OR = 2.56, 1.34–4.90), obesity (OR = 0.53, 0.29–0.95), and hyperlipidemia (OR = 2.36, 1.23–4.51) were associated with appropriate preventive care of both foot and eye.

CONCLUSIONS: Preventive care of diabetic angiopathic complications is frequently inappropriate among outpatients. Particular attention should be paid to obese patients with diabetes.

ARE PATIENT-CENTERED MEDICAL HOME PRACTICES ASSOCIATED WITH BETTER ACCESS TO PRIMARY CARE SERVICES?Jaya Aysola¹; Karin V. Rhodes²; Daniel Polsky¹. ¹University of Pennsylvania, Philadelphia, PA; ²University of Pennsylvania, Philadelphia, PA. (Tracking ID #1940136)

BACKGROUND: Patient-centered medical homes (PCMH) have gained prominence as models designed to promote high quality cost effective primary care. An integral component of the PCMH model is providing enhanced access to primary care. However, the potential effect of PCMHs on access to primary care services has not been examined. Our study evaluated whether PCMH practices in ten US states were associated with better access to care, measured using simulated patient (audit) methodology.

METHODS: We conducted a cross-sectional analysis of publically available data on practices recognized as National Committee Quality Assurance (NCQA) patient-centered medical homes (PCMH) before June 2013 and data collected through a simulated patient (audit) study. Our primary predictor was whether or not a practice was a NCQA recognized PCMH. Our primary outcome measure was the ability of the caller (simulated patient) to schedule a new appointment. Trained field staff contacted primary care offices in 10 US states between November 2012 and March 2013 posing as individuals seeking a new patient appointment with varying insurance status (private, Medicaid, or uninsured). All eligible primary care offices were identified using a proprietary database of existing physician clinics (SK&A physician file). Callers contacted a stratified random sample of primary care practices treating non-elderly adults in each state, for a total of 10572 requests for appointments at 7266 practices. We merged detailed practice characteristics (number of sites, number of providers, average patient volume, number of exam rooms) from SK&A physician file, and geographic socioeconomic and health care resource information (county-level percent persons below poverty level, percent minority, percent non-elderly uninsured, number of office based primary care providers, and Health Professional Shortage Area designation) from the Area Resource File, into our audit data. We performed multivariable logistic regression model, adjusting for the caller’s insurance status and race/ethnicity as well as the practice and geographic characteristics listed above, to estimate the association between PCMH status and our outcome variable. We then fitted a multivariable logistic regression model, including interactions of practice PCMH status and caller’s insurance status, adjusting for the same covariates above.

RESULTS: Out of 10572 requests to practices for appointments, 646 (6.1%) were to NCQA recognized PCMHs. Significant predictors of scheduling new appointments included the caller’s insurance status, number of practice exam rooms, percent minority and percent persons in poverty where the practice was located, and the practice’s NCQA PCMH recognition status. There were no significant interactions between practice PCMH status and caller’s insurance status. Adjusted analysis revealed that calls placed to PCMH practices were significantly more likely to result in scheduled new appointments then calls placed to non-PCMH practices. (Adjusted OR 1.3; 95% CI 1.0, 1.5; p = 0.02).

CONCLUSIONS: Our findings suggest that practices recognized as patient centered medical homes, after robust adjustment for individual caller, practice, and geographic variables, were associated with better access to new primary care appointments for non-elderly adults, those most likely to gain insurance under the Affordable Care Act.

ARE PATIENTS A RELIABLE SOURCE FOR ADVERSE EVENT REPORTING? SURVEY SAYS…Kristy Deep; Megan Minch. Univ of Kentucky, Lexington, KY. (Tracking ID #1939331)

BACKGROUND: Learning from adverse events is an important way to improve healthcare quality and safety. However, events are underreported by healthcare providers. We sought to determine whether patient comments on the post-discharge Healthcare Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey reliably identify adverse events or errors that occur during the hospital stay.

METHODS: HCAHPS surveys with negative comments returned during a 7month period were reviewed. Comments that described a specific adverse event or error were then analyzed by two independent raters. The patient’s medical record was reviewed to corroborate the event. A coding scheme was developed iteratively from the data to categorize the types of adverse events reported by patients. One patient comment could receive multiple codes. We assigned a four-tier harm score for each adverse event: significant harm and/or need for additional treatment, potential harm but outcome unknown, emotional distress/inconvenience, and no harm.

RESULTS: Ninety comments were reviewed, 20 of which were adjudicated as complaints which did not identify quality issues leaving 70 comments for analysis. Of these, 29 (41%) were corroborated in the medical record. An additional 35 (50%) were not corroborated by the medical record but were deemed credible based on the level of detail within the patient’s report or when the described event was unlikely to be documented (i.e. delays in care). Communication errors were most common and found in 27 events. These included communication between healthcare providers and patients as well as among providers. Other frequent event types were patient education deficits (13), delays in clinical care (13), and medication errors (8). Patients also identified procedural complications (3) and lapses in infection control (2). Ten events (14.2%) resulted in significant patient harm and/or required additional treatment. Examples include a failure to recognize an abscess resulting in emergent surgery and a patient developing apnea following administration of a medication. Three of these significant harm events were deemed credible but not documented in the medical record. An additional 10 described situations that were potentially harmful but the impact on the patient was unknown. The remainder of the adverse events or errors resulted in either emotional distress/inconvenience (30) or no harm (20).

CONCLUSIONS: Patient comments on the HCAHPS post-discharge survey are a reliable source for insight into the quality of care patients receive. They may also identify significant issues that are not captured by other safety reporting mechanisms.

ARE THERE RACIAL DISPARITIES IN QUALITY OF TREATMENT FOR LOCALLY ADVANCED PROSTATE CANCER?Rajwanth Veluswamy^1,3; Rebeca Franco²; Kezhen Fei²; Nina Bickell^1,2. ¹Icahn School of Medicine at Mount Sinai, New York, NY; ²Icahn School of Medicine at Mount Sinai, New York, NY; ³Icahn School of Medicine at Mount Sinai, New York, NY. (Tracking ID #1940153)

BACKGROUND: African American men are more likely to develop and die of prostate cancer than white men. Past studies have shown lower rates of invasive treatment among black men. However, these studies included early-stage disease for which treatments may not be beneficial. We undertook this study to assess the quality of prostate cancer treatment among men with locally advanced cancer who could benefit from active therapies.

METHODS: All black and a random sample of white men with Gleason Scores of 7+, diagnosed at an inner-city academic center between 2007 and 2012, were identified from pathology and charts abstracted for clinical, pathologic and treatment data. A Steering Committee of experts in prostate cancer care created quality measures of treatment for locally advanced prostate cancer. Underuse of definitive treatments (radical prostatectomy and radiotherapy) was identified using these quality measures.

RESULTS: Overall, 290 black and 255 white men were identified. The average age of the entire cohort was 59years old (standard deviation [sd] = 8.6, range: 32 to 86) and there was no statistical age difference between races. Black men had more comorbidities (30% vs 9%; p < 0.001) and higher prostate-specific antigen (PSA) levels prior to biopsy (12.4 vs 6.9; p = 0.0027), but whites were more likely to present with stage three disease (30% vs 22%; p = 0.02). Using D’Amico risk criteria, 14% of patients were found to have intermediate risk, while 86% were high risk. The distribution of D’Amico risk between race was not statistical different. Black men were less likely than white men to undergo radical prostatectomy (79% vs 91%, p < .0001), but more likely to get external beam radiation therapy (EBRT, 13% vs 5%; p = 0.0009) or brachytherapy (11% vs 4%; p = 0.0037). Of note, 85% of black and 100% of white men received the recommended radiotherapy (RT) dose (74Gy). Only 50% of men undergoing RT received androgen deprivation therapy (ADT), with no racial difference in ADT use. Overall, 5% of men with locally advanced prostate cancer did not undergo surgery or RT, with no racial difference in underuse. Multivariable analysis using logistic regression showed that patients with intermediate risk were 20 times more likely to experience underuse of definitive treatment as compared to patients with high risk (95% CI: 7.2-50.3). Additionally, every year increase of age was associated with a 1.06 times higher risk of underuse of definitive treatment (95% CI: 1.01–1.12). However, even after accounting for potential confounders, race did not play a role in whether patients received definitive treatment according to quality measures.

CONCLUSIONS: The rate of underuse of definitive treatment in men with locally advanced prostate cancer is low and does not appear to be driven by racial differences. However, underuse is significantly associated with patients of intermediate risk and advancing age. We will further investigate why these patients are not receiving active treatment.

ASKING THE PATIENT ABOUT PATIENT-CENTERED MEDICAL HOMESJaya Aysola^1,2; Rachel M. Werner¹; Judy A. Shea¹. ¹University of Pennsylvania, Philadelphia, PA; ²Children, Philadelphia, PA. (Tracking ID #1939454)

BACKGROUND: To date the patient perspective is not an integral component of any of the existing patient-centered medical home (PCMH) measurement standards, although upcoming guidelines necessitate patient-reported experience and satisfaction data in evaluations of practice and provider performance. Despite this, patient understanding of PCMH and their perspectives on their role in the evaluation of practices implementing this model is not known. Therefore, we sought to characterize patients’ understanding of the PCMH model, experiences with PCMH practices after transformation, and perceptions on approaches to incorporate patient feedback.

METHODS: We conducted semi-structured interviews of patients from NCQA (National Committee Quality Assurance) recognized PCMH practices at University of Pennsylvania (UPENN). We first surveyed, scored, and ranked all UPENN affiliated PCMH adult primary care practices with three or more providers (n = 23) using a previously validated scale that quantifies the overall degree of PCMH practice transformation. We then identified the practices, representing the four highest and lowest PCMH transforming practices. To determine if the degree of PCMH practice transformation was related to patient responses, participants were recruited from a pool of adult patients with hypertension and/or diabetes from these eight selected practices. We conducted semi-structured, one-on-one telephone interviews with patients consisting of both open- and closed-ended questions about their experiences with care, and perceptions and understanding of key PCMH domains. Interviews were recorded, transcribed, and imported into NVivo 10 for coding and analysis, using a modified grounded-theory approach.

RESULTS: Of the 19 interviews completed, 52% were from high-transforming practices, 48% from low. Fifty-seven percent of the sample was African American and the mean age was 61. Responses did not appear to vary between patients from high and low PCMH transforming practices. None of the participants were aware that their clinic was a PCMH, regardless of whether they were from a high- or low-PCMH transforming practice and most did not know what a PCMH was. Of the key aspects of the PCMH, most participants identified a team-based approach and strong patient-provider partnership as more important to them than enhanced access to care. The majority could not identify additional personnel involved in assisting their PCP in their care. Respondents had differing perspectives about their own role in decision making, with many relying on their PCP to make their treatment decisions. Emerging themes included the following: 1) Patients were unaware of what a medical home is and if their practice was one, irrespective of the practices’ degree of PCMH transformation. 2) The patients’ lack of knowledge regarding PCMH concepts did not appear to impact their experiences with care and their uniformly positive relationship with their primary care provider. 3) The PCMH domains patients reported as most important for care delivery in theory did not always relate to their preferences in reality. For example, participants reported team-based care as one aspect of PCMH most important to them, yet when discussing their specific experiences at their practice, the majority could not recall any additional personnel teaming with their provider to participate in their care. The majority of participants also reported a strong patient-provider partnership as the other aspect of primary care delivery most important to them; however, when detailing specific preferences, many were less interested in partnering with their provider to share responsibility in medical treatment decisions.

CONCLUSIONS: Our study aimed to examine patient understanding and perceptions of care within PCMHs and whether their experience varied depending on the degree of PCMH practice transformation. We found that patients uniformly lacked awareness of PCMH concepts and the fact their practice was a PCMH, regardless of the degree of transformation reported by the practice. Moreover, this lack of understanding did not appear to impact their experiences with care or overwhelmingly positive relationships with their providers. As we continue to redesign primary care delivery with an emphasis on patient experience measures as performance metrics, further evaluation is needed to determine what if any aspects of how we structure practices relates to patient experience and satisfaction with care.

ASSESSING CONTENT AND QUALITY OF ADVANCE CARE PLANNING DOCUMENTATION IN AN INTERNAL MEDICINE RESIDENT CLINICKatherine A. Wang¹; Jacqueline K. Yuen^2,3. ¹Icahn School of Medicine at Mount Sinai, New York, NY; ²Icahn School of Medicine at Mount Sinai, New York, NY; ³Icahn School of Medicine at Mount Sinai, New York, NY. (Tracking ID #1913261)

BACKGROUND: Despite widespread recognition of the need for advance care planning (ACP), there is a troubling communication gap between patients and physicians. One large multicenter study found that of patients who preferred not to be resuscitated, only 52% had DNR orders. Primary care physicians, given their longitudinal relationships with patients, are ideally positioned to elicit patients’ values and goals. Meaningful documentation of these conversations is essential to ensure appropriate communication with other health care providers and to guide future decision-making. Educational interventions have been shown to improve residents’ knowledge, skills, and comfort level with ACP and to increase documentation of advance directives, but little data exists about improving the quality of documentation. A novel month-long curriculum was piloted with PGY-2 Internal Medicine (IM) residents from January-June 2013 to improve resident skills with ambulatory ACP discussions and documentation. The objective of this study was to assess the impact of an educational intervention and electronic medical record (EMR) template on the content and quality of documentation of ACP discussions.

METHODS: The setting was an urban academic ambulatory resident clinic. Participants were 41 PGY-2 IM residents who each identified patients with at least one chronic illness and a high likelihood of hospitalization within a year. Residents held visits to address ACP with these patients during the one-month curriculum and documented conversations. An EMR template including prompts for topics introduced in the curriculum (e.g. health care proxy (HCP), disease trajectory, values) was developed for optional use by residents. Investigators conducted a retrospective chart review of all identified patients and measured the frequency of certain discussion elements such as HCP and prognosis. Fisher’s exact test was used to compare the templated and free-text notes. Twenty reviewers from IM and Emergency Medicine will rate the notes for usefulness in clinical decision-making in a hypothetical case scenario using a Likert scale, and a qualitative analysis comparing elements of the most highly-rated notes will be performed.

RESULTS: 41 residents received the ACP curriculum. Twenty-nine residents (71%) documented a total of 34 ACP discussions (Table). The other 12 residents did not document an ACP discussion because the patient did not show up (n = 2) or for unknown reasons (n = 10). Of the 34 patients, the mean age was 71.8years (SD 13.0), 65% were female, and the most common conditions were hypertension (91%), diabetes (62%), chronic kidney disease (32%), CHF (24%), COPD (18%), CAD (18%), and cirrhosis (12%). Seventy-one percent had a readmit risk score of high or very high (based on number of hospitalizations in last 6months and certain high risk diagnoses). Of the 34 ACP notes, 6 (18%) used the template and 28 (82%) did not. HCP was nearly always addressed (97% of all conversations), and a signed HCP form was scanned into the chart frequently (56% of all conversations). Twenty-four percent of notes documented a discussion of goals of care; this was more likely with residents using a template (83% vs 11%, p = 0.001). Qualitative analysis results are pending.

CONCLUSIONS: This study assessed the impact of an educational intervention and EMR template on the content and quality of ACP documentation by IM residents. Regardless of template use, certain discussion elements such as HCP were addressed in nearly all ACP notes, while disease trajectory and prognosis were infrequently addressed. However, residents who used the template were significantly more likely to explore goals of care with patients. The limitations of this study are the lack of a control population who did not receive the curriculum and the small number of residents who opted to use the template. Analysis of the clinical usefulness of ACP documentation is still pending and will identify characteristics of notes that are most helpful for future decision-making. Findings will contribute to future iterations of the resident curriculum to improve ACP communication and documentation skills.

Comparison of ACP discussion elements between templated and free-text notes

* E.g. values, activities, fears, and prior experiences with illness. In the template, prompts include questions such as: “What is important to you in life? What activities or experiences are important for you to live well? What fears or worries do you have about your health?” ^ E.g. code status, artificial nutrition/hydration

ASSESSING LONG TERM OUTCOMES OF A TEACHING COMMUNICATION SKILLS COURSE FOR FELLOWS AND FACULTYSarah B. Merriam; Robert M. Arnold; Carla Spagnoletti. University of Pittsburgh Medical Center, Pittsburgh, PA. (Tracking ID #1936910)

BACKGROUND: The ability to communicate effectively with patients is a key skill for practicing physicians. The importance of effective patient-doctor communication has received increasing attention in the field of medical education. Since 2003, the Clinician Educator Training Program at the University of Pittsburgh has offered a master’s level course entitled “Teaching Communication Skills” designed to provide participants with the skills needed to effectively teach clinicians communication skills. This study’s objective is to evaluate how this course influenced participants’ subsequent teaching of communication skills and related scholarly endeavors.

METHODS: The 65 fellows and faculty who completed the “Teaching Communication Skills” course between 2003 and 2011 were surveyed electronically. The survey assessed participants’ perceived importance and effectiveness of the course and its instructional methods. It also evaluated the use of these teaching skills in their current teaching practice. Demographic characteristics of the respondents as well as their career descriptions and academic productivity were assessed. Descriptive statistics were used to report survey responses.

RESULTS: A total of 53 of 65 course participants completed the survey (response rate 80%). When asked why they took the course, 38% indicated that it was highly recommended by peers and 85% indicated that they anticipated a future role as a clinician educator and thought the course would be helpful in attaining this goal. Almost all (98%) rated both the course and the individual communication-specific teaching skills taught in the course to be effective in the development of their teaching skills (on a 5-point Likert-type scale where 1 = ineffective and 5 = effective). When participants were asked to list which of the course’s instructional methods were most beneficial in developing their ability to teach patient-doctor communication skills, most frequent responses included: exposure to the “primary teaching method” (a learner-centered small group facilitation method commonly used to teach communication skills; n = 14), the process of direct observation and feedback (n = 22), and small group role play with standardized patients (n = 52). All instructional methods taught in the course were reported to be increasingly utilized in their teaching practice since taking the course, with 77% using standardized patients, 89% using small group work, and 83% using role-play at least annually. Forty-nine (93%) graduates described their current position as academic. Of those individuals, 72% described roles as clinician-educators, 11% as clinician-researchers, and 6% as clinician-administrators. Since taking the course, 72% of respondents have either developed or taught in a communication curriculum. Thirteen (25%) have held a leadership or administrative position directly related to patient-doctor communication. Forty-four (83%) have since pursued research or other scholarly activity related to communication: 10 obtained grant funding, 28 presented their work at a regional or national conference, and 21 wrote peer-reviewed manuscripts.

CONCLUSIONS: This novel course was highly rated by the fellows and faculty who participated. A majority of course participants went on to design or teach communication curricula. Courses such as this one, designed to instruct educators how to teach communication skills, may be valuable to medical schools and other training programs that want to improve their faculty’s ability to effectively deliver this curricular content to trainees.

ASSESSING PATIENT ATTITUDES AND PRACTICES REGARDING INFORMATION TECHNOLOGY IN A PATIENT CENTERED MEDICAL HOME: ARE WE READY FOR TELEMEDICINE?Jeanne Morley^1,2; Daniel J. Coletti^1,2; Laura Harrison¹; Joseph Conigliaro^1,2. ¹North Shore LIJ Health Systme, Great Neck, NY; ²Hofstra North Shore-LIJ School of Medicine at Hofstra University, Hempstead, NY. (Tracking ID #1936175)

BACKGROUND: Delivering care in a Patient Centered Medical Home (PCMH) requires an understanding of patient comfort with and access to information technology (IT), as use of that technology in the general population evolves and expands. According to a recent Pew Research Center report, 56% of Americans owned a smartphone and 34% used a tablet computer. Nearly 165 million Americans use social media, and as of 2012, one in three cell phone owners reported using their phones to look up health information. While email, text messaging, video chat, and social media have transformed how Americans communicate, most medical practices have failed to keep pace with these changes. Although about one third of Americans prefer getting text messages to phone calls and almost three quarters use email daily, the majority of physician offices still exclusively communicate by telephone with patients to make appointments, review lab results and answer questions. While clearly desirable, implementing new models of patient communication requires an understanding of patient access to technology. This is particularly salient in primary care settings that treat the underserved, who may lack access to novel and expensive forms of technology. The objectives of this investigation were to describe patient patterns of using IT in general, and to examine patient preferences and attitudes to using IT as part of their health care. We also wished to assess whether attitudes were associated with respondent age and sex.

METHODS: We completed a descriptive study as part of a larger feasibility study to prepare for deployment of internet-based patient communications within a large PCMH. The study was approved by our Institutional Review Board. We created a 12-item questionnaire entitled Patient Attitudes to Information Technology (PAIT) to assess the frequency of use of communication modalities such as email, text messaging, video chat, and attitudes towards IT. Our sample included patients seen for routine visits, complete physical exams, and sick calls during July and August of 2013. Questionnaires were completed in the office waiting room prior to a medical appointment. The office is a busy academic primary care practice, and designated Level III PCMH. Our patient population is diverse, and spans a wide range of educational and socioeconomic strata. Roughly half have Medicare or private insurance, and the other half are uninsured or have Medicaid. The questionnaire was offered to all patients on a voluntary basis. No protected health information was collected. Chi-square analyses, T-tests for independent means and Analysis of Variance (ANOVA) were deployed as appropriate to test for differences in attitudes according to demographic variables.

RESULTS: A total of 215 questionnaires were completed. Mean age of the sample was 48.29years (SD = 18.26, range = 18–90) and was 62.5% female. On the day of their visit, 50.7% of the patients surveyed had already used a PC, 55.5% had used email, 79.2% made a mobile phone call, 59.9% had sent or received a text message, and 18.6% had used a video chat application. When assessment was expanded to the past week, the numbers increased to 78.4%, 75.9%, 91.5%, 72.3%, and 49.1% respectively. Patients were also asked to score their agreement with four statements regarding IT on a scale of 0–10. On average, our patients reported moderate-to-high levels of agreement that they “have a lot of experience using smart phones” (mean = 6.01 out of 10), and were “the kind of person who like trying new electronic devices (mean = 6.86 out of 10). We also found high levels of agreement with the statement, “I would be interested in learning more about ways to use information technology…to communicate with my medical team,” (M = 7.0 out of 10). Although patterns of use and comfort with technology were high overall, statistically significant age-related effects were observed across all attitudinal variables (p < .001 for all analyses), with younger respondents more likely to express higher levels of comfort and interest in IT.

CONCLUSIONS: Results from this investigation suggest high levels of both readiness and interest among the patients in our PCMH to use IT to augment current communication with their medical team. Observed age-related effects suggest that older patients, while generally interested in using IT to facilitate healthcare communication, may need additional support to increase their knowledge and comfort level prior to implementing new forms of interaction with their healthcare team. We were particularly surprised and encouraged by the frequent use of video-based applications among our respondents. Using this modality to facilitate more integrated care seems promising and realistic. Areas to explore could be more frequent but remote video monitoring of our high risk patients, extending the range of directly observed therapy, and support for patients involved in modification of high-risk behaviors, such as smoking cessation, ETOH reduction, and substance misuse.

ASSESSING THE IMPACT OF A RESIDENT COORDINATED TRANSITIONAL CARE ROTATION ON RESIDENT ATTITUDES TOWARDS PATIENT TRANSITIONS AND HEALTH MANAGEMENTAlexis Eastman^1,3; Elizabeth Chapman^1,2; Laury Jensen³; Holly I. Bottoms⁴; Mary Thompson²; Bennett Vogelman^1,2; Amy J. Kind^2,3. ¹University of WI Hospitals and Clinics, Madison, WI; ²University of Wisconsin, Madison, WI; ³Madison VA Geriatric Research Education and Clinical Center, Madison, WI; ⁴William S. Middleton Memorial Veterans Hospital, Madison, WI. (Tracking ID #1938974)

BACKGROUND: Although the Accreditation Council for Graduate Medical Education (ACGME) and others have called for improved medical resident education in the area of care transitions, few well-described formal education programs are available in the published literature to accomplish this goal. Pragmatically, instituting a formal transitions training program is challenging due to supervision and site-specific limitations and work hour requirements. Our objective was to evaluate whether Resident Coordinated Transitional Care (RC-TraC), an adaptation of the previously tested phone-based VA Coordinated Transitional Care (C-TraC) program, impacted resident attitudes related to transitional care.

METHODS: In 2009, the University of Wisconsin Internal Medicine residency program established RC-TraC, a phone-based care transitions training experience for residents randomly assigned to VA primary care continuity clinics. Residents with other primary care continuity clinics did not have a formal care transitions training experience. During this rotation, residents act as discharge care managers for hospitalized patients from the resident clinic panel, similar to the RN case manager position in the C-TraC program. Prior to discharge, the RC-TraC residents counsel patients on symptoms that should warrant medical attention. Post-discharge, the RC-TraC residents complete protocolized phone calls weekly for 30days wherein symptoms are reviewed, medications reconciled and appropriate follow-up appointments are verified. In spring of 2013, all UW internal medicine residents were electronically surveyed to determine their attitudes regarding care transitions, including residents who did (N = 31) and did not experience RC-TraC (N = 56). Responses were collected on a 4-item modified Likert scale (almost never, infrequently, often, almost always). For purposes of analysis, “almost never/infrequently” and “often/almost always” were combined. Frequencies were compared using Chi-squared tests.

RESULTS: Overall survey response rate was 83/87 (95%), with 29/31 (94%) RC-TraC and 54/56 (96%) non-RC-TraC residents responding. Ninety-seven percent of residents who had experienced RC-TraC at least once identified that they had received in-residency transitions training, while only 24% of non-RC-TraC residents felt they had received transitions training (p < 0.001). When compared to non-RC-TraC residents, RC-TraC residents were more likely to believe patients could “almost always/often” identify their pre-admission meds (45% RC-TraC versus 18% non-RC-TraC, p = 0.01), identify their pre-admission diagnoses (55% vs 35%, p = 0.08) and know how to contact providers post-discharge (52% vs 31%, p = 0.07). Yet, there was no statistical difference between RC-TraC and non-RC-TraC residents’ beliefs on patients’ abilities to identify post-discharge medications and hospital diagnoses.

CONCLUSIONS: Findings suggest that the RC-TraC program is identified by residents as a formal care transitions training experience and that it alters resident perceptions of patients’ baseline capacity for health management. This may suggest that RC-TraC increased residents’ appreciation for a patient’s capacity to participate in their health care. RC-TraC is a feasible care transitions training program, but needs additional testing of its impact on resident knowledge and patient outcomes.

ASSESSING THE UNDERSTANDABILITY AND ACTIONABILITY OF AFTER VISIT SUMMARIES FROM PRIMARY CARE CLINICSAlex Federman¹; Christopher Salmon¹; Ravishankar Ramaswamy¹; Sereena Singh¹; Allison L. Russell²; Michael S. Wolf². ¹Icahn School of Medicine at Mount Sinai, New York, NY; ²Feinberg School of Medicine, Northwestern University, Chicago, IL. (Tracking ID #1936547)

BACKGROUND: The after visit summary (AVS) is a document given to patients to summarize key elements of the clinical encounter. The AVS has the potential to serve as an organizing and informational tool for patients, and has gained importance as new federal and state health policies, like Meaningful Use, work to advance patient centered care. Despite this, little is known about the AVSs currently in use in ambulatory practices. In this study, we sought to characterize the content and formatting of AVSs from a primary care practices and determine understandability and actionability of the information they provide.

METHODS: We identified National Committee for Quality Assurance (NCQA) accredited level III patient centered medical homes from the NCQA public use database and through snow-ball contacts. A convenience sample of these practices was contacted by email and telephone. Clinic directors or clinician representatives underwent a semi-structured interview about their experience with their practice’s AVS and were asked to email or fax de-identified samples of actual AVSs from their practice. AVS samples were analyzed for content, length, and reading grade level. They were then assessed for understandability and usability using the Patient Education Materials Assessment Tool (PEMAT). Two raters independently evaluated the AVS samples.

RESULTS: We evaluated AVS documents from 18 primary care practices in 14 states. Number of pages ranged from 1 to 9 (median, 2) and the mean and median font size was 10. Content included medication lists (100%), specific instructions and referrals requested or procedures performed (83%), diagnoses (72%), vital signs (67%), allergies (56%), appointments (50%), goals of care (22%), and immunization history (17%). On the PEMAT, the average levels of understandability and actionability were 62% and 80%, respectively, indicating poor understandability and fair actionability. The median reading grade level was 12 (range 8–19) by the Flesch-Kincaid formula. Interviewed clinicians generally believed that their AVSs were of little use to patients except for the medication lists. None reported providing AVSs in languages other than English.

CONCLUSIONS: After visit summaries may not be achieving their intended purpose of providing patients with accessible and useful information regarding their care, nor meeting Meaningful Use requirements for content, even in NCQA-certified level III patient centered medical homes. Improvements in formatting, wording, and content of AVSs are needed to advance patient-centered care.

ASSESSMENT OF THE “SPEAKING UP” CULTURE: A NEW DOMAIN FOR PATIENT SAFETY CULTURE SURVEYSWilliam Martinez¹; Jason M. Etchegaray²; Lisa Lehmann^3,5; Eric J. Thomas²; Sigall K. Bell^4,5. ¹Vanderbilt University Medical Center, Nashville, TN; ²Univesity of Texas Medical School at Houston, Houston, TX; ³Brigham and Women’s Hospital, Boston, MA; ⁴Beth Israel Deaconess Medical Center, Boston, MA; ⁵Harvard Medical School, Boston, MA. (Tracking ID #1939052)

BACKGROUND: Despite over a decade focusing on identifying the key components of patient safety culture relevant to healthcare settings, tangible safety progress has been slow. Current safety culture assessments fail to measure an important emerging competency, “speaking up” about patient safety threats, a skill that may be particularly challenging for residents responding to conforming pressures of the medical hierarchy. Thus, we sought to (1) develop and test survey items that measure the speaking up culture in the clinical work environment, (2) examine relationships among speaking up culture, teamwork culture, error disclosure culture and safety culture and (3) establish predictive validity for survey items measuring speaking up culture.

METHODS: All (667) surgical and medical residents from two large academic medical centers were invited to complete an anonymous, electronic questionnaire. The questionnaire asked respondents about personal experience speaking up about threats to patient safety, and contained six items developed by patient safety experts to assess the speaking up culture in the clinical environment. It also included previously validated measures of safety culture including the safety and teamwork dimensions of the Safety Attitudes Questionnaire (SAQ) and the general dimension of the Error Disclosure Scale. Factor analysis was used to test whether the six speaking up items were consistent with the concept of speaking up about threats to patient safety and to create a summary score in order to examine the relationship of speaking up culture to other patient safety domains and compare speaking up culture between groups.

RESULTS: The overall response rate was 52% (349/667). Factor analysis supported a single 6-item speaking up scale and a scale score was created by summing the 5-point Likert-scale responses to individual items (score range: 5–30, Cronbach α = 0.75). Speaking up culture was a measure unique from safety culture, teamwork culture, and error disclosure culture scales (correlations less than r = 0.85). Speaking up culture predicted residents’ self-reported, personal experience speaking up about patient safety threats during their last inpatient month (OR 1.13, 95%CI 1.05–1.22), while teamwork (OR 1.02, 95%CI 0.91–1.15), error disclosure (OR 0.98, 95%CI 0.92–1.05) and safety culture (OR 0.96, 95%CI 0.92–1.01) did not. Surgical residents perceived their clinical work environment’s culture to be less supportive of speaking up than medical residents (mean scale score: 21.22 vs. 22.19, P = 0.045). Compared to medical residents, fewer surgical residents felt (A) encouraged by colleagues to speak up about patient safety concerns (76% vs. 62%; P = 0.006) and (B) that speaking up about patient safety concerns results in meaningful change (67% vs. 55%; P = 0.03).

CONCLUSIONS: Our speaking up scale has sound psychometric properties and is a new measure of speaking up culture regarding patient safety threats in the clinical work environment that predicts self-reported speaking up about safety threats among residents. This measure fills an existing gap in assessments of organizational safety by measuring the extent to which the culture within a clinical work environment supports clear, direct, real-time communication between colleagues when a threat to patient safety is identified, a critical aspect of safety culture that is not currently assessed by traditional safety culture instruments. The culture may be less supportive of residents speaking up in surgical environments than in medical environments.

ASSOCIATION BETWEEN AFTER-VISIT CLINICAL SUMMARIES AND CHRONIC DISEASE MANAGEMENTRebecca G. Mishuris^1,2; Michael Healey^1,2; Endel J. Orav^1,2; Jennifer Haas^1,2. ¹Brigham and Women’s Hospital, Boston, MA; ²Harvard Medical School, Boston, MA. (Tracking ID #1937817)

BACKGROUND: Electronic health records (EHRs) have been adopted by 72% of US physicians. The Medicare and Medicaid EHR Incentive Program, or “meaningful use”, incents physicians to adopt EHRs and perform certain tasks with them, including the provision of after-visit clinical summaries following every office visit. These summaries must at least contain the patient’s problem list, medication list, allergies, and recent lab results. It is unknown how these visit summaries, and the written information they provide to the patient, will impact chronic disease management.

METHODS: We used patient data from 14 adult primary care clinics associated with a Boston teaching hospital from October 2010 to October 2013 to determine whether providing after-visit clinical summaries is associated with an improvement in diabetes management, as measured by hemoglobin A1c (HbA1c). After-visit clinical summaries were completely implemented across all clinics by October 2012, which was used as the cut-off time for pre- and post-implementation analysis. After-visit clinical summaries are provided to patients after every clinic visit, either in person, by mail or through an electronic patient portal. Patients were included in the analysis if they were 18 to 74years of age, seen at one of the clinics at least once and were prescribed an anti-hyperglycemic agent for diabetes management during the period from October 2010 to October 2013. We will perform an interrupted time series analysis with repeated measures to look for a difference in change of HbA1c from the time prior to implementation of the after-visit clinical summaries to the time after the implementation of the after-visit clinical summaries. We will use clustering to account for the effect of the clinic or provider and control for patient characteristics, how the after-visit clinical summary was provided and the number of instances that the after-visit clinical summary was provided.

RESULTS: There were 6935 patients in this cohort; 6281 patients were seen prior to the implementation of the after-visit clinical summary and 5414 patients were seen after the implementation of the after-visit clinical summary. As of October 2013, 94.8% of physicians were meeting the meaningful use requirement of providing after-visit clinical summaries for greater than 50% of clinic visits, and patients were provided an after-visit clinical summary for 84.8% of all clinic visits. Average HbA1c prior to implementation was 7.7, ranging from 3.8 to 18.9. Average change in HbA1c during this period was −0.03, ranging from −12.2 to 13.3. Average percentage change in HbA1c during this period was 1.6%, ranging from −64.5 to 97.0%. Average HbA1c after implementation was 7.7, ranging from 4.1 to 18.9. Average change in HbA1c during this period was 0.06, ranging from −13.1 to 8.8. Average percentage change in HbA1c during this period was 3.8%, ranging from −69.3 to 85.0%. The study will have a power of 80% to detect a 0.05 change in HbA1c with 95% confidence and a power of 95% to detect a 0.1 change in HbA1c with 95% confidence.

CONCLUSIONS: After-visit clinical summaries have been successfully implemented in our adult primary care clinics. Average hemoglobin A1c control has not changed from the period preceding implementation of these patient after-visit clinical summaries to the period following implementation. Further analysis is required to determine whether there has not yet been a long enough follow-up period to see a significant change in chronic disease control associated with after-visit clinical summaries or whether there needs to be further emphasis placed on the importance of the information provided in the after-visit clinical summary to effect a change in chronic disease control.

ASSOCIATION BETWEEN BMI AND QUALITY OF CARE AMONG INDIVIDUALS WITH HIV, 2007–2010Erin Fennern¹; Somnath Saha¹; Stephen Berry²; Rochelle Fu¹; Baligh Yehia³; Kelly Gebo²; Allison Agwu²; Philip T. Korthuis¹. ¹Oregon Health and Science University, Portland, OR; ²Johns Hopkins University School of Medicine, Baltimore, MD; ³University of Pennsylvania, Philadelphia, PA. (Tracking ID #1940304)

BACKGROUND: Prior studies suggest health care providers harbor biased attitudes toward obese patients, which could affect the quality of care received. HIV providers are increasingly encouraged to adhere to quality indicator (QI) performance measures. We sought to assess differences in the care provided to HIV patients by BMI category.

METHODS: We abstracted nine HIV QIs (Table) from medical records of 6,031 patients (10,896 total person-years) age >18years engaged in care (≥2 clinic visits per calendar year) at seven clinics in the HIV Research Network between 2007 and 2010. Underweight patients were excluded. Patient BMI was categorized as normal (18.5 ≤ 24.9), overweight (25 to 29.9), stage 1 obese (30 to 34.9), stage 2 obese (35 to 39.9), and stage 3 obese (≥ 40). Using generalized estimating equations, logistic regression examined the association between BMI and receipt of each QI, and linear regression examined associations between BMI and the mean percent of QIs received, adjusting for age, race, sex, HIV risk behavior, and number of primary care visits in a year. We used R-squared to assess linear trends in odds ratios (ORs) and beta coefficients across BMI category.

RESULTS: Overall, patients received 65% of eligible QIs in a given year. Those with stage 3 obesity received slightly lower eligible QIs vs. normal weight patients (64.3% vs. 65.7%, adjusted β = −0.04, 95% CI −0.06, −0.02), with a moderate downward linear trend as BMI increased (R2 = 0.691). With regard to specific QIs, patients with stage 3 obesity received less screening for gonorrhea/Chlamydia (GC) (aOR = 0.76, 95% CI 0.60 to 0.96) and syphilis (aOR = 0.69, 95% CI 0.56 to 0.86) relative to normal weight patients. As BMI category increased, linear trends were observed for decreasing syphilis, GC, and HCV screening, and increasing CD4 count monitoring and lipid screening (Table). Receipt of ART, PCP and MAC prophylaxis were comparable across BMI categories.

CONCLUSIONS: Overall, HIV quality of care differed only slightly across BMI categories. Providers prescribed ART regardless of BMI, but performed less sexually transmitted infection screening and more lipid screening as BMI increased, suggesting providers may consciously or unconsciously prioritize some QIs over others in patients with higher BMI. Additional studies are needed to determine the source of these differences.

Adjusted Associations Between BMI Category and Receipt of each QI and Mean Percent QI Received, if Eligible.Adjusted Associations Between BMI Category and Receipt of each QI and Mean Percent QI Received, if Eligible.

ASSOCIATION BETWEEN EMERGENCY DEPARTMENT USE AND THE DISCRIMINATION PERCEIVED BY FREQUENT USERSStéphanie Baggio¹; Patrick Bodenmann²; Jean-Bernard Daeppen²; Bernard Burnand²; Olivier Hugli²; Ornella Ruggeri²; Jean-Blaise Wasserfallen²; Karine Moschetti²; Philippe Staeger²; Séverine Alary²; Katia Iglesias²; Marina Canepa Allen². ¹Lausanne University, Lausanne, Switzerland; ²Lausanne University Hospital, Lausanne, Switzerland. (Tracking ID #1930263)

BACKGROUND: Frequent Emergency Department (ED) users are often vulnerable patients with many risk factors. Furthermore, they are often perceived as unwelcome patients, causing EDs’ overcrowding, and being time-consuming, “illegitimate” users. However, data are scarce regarding how frequent ED users deal with this perception and particularly regarding their related feelings of discrimination. This study investigated the relationship between frequent ED use and perceived discrimination during such use in Switzerland.

METHODS: Data were part of a baseline assessment of a randomized controlled trial designed to improve frequent users’ quality of care in Lausanne University Hospital, a French-speaking Swiss tertiary care hospital, with over 50,000 annual ED visits. In total, 250 frequent ED users (≥5 attendances during the previous 12months, ≥ 18years), were interviewed between May 2012 and July 2013. From a previously published questionnaire, 15 dichotomous sources of discrimination were assessed, and grouped into five dichotomous causes: (i) any cause, (ii) immigration-related, (iii) age/gender, (iv) physical characteristics/sexual orientation, (v) social/economic status. Prevalence rates of perceived discrimination were computed first, then associations between the number of ED attendances and perceived discrimination were assessed with multivariate models (logistic regressions) controlling for socio-demographic characteristics and health-related issues (i.e. age, gender, education, nationality, proficiency in French, health literacy, and somatic-, mental-health-, behavioral-, and social issues).

RESULTS: Participants were 46.2 ± 18.9years old on average, 57.2% were men, 47.8% Swiss, 17.7% European, and 34.5% non-European. In total, 11.6% of the frequent ED users were asylum seekers or undocumented immigrants. Prevalence rates of perceived discrimination showed that 35.2% reported at least one source of discrimination. Discrimination related to physical characteristics/sexual orientation (20.8%), social/economic position (16.0%), and immigration (16.0%) was more frequent than discrimination related to age and gender (6.8%). Multivariate models showed that when the number of ED visits increased, frequent users were more likely to feel discriminated against (any cause: odd-ratio [OR] = 1.19, p = 0.023), particularly regarding social/economic status (OR = 1.28, p = 0.011) and physical characteristics/sexual orientation (OR = 1.19, p = 0.023). Contrariwise, discrimination related to immigration and age/gender was not significantly related to the number of ED attendances (p = .803 and p = 0.266).

CONCLUSIONS: Frequent ED users appeared to perceive frequent discrimination (35.2% compared to 11.1% of non-frequent ED users hospitalized in another French-speaking Swiss tertiary hospital), associated with an increased number of ED attendances. Moreover, they did not feel discriminated against due to intrinsic characteristics or race-based sources (e.g. age, gender, nationality, or skin color), but rather for social and lifestyle factors (e.g. income, education, physical appearance, or sexual orientation), even if foreigners, asylum seekers, and undocumented immigrants were highly represented in the sample. The conclusions of this study should help health care practitioners and managers to be more cautious about potential discrimination against frequent ED users, as they may be unwittingly perpetuating health care disparities.

ASSOCIATION BETWEEN OFFICE-BASED MEDICAID ACCEPTANCE RATES AND SAFETY-NET SERVICE UTILIZATION BY MEDICAID PATIENTSKimberly Cartmill; Arnab K. Ghosh; Meroe B. Morse; Shawn L. Tyler. New York University, New York, NY. (Tracking ID #1929579)

BACKGROUND: The Patient Protection and Affordable Care Act (ACA) expands Medicaid coverage, which may increase utilization of healthcare safety net facilities such as emergency departments (EDs) and hospital outpatient departments (OPDs). To ease their burden, the ACA raises Medicaid reimbursement rates to encourage primary care physicians to accept Medicaid patients. It is uncertain if an increase in outpatient Medicaid acceptance rates is associated with a decrease in hospital OPD and ED visits among Medicaid patients. This study uses a robust national survey to determine the association between office-based Medicaid acceptance rates and the percent of Medicaid OPD and ED visits.

METHODS: Data from two nationally representative surveys, NAMCS and NHAMCS, were used to measure Medicaid acceptance rates among office-based physicians and percent of Medicaid ED and OPD visits from 2003 to 2010. 231,000 physicians responded to NAMCS. 269,000 OPD visits and 257,000 ED visits were used in NHAMCS. Trends in acceptance rates and percent Medicaid OPD and ED visits were analyzed by year (2003–2010) and across four broad census regions covering the country. Thirty-two point estimates, each representing one region and 1year, were used to test the relationship between office-based acceptance rates (NAMCS) and hospital-based visits (NHAMCS). Pearson’s correlation coefficient was applied to assess magnitude, direction and significance of the association.

RESULTS: Response rates for the specified questions studied were 97% for acceptance rates; 94% for expected method of payment for OPD or ED visits. Nationally, no statistically significant trend in percent of office-based physicians accepting Medicaid was found (adjusted F 2.16, p = 0.62) between 2004 and 2010 (2003 excluded as outlier). These rates did not differ by year within three of the four regions. Percent Medicaid, OPD, and ED visits did not change significantly over the 7-year period nationally or within regions. When Medicaid acceptance rates were compared with percent Medicaid ED visits using point estimates, there was no significant correlation (R² = 0.05; p = 0.87). However, there is a strong negative correlation between percent office-based Medicaid acceptance and percent Medicaid OPD visits (R² = −0.59, p < 0.01).

CONCLUSIONS: In nationally representative data sets, office-based physician acceptance rates of Medicaid patients were inversely associated with visits to hospital outpatient departments, but not hospital emergency departments. Hospital outpatient departments may serve as a buffer to prevent emergency department utilization. This requires further investigation with better national surveys to assess the impact of health policy interventions.

ASSOCIATION BETWEEN OUTPATIENT “NO-SHOWS” AND SUBSEQUENT ACUTE CARE UTILIZATIONAndrew S. Hwang¹; Steven J. Atlas²; Patrick R. Cronin³; Jeffrey M. Ashburner²; Sachin J. Shah²; Wei He²; Clemens S. Hong². ¹Tufts University School of Medicine, Boston, MA; ²Massachusetts General Hospital, Boston, MA; ³Massachusetts General Hospital, Boston, MA. (Tracking ID #1934637)

BACKGROUND: To reduce healthcare costs, healthcare systems are focusing on identifying the relatively small percentage of patients who account for a disproportionate share of hospitalizations and emergency department (ED) visits. We hypothesized that primary care patients with a higher propensity for missed outpatient appointments, or “no-shows,” will have higher acute care utilization over the following 3years.

METHODS: We calculated the “no-show propensity factor” (NSPF) for 140,947 patients seen in an academic primary care practice network from 2007 to 2009 using 5years (2005–2009) of outpatient appointment data. NSPF transforms a patient’s count of arrivals and no-shows into a score that represents appointment adherence, correcting for patients with few appointments to avoid undue influence on the model. We divided patients into two groups: patients in the highest quartile and patients in the lower 3 quartiles of NSPF, and assessed subsequent hospital and ED utilization over 3years (2010–2012). We evaluated the association between NSPF and our two continuous outcomes using Poisson regression models adjusting for number of outpatient visits in 5years (2005–2009). Using a “change-in-estimate” strategy, we included other patient characteristics (age, gender, race, language, insurance, neighborhood median household income, Charlson comorbidity index, diagnosis of depression and alcoholism, and hospitalization and ED visits in 2009) and primary care provider characteristics (gender, years in practice, and whether the provider practiced at a community health center) in the models if they altered the incidence rate ratio (RR) by >5%. We also ran models adjusting for all patient and provider characteristics and performed a subgroup analysis to examine the association between NSPF and subsequent hospitalization in patients with high-risk chronic illnesses (congestive heart failure, chronic obstructive pulmonary disease, diabetes mellitus, atrial fibrillation, coronary artery and cerebrovascular disease).

RESULTS: Compared to patients in the lower 3 quartiles of NSPF, patients in the highest quartile were significantly (P < 0.001 for all) younger (45 vs. 52years) and more likely to be male (44% vs. 42), non-white (37% vs. 16%), non-English speaking (17% vs. 6%), and insured by Medicaid or uninsured (22% vs. 9%), live in neighborhoods with low median household income ($50,612 vs. $63,781) and have higher rates of ED visits (192 vs. 84 visits/1,000 patients) and admissions (94 vs. 68 hospitalizations/1,000 patients) in 2009. In the subsequent 3years (2010–2012), patients in the highest quartile of NSPF had significantly (p < .0001 for all) higher unadjusted rates of acute care utilization (77 vs. 66 hospitalizations/1,000 patients/year and 153 vs. 73 ED visits/1,000 patients/year) than patients in the lower 3 quartiles. In addition, Poisson regression analysis demonstrated that compared to patients in the lower 3 quartiles of NSPF, eligible patients in the highest quartile had a 19% increased rate of ED visits in 3years (adjusted RR 1.19 [1.05–1.34] adjusting for number of outpatient visits, age and ED visits and hospitalizations in 2009), but not a significantly increased rate of hospitalization in 3years (adjusted RR 1.08 [0.996–1.18] adjusting for number of outpatient visits, age, and ED visits in 2009). Including all variables in the regression models did not change the significance or direction of the findings, and subgroup analysis among patients with high-risk comorbidities (congestive heart failure, chronic obstructive pulmonary disease, diabetes mellitus, atrial fibrillation, coronary artery and cerebrovascular disease) did not reveal a significant association between NSPF and hospitalization.

CONCLUSIONS: NSPF was an independent predictor of subsequent ED utilization. Using NSPF may help primary care delivery systems identify patients at increased risk for frequent utilization of the ED and tailor programs to increase primary care engagement and reduce ED overutilization.

ASSOCIATION BETWEEN OUTPATIENT “NO-SHOWS” AND SUBSEQUENT CLINICAL OUTCOMESAndrew S. Hwang¹; Steven J. Atlas²; Patrick R. Cronin³; Jeffrey M. Ashburner²; Sachin J. Shah²; Wei He²; Clemens S. Hong². ¹Tufts University School of Medicine, Boston, MA; ²Massachusetts General Hospital, Boston, MA; ³Massachusetts General Hospital, Boston, MA. (Tracking ID #1934608)

BACKGROUND: To improve care and achieve performance targets, healthcare systems seek to identify patients at high risk for not achieving preventive cancer screening and chronic disease management goals. We hypothesized that patients with a higher propensity for missed outpatient appointments, or “no-shows,” would have lower colorectal cancer (CRC) screening and low-density lipoprotein (LDL) goal attainment over the following year.

METHODS: We calculated the “no-show propensity factor” (NSPF) for 140,947 patients seen in an academic primary care practice network from 2007 to 2009 using 5years (2005–2009) of outpatient appointment data. NSPF transforms a patient’s count of arrivals and no-shows into a score that represents appointment adherence, correcting for patients with few appointments to avoid undue influence on the model. We divided patients into two groups: patients in the highest quartile and patients in the lower 3 quartiles of NSPF, and assessed CRC screening completion among patients (52–75years) without prior colectomy and LDL goal (LDL < 100mg/dl) attainment in patients with diabetes or coronary artery disease at 1year follow-up (2010). We evaluated the association between NSPF and our binary outcomes using logistic regression adjusting for number of visits. Using a “change-in-estimate” strategy, we included patient characteristics (age, gender, race, language, insurance, neighborhood median household income, Charlson comorbidity index, and diagnosis of depression or alcohol use disorder) and primary care provider characteristics (gender, years in practice, and whether the provider practiced at a community health center) in the models if they altered the odds ratio by >5%. We also ran models adjusting for all variables.

RESULTS: Compared to patients in the lower 3 quartiles for NSPF, patients in the highest quartile were significantly (P < 0.0001 for all) younger (45 vs. 52years) and more likely to be male (44% vs. 42), non-white (37% vs. 16%), non-English speaking (17% vs. 6%), lack insurance (7% vs. 4%), live in neighborhoods with low median household income ($50,612 vs. $63,781), receive primary care at a community health center (15% vs. 7%), have a diagnosis of depression (20% vs. 11%) or alcohol use disorder (3% vs. 1%), and have higher number of outpatient visits over 5years (27 vs. 21). Compared to eligible patients in the lower quartiles for NSPF, eligible patients in the highest quartile were significantly more likely to have not completed CRC screening (adjusted OR 1.87 [1.77–1.97] adjusting for number of visits and neighborhood median household income). Compared to patient in the lower quartile for NSPF, eligible patients in highest quartile were significantly more likely to have an above goal LDL (adjusted OR 1.70 [1.57–1.83] adjusting for number of visits and age). Inclusion of all variables in the models did not change the significance or direction of the effect, and the effect changed by less than 6% for both outcomes.

CONCLUSIONS: NSPF was a strong, independent predictor of subsequent colonoscopy completion and LDL goal attainment. Using NSPF may help healthcare systems identify patients at increased risk for non-adherence to recommended care for targeted interventions to improve care and achieve performance targets.

ASSOCIATION BETWEEN USUAL SOURCE OF CARE AND CHRONIC DISEASE HOSPITALIZATION IN CALIFORNIA MEDICAID PATIENTSHugo A. Torres¹; Ed Yelin². ¹Cambridge Health Alliance, Cambridge, MA; ²University of California, San Francisco, San Francisco, CA. (Tracking ID #1938741)

BACKGROUND: The costs of Medicaid are driven in large part by the prevalence and costs of chronic diseases. Therefore, achieving improved control of chronic diseases by reducing the rates of hospital use associated with them is vital to the future successful expansion of the Medicaid program. Prior research links primary care access, often measured by usual source of care (USOC), with lower rates of hospitalizations for many chronic diseases, called ambulatory care sensitive Conditions (ACSC). Medicaid patients in general have high rates of ER utilization and hospitalization for ACSC relative to the uninsured and privately insured. This study aims to determine if the difference in ACS hospitalizations between Medicaid beneficiaries and low-income privately-insured individuals can be explained by the presence of a USOC.

METHODS: This is a cross-sectional study utilizing the 2009 and 2011 California Health Interview Survey (CHIS). The population studied is respondents to the CHIS, aged 18–64, earning less than 300% of the Federal Poverty Level (N = 7564). The principal independent variable is Medicaid vs. Private insurance. The outcome is an ED visit or hospitalization for an ACSC. We use logistic regression to control for demographic variables that have been previously shown to affect health access. The analysis determines if Medicaid insurance was associated with rates of ED use or hospitalization; USOC was added to the model, first as an independent variable, then as an interaction term, to determine if the magnitude of the effect of insurance status on chronic disease ED use or hospitalization was altered by primary care access.

RESULTS: Medicaid patients were likelier to be female (60% vs. 53%), younger (mean age 43.6 vs. 45.1), in poorer self-rated health, unmarried, and more likely to have children than their counterparts with private insurance (p < 0.05 for all comparisons). They were also more likely to report a disability and to lack a usual source of care (OR = 0.72, 95% CI 0.54–0.95, p = 0.02). The unadjusted rates of ED use and hospitalization for ACS conditions were 46.8% vs. 30.9% for privately insured patients (OR: 1.96, 95% CI 1.60–2.41, p < 0.0001). The OR adjusted for the above was lower but still significant at 1.54 (95% CI 1.2–2.0, p < 0.0001). The association remained after controlling for USOC (OR 1.55, p < 0.0001). An interaction term for USOC and Medicaid was positive and significant (OR 1.79, p = 0.05), indicating an increased effect of Medicaid on ED use/hospitalization for ACS conditions.

CONCLUSIONS: This analysis demonstrates a correlation between Medicaid (vs. private) insurance and ED use/hospitalization among people with ACSC, and that possession of primary care access in the form of having a usual source of care did not affect this relationship in the expected pattern; in fact, having a usual source of care increased the association. This is consistent with data from Oregon that found increased primary care use and ED use among Medicaid patients. The explanation for this relationship is likely complex. Medicaid patients may face hardship to access their provider, and are also likelier to have mental health and social issues which complicate their care. This research highlights the need for expanded efforts to understand and treat the causes of poor chronic disease outcomes in Medicaid patients. To do so would be to improve their health status outcomes, decrease ED visits and hospitalizations, and improve quality of life.

ASSOCIATION OF ADHERENCE MEASURES WITH PHYSICAL ACTIVITY OUTCOMES IN AN ONLINE WEIGHT LOSS TRIAL: RESULTS FROM THE OCELOT STUDYBethany Scanlan; Molly B. Conroy; Dana L. Tudorascu; Irina Karpov; Rachel Hess; Gary Fischer; Laurey R. Simkin-Silverman; Kathleen M. McTigue. University of Pittsburgh, Pittsburgh, PA. (Tracking ID #1931481)

BACKGROUND: Lifestyle programs for weight loss and physical activity (PA) promotion have traditionally been delivered in person. Adherence, defined by number of sessions attended, has been shown to correlate with weight loss and PA improvements. With growing use of health information technology, an understanding of adherence to online interventions and the relationship between adherence and PA is needed. We hypothesize that participants adherent to an online intervention will have higher PA levels at 6 and 12months than those who are non-adherent.

METHODS: The Online Counseling to Enable Lifestyle-focused Obesity Counseling in Primary Care (OCELOT-PC) study is a comparative effectiveness study of three online lifestyle interventions. We recruited 373 primary care patients, 257 of whom were randomized to a 12-month structured online adaptation of the Diabetes Prevention Program’s lifestyle intervention. The remaining patients had access to online resources alone and served as an active control arm. The behavioral online intervention included: lessons; self-monitoring (SM) of weight, eating pattern, and PA; feedback from a lifestyle coach; and links to community resources. We defined a current program user as a participant with a log-in in past 28days. Non-users are those with no log-in for ≥28days and no response to either of two weekly phone calls from the coach (initiated after the participant has not logged in for ≥14days). Adherence was also based on other aspects of program use including: log-in in past week, weight entered in past week, eating pattern and PA SM (including number of days of SM in past week), and number of lessons completed. PA was determined by steps recorded on a pedometer worn for 2weeks at baseline, 6months, and 12month. Adherence measures were correlated with PA outcome using Pearson correlations for continuous variables; means were compared using t-tests.

RESULTS: We were able to calculate program use and other adherence measures for 255 participants who were given access to the structured online intervention. Eighty percent of participants were women, 77% white, and 62% had education ≥ college; participants’ average age was 49.2years. Participants had an average (SD) of 4127 (1918) steps per day and weight of 107.1 (21.6) kg at baseline. At 6months, patients who were current program users had a mean (SD) of 3955 (2467) steps per day; those who were non-users had a mean of 1888 (1930) steps (p < 0.0001). Those with a log-in in the past week, weight entered in the past week, or any diet or PA SM in the past week also showed statistically significant greater numbers of steps per day compared to those who were non-adherent to these measures (all p < 0.001). Number of lessons completed, mean number of days of dietary SM in past week, and mean number of days of PA SM in past week were positively correlated with PA (r = 0.4; r = 0.37; r = 0.39; all p < 0.0001). Similar trends were observed for PA at 12months. At 12months, current program users had a mean (SD) of 4384 (2755) steps; non-users had a mean of 2689 (2348) steps (p < 0.001).

CONCLUSIONS: Participant adherence to the online intervention was associated with PA outcome. Those who were adherent to the tasks of the behavioral online intervention had significantly more steps at 6 and 12months compared to those who were non-adherent. Further research should focus on which measure of adherence best relates to PA and other behavioral outcomes, as well as mechanisms to increase adherence to online interventions.

ASSOCIATION OF CARDIOVASCULAR HEALTH SCREENING WITH MORTALITY, CLINICAL OUTCOMES, AND HEALTH CARE COSTDong Wook Shin¹; Hyejin Lee¹; BeLong Cho¹; Juhee Cho²; Ju Young Kim¹; Eliseo Guallar². ¹Seoul National University Hospital, Seoul, Republic of Korea; ²Bloomberg school of public health, Baltimore, MD. (Tracking ID #1939911)

BACKGROUND: Since several CVD risk factors are modifiable, prevention programs routinely screen for hypertension, diabetes, and dyslipidemia expecting that early detection and treatment of CVD-related health conditions will decrease the burden of CVD. The effectiveness of screening programs for the CVD-related health conditions on health outcomes and healthcare utilization, however, is unclear. In South Korea, a country with universal healthcare coverage, the Korean National Health Insurance (KNHI) Corporation provides a biennial CVD health screening program to all national health insurance members over 40years of age free of charge. We aimed to determine whether a nationwide CVD health screening program is associated with CVD incidence, mortality, healthcare utilization and costs.

METHODS: We randomly selected a 3% sample (n = 621 350) of all KNHI members 40years of age or older as of December 31, 2002. We excluded participants with cancer, diabetes, hypertension, dyslipidemia, or any related CVD including stroke and myocardial infarction (n = 170 490). Study participants were followed-up from January 1, 2005 through December 31, 2010. Propensity score matching was used to improve comparability of screened and non-screened participants. (n = 149 243 matched pairs)

RESULTS: The hazard ratios for CVD mortality, all-cause mortality, and incident myocardial infarction, cerebral infarction, and cerebral hemorrhage during comparing participants who attended a screening exam during 2003–2004 compared to those who did not were 0.55 (95% CI: 0.50–0.60), 0.63 (95% CI: 0.61–0.66), 0.85 (95% CI: 0.76–0.95), 0.82 (95% CI: 0.77–0.87), and 0.71 (95% CI: 0.65–0.79). Screening attenders had higher rates of newly diagnosed hypertension and dyslipidemia (p < 0.001), lower inpatient days of stay and cost and lower outpatient cost compared to non-attenders.

CONCLUSIONS: CVD health screening was associated with lower rates of CVD and all-cause mortality and CVD events, higher detection of CVD-related health conditions, and lower healthcare utilization and costs. In the absence of large randomized controlled trials, our findings are suggestive of the effectiveness of CVD screening programs.

ASSOCIATION OF ILLNESS BELIEFS AND END-OF-LIFE TREATMENT PREFERENCES AMONG LUNG CANCER PATIENTSJenny J. Lin; Jessica Lake; Juan Wisnivesky. Icahn School of Medicine at Mount Sinai, New York, NY. (Tracking ID #1939176)

BACKGROUND: Patient preferences about end of life care regarding resuscitation and intubation may be influenced by illness beliefs, family/social factors or physician communication. We assessed the association between patient-physician communication and cultural beliefs with patients’ preferences for end-of-life care among recently-diagnosed lung cancer patients.

METHODS: Patients with recently-diagnosed lung cancer were recruited from four medical centers in New York City from 2008 to 2011. Participants were surveyed about beliefs regarding lung cancer and cancer treatment, fatalism, medical mistrust and physician communication regarding treatment, symptoms and needs. Univariate analysis was used to assess which factors were associated with end-of-life preferences.

RESULTS: Of the 352 lung cancer patients enrolled in the study, 168 (48%) were men; 21% were black, 20% were Hispanic and 59% were white. Overall, 132 (38%) did not want to be intubated and 88 (25%) did not want to be resuscitated. Approximately 30% of the participants did not have a preference regarding either intubation or resuscitation. Overall, 93% of patients who did not want resuscitation also did not want intubation, while 25% patients who wanted to be resuscitated did not want to be intubated. Those who were younger or who had lower educational attainment were more likely to want more aggressive care (resuscitation and intubation, p < 0.05 for all comparisons). There was no difference in end-of-life care preferences by lung cancer stage (p > 0.05 for all comparisons). Patients who believed that CPR should always be performed and those who expected their cancer to be cured were more likely to want aggressive care (both resuscitation and intubation, p < 0.04 for all comparisons). Additionally, those who agreed with the statement that “everything that happens is part of God’s plan” (p < 0.01) or who believed there was a >50% of restarting the heart when in cardiac arrest (p = 0.03) were more likely to want resuscitation. Patients who reported that their physicians discussed issues regarding life sustaining treatment were less likely to want resuscitation, whereas those who reported that their physicians talked about emotional symptoms were more likely to want more aggressive care (p < 0.01 for all comparisons).

CONCLUSIONS: Lung cancer patients’ beliefs about cancer curability, resuscitation efficacy and fatalism impact their preferences about intubation and resuscitation. Physician discussions about life sustaining treatment may facilitate decisions about end-of-life preferences. Clinicians should ensure open conversations about patient beliefs about cancer prognosis and end-of-life care when discussing lung cancer treatment.

ASSOCIATION OF INFLUENZA VACCINATION COVERAGE IN YOUNGER ADULTS WITH INFLUENZA ILLNESS IN THE ELDERLYGlen Taksler¹; Michael B. Rothberg¹; David M. Cutler^2,3. ¹Cleveland Clinic, Cleveland, OH; ²Harvard University, Cambridge, MA; ³National Bureau of Economic Research, Cambridge, MA. (Tracking ID #1928183)

BACKGROUND: Older adults suffer the majority of influenza morbidity and mortality, but influenza vaccine effectiveness declines with age. It is unknown whether vaccination of nonelderly adults confers additional disease protection upon the elderly.

METHODS: Using the Behavioral Risk Factors Surveillance System Survey, we estimated countywide influenza vaccination coverage among adults aged 18–64year in 313 counties comprising 56.5% of the US population, in each influenza season (Oct–May) of 2002–2010. We linked these data with all Medicare claims for a random sample of 3,090,857 beneficiaries aged ≥65 y residing in these counties, and identified seasonal influenza-like illness (ILI) in each individual during the same years (N = 11,331,129 person-years). Using ICD-9 codes, we identified four types of ILI, from most specific to least specific definition: influenza (primary diagnosis), influenza (primary or secondary diagnosis), pneumonia or influenza, and broader ILI including pneumonia, influenza, bronchitis, cough, or upper respiratory infection. Using logistic regression models, we examined the association between countywide vaccine coverage among adults aged 18–64year and diagnosis of ILI in Medicare beneficiaries, adjusting for comorbidity, influenza season severity, demographics, countywide health and socioeconomic variables, and dummy variables for influenza season and state of residence. We repeated the analysis for hospitalizations for ILI. Results were stratified by documented receipt of a seasonal influenza vaccine in each Medicare beneficiary. An adjusted odds ratio (AOR) for influenza diagnosis in the elderly that varied inversely with vaccine coverage among nonelderly adults would suggest herd immunity. Additionally, we expected to observe a lower AOR with more specific definitions of ILI than with less specific definitions, and during peak (Dec–Mar) vs. non-peak (Oct–Nov, Apr–May) months of influenza season.

RESULTS: Increases in countywide vaccine coverage among 18–64year-olds were associated with lower adjusted odds of influenza-like illness in the elderly. Compared with elderly residents of counties with ≤15% of 18–64year-olds vaccinated, the AOR for a primary diagnosis of influenza was 0.90 (95% CI = 0.87–0.94) for elderly residents of counties with 16%–20% of 18–64year-olds vaccinated, 0.84 (95% CI = 0.81–0.88) for elderly residents of counties with 21%–25% of 18–64year-olds vaccinated, 0.78 (95% CI = 0.74–0.82) for elderly residents of counties with 26%–30% of 18–64year-olds vaccinated, and 0.77 (95% CI = 0.73–0.81) for elderly residents of counties with ≥31% of 18–64year-olds vaccinated (P for trend <0.001). Weaker associations were observed for less specific definitions of influenza; elderly residents of counties with ≥31% of 18–64year-olds vaccinated had an AOR of 0.78 (95% CI = 0.74–0.81) for a primary or secondary diagnosis of influenza, 0.96 (95% CI = 0.95–0.98) for pneumonia or influenza, and 0.95 (95% CI = 0.94–0.96) for the broader definition of ILI, compared with elderly residents of counties with ≤15% of 18–64year-olds vaccinated. The association was stronger among vaccinated elderly (AOR for primary influenza in counties with ≥31% vs. ≤15% of 18–64year-olds vaccinated = 0.63, 95% CI = 0.59–0.68) than for unvaccinated elderly (AOR = 0.84, 95% CI = 0.79–0.89) (P for difference < 0.001). The association was also stronger in peak months (AOR for primary influenza in counties with ≥31% vs. ≤15% of 18–64year-olds vaccinated = 0.75, 95% CI = 0.70–0.79) than in non-peak months (AOR = 0.84, 95% CI = 0.78–0.90) (P for difference = 0.002). For hospitalizations, overall associations were in the expected direction, but trends were not significant due to fewer hospitalizations.

CONCLUSIONS: In a large, nationwide sample of Medicare beneficiaries, influenza vaccination among 18–64year-olds was inversely associated with influenza-like illness in the elderly.

ASSOCIATION OF LONG-TERM OPIOID AND/OR BENZODIAZEPINE RECEIPT AND MEDICATION COUNT WITH MORTALITY AMONG HIV-INFECTED AND UNINFECTED PATIENTSDaniel Weisberg¹; Kirsha S. Gordon²; William Becker^3,2; E. Jennifer Edelman³; J. R. Gaither¹; Robert D. Kerns^1,2; Janet P. Tate²; Amy C. Justice^1,3; David A. Fiellin³. ¹Yale University, New Haven, CT; ²VA Connecticut Healthcare System, West Haven, CT; ³Yale University, New Haven, CT. (Tracking ID #1936560)

BACKGROUND: In the United States, prescribing of controlled substances, including opioid analgesics and benzodiazepines (BZDs), is on the rise. Opioids and BZDs are known to have synergistic effects on sedation and respiratory depression leading to overdose. Those receiving long-term (LT) opioids and BZDs, older patients and those treated for HIV infection deserve particular attention due to greater exposure, the prevalence of polypharmacy and a diminished capacity for drug metabolism and elimination. We previously demonstrated an association between polypharmacy and mortality among HIV infected and uninfected (HIV+/−) patients. The prevalence and harms of LT opioid and/or BZD receipt beyond polypharmacy in HIV+/− patients are unknown.

METHODS: We used data from the Veterans Aging Cohort Study-Virtual Cohort to analyze the association between opioid/BZD receipt and all-cause mortality. Patients with cancer diagnoses, and HIV+ patients not on anti-retroviral therapy (ART) were excluded. LT medication receipt was defined (using pharmacy fill/refill data) as greater than 90 consecutive days of therapy allowing a 30-day gap. Medication count was determined by average LT concurrent medications over the study period, excluding ART, opioids and BZDs. All-cause mortality was based on the Veterans Health Administration Vital Status File. To account for confounding by indication, propensity scores for opioid/BZD receipt were generated using a logistic regression model including demographics, smoking, pain diagnoses, HIV and HCV status and other medical and psychiatric conditions. Individuals were 1:1 matched by propensity score, using a greedy algorithm. Cox regression was used to generate hazard ratios for all-cause mortality. In sensitivity analysis, separate models were run by HIV status.

RESULTS: From 64,441 eligible individuals, with 14550 opioid and/or BZD recipients, 12,881 pairs were included in the matched sample. The propensity score model had good discrimination (c = .76) and produced a well-balanced sample (Table 1). There were 758 deaths. Both opioid and BZD receipt were associated with a >30% increase in mortality; patients receiving both had a >50% increase in mortality. Each additional medication was associated with 4% higher mortality risk. Results from HIV+/− strata were similar to overall findings (Table 2).

CONCLUSIONS: LT opioid and BZD receipt and average medication count each showed a significant independent risk for mortality, after adjustment for confounding by indication using propensity score matching. The incremental harm associated with opioid and/or benzodiazepine receipt was greater than would be expected due to their contribution to polypharmacy alone. These observational findings suggest the need for designing and testing interventions prioritizing the safe use of opioids and BZDs.

Table 1: Propensity score matching

Table 2: Mortality, Hazard Ratio (95% Confidence Intervals)

ASSOCIATION OF OPIOID USE WITH FALLS AND FRACTURES AMONG OLDER MEN WITH MUSCULOSKELETAL PAINErin E. Krebs^1,2; Misti Paudel²; Brent C. Taylor^1,2; Douglas Bauer⁴; Howard A. Fink^1,2; Nancy E. Lane³; Kristine E. Ensrud^1,2. ¹Minneapolis VA Health Care System, Minneapolis, MN; ²University of Minnesota, Minneapolis, MN; ³University of California at Davis, Sacramento, CA; ⁴University of California at San Francisco, San Francisco, CA. (Tracking ID #1936417)

BACKGROUND: Older adults are disproportionately affected by painful musculoskeletal conditions and receive more opioid analgesics than younger adults, but insufficient data are available about potential benefits and harms of opioid therapy in older adults. The objectives of this study were to evaluate potential adverse effects of opioid analgesics on prospectively assessed falls and fractures among older men with persistent musculoskeletal pain.

METHODS: This is a secondary analysis of data from the Osteoporotic Fractures in Men (MrOS) study, a longitudinal cohort study of 5994 community-dwelling men age ≥65years. This analysis includes MrOS participants who reported having back, hip, or knee pain most or all of the time at baseline. Medication exposure and covariate data were collected from participants at baseline and two follow-up visits. Participants were contacted every 4months to assess fall and fracture outcomes; follow-up contacts were >99% complete. For falls, the outcome was one or more falls reported during each 4-month period. Fracture outcomes were incident events confirmed with radiographic reports and evaluated in 3 categories: any clinical fracture; major osteoporotic fracture (fractures of the hip, wrist, spine and shoulder); and hip fracture. Propensity scores (PS), representing the conditional probability of receiving opioids, were estimated for each visit using a logistic regression model that included age, BMI, smoking, medical comorbidities, frailty status, activity level, cognitive performance, pain location, and other covariates. To achieve adequate balance of covariates between opioid users and non-users within PS quintiles, men with PS < 0.03 (meaning <3% conditional probability of receiving opioids) were excluded from the primary analyses. Generalized estimating equations (GEE) were used to model the relationship between repeated measures of opioid use and falls, controlling for time-varying PS quintile. Cox proportional survival analyses were used to model associations between opioid use and incident fractures.

RESULTS: The analytic cohort included 2,732 men with back, hip, or knee pain at baseline. A total of 351 men (12.8%) reported opioid use at one or more study visits. Over 9.1 (SD 4.0) years of follow-up, 2069 participants reported at least one fall, 318 had any clinical fracture, 149 had a major osteoporotic fracture, and 60 had a hip fracture. In the primary model, the relative risk of falling was 1.10 (95% CI: 0.97, 1.24) for opioid users compared with non-users. Point estimates for all fracture outcomes were higher for opioid users than non-users, but none of the associations were statistically significant (Table).

CONCLUSIONS: We found non-significantly higher fall and fracture rates for opioid users compared with non-users. This analysis of MrOS data has major advantages over prior studies, including detailed assessment of potential confounders and rigorous prospective ascertainment of fall and fracture outcomes. Study limitations include lack of detailed opioid dose data and infrequent medication assessment. Power was limited for major osteoporotic fracture and hip fracture events, so we cannot exclude a moderate effect of opioid use on these outcomes.

Hazard ratio (HR) of fracture outcomes for opioid users vs. non-users

ASSOCIATIONS BETWEEN INTERNAL MEDICINE RESIDENT SCHOLARSHIP AND CLINICAL PERFORMANCELuke A. Seaburg; Amy T. Wang; Colin P. West; Darcy Reed; Andrew J. Halvorsen; Gregory Engstler; Amy Oxentenko; Thomas J. Beckman. Mayo Clinic, Rochester, MN. (Tracking ID #1936416)

BACKGROUND: The Accreditation Council for Graduate Medical Education requires that resident physicians participate in scholarly activities. Studies have shown that dedicated research time during residency is associated with increased publication, and residents who publish are more likely to select careers in academic medicine and achieve higher academic rank. However, it is unknown whether research productivity among residents is related to broad measures of clinical achievement during residency training. Therefore, we examined associations between the quantity of Mayo Clinic internal medicine residents’ peer-reviewed publications and measures of their knowledge (ABIM certification examination scores), skills (mini-clinical examination scores) and multi-source evaluations of performance.

METHODS: This was a longitudinal study of 312 post-graduate-year three (PGY-3) residents at the Mayo Clinic in Rochester, Minnesota from 2006 to 2012. To quantify scholarship, we identified peer-reviewed articles published in Ovid MEDLINE between July of each resident’s match year and the end of their graduation year. Outcomes included ABIM certification examination scores, mini clinical examination (mini-CEX) scores, and validated multi-source assessments of performance by resident-peers, faculty, and non-physicians. Scores for the clinical performance assessment items were averaged to form an overall score ranging from 1 to 5. Covariates included sex (male/female), medical school (U.S. versus international medical graduate), track (categorical versus clinician-investigator), PhD (yes/no), age at PGY-3 start, and percent correct on PGY-3 ITE examination. Associations between quantity of resident publications and first author status—and ABIM, mini-CEX and clinical performance assessment scores—were determined using multivariate linear regression analysis. The threshold for statistical significance was set at p < 0.01 to account for multiple comparisons.

RESULTS: Of the 312 PGY-3 residents, 199 (63.8%) were male, 263 (84.3%) were U.S. medical graduates, 282 (90.4%) were categorical, and 11 (3.5%) had PhD degrees. Their average age at PGY-3 start was 29.7years. The residents published 667 papers, of which 462 were research papers, 204 were case reports, and 288 were first-authored. On adjusted analysis, multi-source evaluations of clinical performance were significantly associated (beta; p-value) with the numbers of research articles (0.013; 0.003), overall publications (0.013; 0.0009), and first-authored articles (0.017; 0.007). There were no statistically significant associations between quantity of publications and ABIM certification examination or mini-CEX scores.

CONCLUSIONS: This may be the first study to demonstrate that, among resident physicians in training, scholarly achievement is associated with clinical performance as determined by observation-based assessments from resident peers, faculty, and non-physicians. Furthermore, performance assessment scores were linked to the publication of journal articles that were first-authored and research-focused. These findings suggest that scholarly productivity may be an indicator of clinical performance during residency training, and underscore the importance of research programs in graduate medical education.

ASSOCIATIONS BETWEEN MEDICAL HOME CHARACTERISTICS AND SUPPORT FOR PATIENT ACTIVATION IN THE SAFETY NET: UNDERSTANDING DIFFERENCES BY RACE, ETHNICITY, AND HEALTH STATUSRobert S. Nocon¹; Yue Gao¹; Kathryn E. Gunter¹; Janel Jin²; Lawrence P. Casalino³; Michael T. Quinn¹; Sarah Derrett⁴; Wm Thomas Summerfelt⁵; Elbert S. Huang¹; Sang Mee Lee⁶; Marshall Chin¹. ¹University of Chicago, Chicago, IL; ²Johns Hopkins Bloomberg School of Public Health, Baltimore, MD; ³Weill Cornell Medical College, New York, NY; ⁴Massey University, Palmerston North, New Zealand; ⁵Advocate Healthcare, Chicago, IL; ⁶University of Chicago, Chicago, IL. (Tracking ID #1936110)

BACKGROUND: Few studies have evaluated whether the patient-centered medical home (PCMH) supports patient activation, and existing evidence is mixed. No studies have evaluated whether support for patient activation differs among racial and ethnic groups in a PCMH. This is critical because activation is lower on average among minority patients. We sought to assess the association between clinic PCMH characteristics and patient perception of clinic support for patient activation, and whether that association varies by patients’ self-reported race, ethnicity, and health status.

METHODS: We conducted a cross-sectional analysis of PCMH characteristics and patient-assessed clinic support of patient activation in 24 safety net clinics across five states. PCMH characteristics were measured via surveys of 271 providers and staff. The provider and staff survey produced a 0 (worst) to 100 (best) PCMH score based on a scale created by the authors and described previously in the literature. Clinic scores were created by averaging provider/staff responses. Clinic support of patient activation was measured via surveys of 1,656 patients. The patient survey used the patient activation scale of the Patient Assessment of Care for Chronic Conditions to produce a 0 (worst) to 100 (best) score for clinic support for patient activation. Patient race, ethnicity, and health status were based on self-report in the patient survey. To investigate the relationship between PCMH characteristics and patient activation, while allowing for a clustering effect of patients within clinics, we fitted multivariate models using generalized estimating equation models with an exchangeable correlation structure. We analyzed interactions terms to assess how the association of PCMH characteristics and patient activation varied by race/ethnicity and health status subgroups. We interpret the association in terms of a 10-point change in PCMH score, a difference we found to be operationally meaningful in previous work.

RESULTS: We received 214 (79.0%) provider and staff survey responses and 735 (44.4%) patient survey responses. Mean PCMH score among the 24 clinics was 58.7 (SD = 6.4). The mean score for patient perception of clinic support of patient activation was 68.8 (SD = 30.0). Across all patients, a 10-point higher PCMH score was associated with a 5.6-point higher score for patient perception of clinic support for patient activation (95% CI, 1.3–9.9). The association between PCMH score and patients’ perception of clinic support for patient activation was particularly strong among minority patients in fair or poor health: a 10-point higher PCMH score was associated with a 16.2-point (CI 1.7–30.6) higher score for clinic support of patient activation among Hispanic patients in fair or poor health and a 34.1-point higher score among black patients in fair or poor health (CI 9.1–59.1). The effect of PCMH score on patient activation score among black patients with poor/fair health status was statistically significantly different from the effect seen among non-Hispanic white patients in good or better health.

CONCLUSIONS: In a population of safety net patients, PCMH characteristics showed a moderate, positive association with patients’ perception of clinic support for activation; the magnitude of association was notably larger for minority patients in poor/fair health status. The PCMH may be promising for reducing disparities in patient activation for ill racial and ethnic minority patients.

ATTITUDES TOWARD INTERNAL MEDICINE RESIDENT PROFESSIONALISM TRAININGLaura Nichols; Kathlyn E. Fletcher; Brent Nichols. Medical College of Wisconsin, Wauwatosa, WI. (Tracking ID #1934886)

BACKGROUND: Professional conduct is a core competency of medical practice and residency training. Defining, teaching and evaluating professionalism is challenging and requires collaboration between residents and faculty. In order to facilitate this effort, it is important to understand the beliefs and attitudes of those involved in the process. It was our objective to understand current modalities used, attitudes toward currently utilized curriculum and potential areas for improvement in professionalism training.

METHODS: Two surveys (one for residents and another for program directors) were developed using previously identified methods of teaching and evaluating professionalism. We distributed the survey link via email. Data was analyzed for associations using a chi-squared test.

RESULTS: Sixty-seven program directors and 589 residents completed the respective surveys. The majority (94%) of residents felt that professionalism education was relevant to their development as physicians. Forty-two percent of programs reported having a formal professionalism curriculum, though the majority of program directors (86.6%) and residents (87.1%) felt that residents could be taught professionalism. Residents who reported having a formal professionalism curriculum were significantly more likely to rate the quality of their professionalism training as adequate (27%) or very good (28%) (P < 0.0001). Commonly utilized evaluation modalities included faculty evaluation forms (97.0%) and completion of administrative tasks (85.1%). Role modeling was the most utilized teaching modality (80.3%). Residents noted faculty evaluation (66.0%) and ancillary staff evaluation (61.9%) to be most effective evaluation methods, while Objective Structured Clinical Examinations (OSCE) and traditional examinations ranked lowest. Role modeling (66%) and faculty mentorship (48.7%) were felt by residents to be the most useful methods of teaching professionalism.

CONCLUSIONS: We examined attitudes of residents and program directors toward professionalism curricula to aide in the formation of effective professionalism education. Consistent with previous literature, formal didactics and examinations were noted to be less effective than staff evaluations and role modeling. Less than half of programs reported having professionalism curricula; however, residents who had a formal curriculum were more likely to report high quality professionalism teaching. A multifaceted approach with a focus on faculty and ancillary staff evaluation, role modeling and mentorship is likely to be the most well-received curriculum.

AUTOMATED ELECTRONIC SEPSIS ALERTS: A SYSTEMATIC REVIEWAnil N. Makam¹; Oanh K. Nguyen¹; Andrew Auerbach². ¹University of Texas Southwestern, Dallas, TX; ²University of California San Francisco, San Francisco, CA. (Tracking ID #1938467)

BACKGROUND: Timely treatment of sepsis, including adequate fluid resuscitation and appropriate antibiotic administration, decreases morbidity and mortality. However, there are diagnostic and therapeutic delays in deploying early-goal directed therapies due to the resource intensive nature of these protocols. To circumvent these delays, automated, electronic sepsis alerts are being implemented to facilitate the delivery of timely and effective care. We sought to systematically evaluate whether automated, real-time electronic sepsis alerts can accurately identify sepsis, and improve process measures and outcomes for patients with sepsis.

METHODS: We systematically searched MEDLINE, Embase, The Cochrane Library, and CINHAL from database inception till August 13, 2013. Two independent reviewers selected studies that empirically evaluated one or both of our prespecified objectives. We excluded studies of bacteremia, nosocomial infections, or rapid response interventions. Two independent reviewers extracted data and assessed the risk of bias. Accuracy of sepsis detection was measured by sensitivity, specificity, positive and negative predictive values and likelihood ratios. Efficacy was assessed by changes in various process measures (i.e. time to antibiotics) and outcomes (length of stay, mortality). Heterogeneity in setting, design, and definitions of the alert precluded a meta-analysis.

RESULTS: We searched 1,091 citations and identified six studies for inclusion, four for the identification of sepsis (n = 34,328) and four for the evaluation of automated, real-time sepsis alerts (n = 6,723). The definition of sepsis alert thresholds varied, but included abnormal systemic inflammatory response syndrome (SIRS) criteria or abnormal SIRS criteria plus organ dysfunction. The accuracy of alerts to identify sepsis varied greatly depending on the clinical setting and the definition of sepsis used for the alert system, with sensitivity ranging from 36.2 to 98.9%, specificity, 30.7–99.6%, positive likelihood ratio, 1.4–145.8, and negative likelihood ratio, 0.04–0.86. There was modest evidence for improvement in process measures, such as lactate collection and antibiotic administration, among patients in non-ICU settings; however, without corresponding improvements in mortality or length of stay. We found no studies that reported on overuse and potential harms due to false positive alerts.

CONCLUSIONS: Automated sepsis alerts derived from electronic health records may improve care processes but tend to have poor positive predictive value and do not improve mortality or length of stay. Future efforts should develop and study methods for sepsis alert systems which avoid the potential for alert fatigue while improving outcomes.

AUTOPSY IN THE AGE OF ADVANCED DIAGNOSTIC TECHNOLOGYDavid Chia²; Ali Ershadi¹; Majid Sadigh¹; Steven I. Aronin². ¹Western Connecticut Health Network, Danbury, CT; ²Yale, Waterbury, CT. (Tracking ID #1928285)

BACKGROUND: Traditionally, the role of autopsy has had many benefits to the healthcare system, including illumination of the exact cause of death, determination of the rate of clinical errors, assurance of the quality of care, and identification of new diseases or changes in preexisting ones. However, autopsy rates have declined worldwide. In the United States, the in-hospital autopsy rate has fallen to 5–10% from its peak of 70% in the early 1960s. Furthermore, a heavy reliance on advanced diagnostic technologies have given an impression of diagnostic adequacy that invalidates the present need for autopsy. However, despite these advances in technology, the reported discrepancy between clinical and autopsy diagnoses remains unchanged at 10–30%.

METHODS: We conducted a retrospective review of all autopsies performed on patients who died in two community hospitals between January 2008 and December 2012. Major discrepancies between clinical and autopsy diagnoses were classified into two categories: significant errors that had the proper diagnosis been made would have changed management and possibly outcome (type I) and significant errors that had the proper diagnosis been made would not have changed management or outcome (type II).

RESULTS: 2,409 deaths occurred during the study period and autopsies were conducted on 284 patients (11%). Autopsies performed on fetuses, patients who were dead on arrival to the hospital, and those with inadequate documentation were excluded from our study. Major discrepancies were noted in 20% (33/164) of cases of which 79% (26/33) were type I errors. The most commonly misdiagnosed conditions included pneumonia (30%), myocardial infarction (18%), pulmonary embolism (9%) and pneumonitis (9%). Nearly half of patients (17/33) with major discrepancies died under unclear clinical circumstances. Autopsy established the cause of death in 88% (15/17) of these cases. Major discrepancies occurred mainly because either the diagnosis was not considered on the differential diagnosis (55%) or the diagnosis was considered, but the result of appropriate diagnostic evaluation was misleading (30%).

CONCLUSIONS: Despite advances in medical technology, major discrepancies between clinical and autopsy diagnoses continue to occur at relatively high rates. The majority of these discrepancies would have changed management decisions and may have altered outcomes if the proper diagnosis had been made. These discrepancies often occurred because the proper diagnosis was either not considered in the differential diagnosis or the result of appropriate diagnostic testing was misleading. As a result, autopsy should continue to have an important role in the healthcare system, while physicians should maintain broad differential diagnoses and be critical of unexpected results on diagnostic evaluation

AWARENESS OF CENTRAL VENOUS CATHETERS: A PATIENT SAFETY SURVEYRachel E. Thompson¹; Natalie Melin¹; Vineet Chopra². ¹University of Washington, Seattle, WA; ²University of Michigan, Ann Arbor, MI. (Tracking ID #1938689)

BACKGROUND: Central lines are commonly used in the care of hospitalized patients. Each day a central line is in place the risk of complications such as bloodstream infection rises. Yet central lines often remain in place after they are no longer clinically useful. Preliminary data suggests one obstacle to the timely removal of central lines may be physician being unaware—as was demonstrated in the case of urinary catheters. The goal of this study is to determine the extent of physician awareness of their patients’ central lines.

METHODS: We interviewed intern, resident, and attending physicians on inpatient medical services at an academic hospital. Patients hospitalized on the medicine non-ICU, medicine ICU, cardiology and hospitalist services were included. Physicians were read a list of patients for whom they had direct responsibility and asked whether each had (1) a central venous catheter (CVC) or (2) a peripherally inserted central catheter (PICC). “Yes” or “no” responses were recorded for both types of central line. Physician responses were cross-checked with the hospital’s validated daily central line report for accuracy.

RESULTS: 375 patient records were accessed and their providers interviewed regarding awareness of presence of a CVC or PICC. There were 218 interviews with interns, 264 with senior residents and 347 with attendings. Fifty-two cases had a CVC and 23 a PICC. Of those with either a CVC or PICC, overall provider recall accuracy ranged from 82 to 93%. On teaching teams, attendings were significantly more likely to miss a central line than residents or interns (p < 0.01). There was no significant difference in missed lines between teaching attendings and hospitalists. Of the ICU attending interviews, lines were present in 36% compared to only 13% among non-ICU. Attendings with work that included ICU care were less likely to miss a line (13.9 v 36.1%, p <0.05). These attendings were also more likely to falsely recall a line when no line was present than the non-ICU attendings (15.4% v 5.3%, p < 0.01).

CONCLUSIONS: Our results demonstrate that many physicians are not aware of whether their patient has a central line. Errors in recall occurred in both directions, potentially suggesting a deep-rooted unawareness. The frequency of lines among patients may influence the direction of recall error. The overall greater awareness observed among interns could relate to smaller case-loads. Each day a physician fails to remove an unnecessary line unjustly exposes the patient to an increasing risk of complications. It is important to target awareness of central lines to improve patient safety.

AWARENESS OF PREDIABETES AND ENGAGEMENT IN DIABETES RISK-REDUCING BEHAVIORSAnjali Gopalan^1,3; Ilona Lorincz⁴; Christopher Wirtalla⁶; Judith A. Long^2,3; Steven C. Marcus^2,5. ¹Philadelphia VA Medical Center, Philadelphia, PA; ²Philadelphia VA Medical Center, Philadelphia, PA; ³Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA; ⁴Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA; ⁵University of Pennsylvania, Philadelphia, PA; ⁶Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA. (Tracking ID #1933194)

BACKGROUND: Numerous high quality studies have demonstrated the benefit of lifestyle modification, including weight loss, exercise, and a healthy diet, in the prevention of diabetes among those with prediabetes. Almost a third of the U.S. adult population meets criteria for prediabetes and without any intervention, 25% will develop diabetes within 3 to 5years. Despite the proven benefits of lifestyle changes and increasing numbers of community-based diabetes prevention programs, only about half of those with prediabetes report engaging in behaviors aimed at preventing the onset of diabetes. One barrier is low patient awareness of being at risk. In this study we aimed to determine whether there is a difference in the performance of diabetes risk-reduction behaviors between individuals who are aware of their prediabetes diagnosis compared to those who are unaware.

METHODS: We conducted a pooled cross-sectional analysis from two consecutive cohorts (2007–2008 and 2009–2010) of the National Health and Nutrition Examination Survey (NHANES). Subjects under age 20, pregnant woman, and those with missing responses to questions relating to diabetes history, physical activity, diet, and weight history were excluded, as were individuals missing all three lab values used to diagnose prediabetes: the hemoglobin A1C (HbA1C), a fasting plasma glucose (FPG), and the two-hour oral glucose tolerance test (OGTT). Patients with a self-reported a history of diabetes were classified as diabetes-aware. The remaining analytic sample was identified using the nationally accepted laboratory criteria to define prediabetes: 5.7%≤A1C<6.5%, 100mg/dL≤FPG<126mg/dL, and/or 140mg/dL≤OGTT<200mg/dL. Patients were classified as prediabetic if they met one of these criteria and did not meet any criteria for diabetes. The prediabetic population was then categorized into prediabetes aware and unaware based on participants’ self-report. We then used multivariate logistic and linear regression to estimate the effect of awareness of prediabetes on the likelihood of the following outcomes: engagement in physical activity, weight-related behaviors, and number of fast food/pizza meals eaten weekly. Any weight related behavior was defined by reported intention to lose weight or maintain weight, while appropriate-weight behavior was defined as an intention to lose weight for those with a BMI ≥ 25 and a desire to maintain weight for those with a BMI < 25.

RESULTS: 34.4% (3,635/10,556) met criteria for prediabetes. 8.5% (310) of those meeting these criteria self-reported a diagnosis of prediabetes. Those who were aware of prediabetes differed significantly from those who were unaware by age (55.6 vs 51.7, p < 0.001), but did not differ based on gender, educational attainment, income, or ethnicity. Subjects aware of prediabetes were also more likely to report a regular source of care (94.2% vs. 85.7%, p < 0.001), have a family history of diabetes (45.3% vs. 35.9%, p = 0,02), have a history of hypertension or hyperlipidemia (58.1% vs 35.1%, p < 0.001 and 53.8% vs. 29.5%, p < 0.001, respectively), and to report that a physician told them they were overweight or obese (57.2% vs. 33.3%, p < 0.001). After adjusting for age, gender, ethnicity, education, number of healthcare visits in the past year, family history of diabetes, being told they were overweight/obese, self-reported limitation on work/activity, BMI, and medications for hypertension and hyperlipidemia, adults who self-reported having prediabetes were significantly more likely to report engagement in any physical activity (any level of intensity) on a weekly basis, adjusted odds ratio (AOR) 1.3, ([95% CI 1.0–1.7]; p = 0.046) and engagement in moderate or vigorous physical activity, AOR 1.3, ([95% CI 1.1–1.5]; p = 0.01). Adults with awareness of prediabetes were also significantly more likely to report engaging in any weight-related behavior AOR 1.8, ([95% CI 1.3–2.5]; p = 0.001) and in appropriate-weight related behavior AOR 1.4, ([95% CI 1.0–2.0]; p = 0.05). There was no difference noted in number of fast-food/pizza meals eaten weekly based on awareness of prediabetes (1.9 meals vs. 1.6 meals, p = 0.5).

CONCLUSIONS: Adults who were aware of meeting diagnostic criteria for prediabetes were significantly more likely to engage in moderate or vigorous weekly exercise and were more likely to report engagement in appropriate weight-related behaviors. While the cross-sectional nature of the data limits conclusions regarding continued adherence to these risk-reduction efforts, engagement is unarguably the first step toward lifestyle changes. Increasing our patients’ awareness of prediabetes could result in increases in exercise and weight management and, most importantly, to decreases in the incidence of diabetes and improved health outcomes.

AZITHROMYCIN IS ASSOCIATED WITH INCREASED CARDIAC EVENTS BUT LOWER MORTALITY FOR OLDER PATIENTS HOSPITALIZED WITH PNEUMONIAEric Mortensen^1,2; Ethan Halm²; Michael J. Fine³; Christopher Johnson^1,2; Antonio Anuzeto⁴. ¹VANTHCS, Dallas, TX; ²University of Texas Southwestern Medical Center, Dallas, TX; ³VA Pittsburgh Health Care System, Pittsburgh, PA; ⁴South Texas Veterans Health Care System, San Antonio, TX. (Tracking ID #1931222)

BACKGROUND: Although clinical practice guidelines recommend combination therapy with macrolides, including azithromycin, as first line therapy for patients hospitalized with pneumonia, recent research suggests that azithromycin may be associated with increased cardiovascular events. The purpose of this study was to examine the association of azithromycin use with all-cause mortality and cardiovascular events for patients hospitalized with pneumonia.

METHODS: We conducted a retrospective national study using Department of Veterans Affairs administrative data of patients hospitalized at any Veterans Administration acute care hospital. We included patients >65years hospitalized with pneumonia in fiscal years 2002–2012. We included only those who received antibiotic therapy concordant with national clinical practice guidelines. Our outcomes included all-cause mortality, cardiac arrhythmias, heart failure, myocardial infarction, and any cardiac event. We used propensity score matching to control for the possible effects of known confounders with conditional logistic regression.

RESULTS: Out of the 77,972 patients identified, our propensity-matched groups were composed of 34,763 azithromycin-exposed and 34,763 matched unexposed. There were no significant differences in potential confounders between groups after matching. We found that 90-day mortality was significantly lower in those who received azithromycin (exposed- 17.4% vs. unexposed- 22.8%, odds ratio [OR] 0.71, 95% confidence interval [CI] 0.68–0.73). However, we found significant increased odds of any cardiac event (43.2% vs 42.2%, OR 1.04, 95% CI 1.01–1.07), myocardial infarctions (4.9% vs. 4.2%, OR 1.17, 95% CI 1.09–1.26), and cardiac arrhythmias (26.5% vs. 25.6%, OR 1.04, 95% CI 1.002–1.07) but not heart failure (26.1% vs. 25.8%, OR 1.02, 95% CI 0.98–1.05).

CONCLUSIONS: In patients hospitalized with pneumonia we found that although azithromycin is associated with a small increase in cardiac events it is also associated with a significant decrease in all-cause mortality. Our study supports the current clinical practice guidelines that recommend the use of azithromycin as part of empiric combination therapy.

BACTERIAL CONTAMINATION OF RESIDENT IPADS: LAME DUCK OR TROJAN HORSE?Tanu S. Pandey¹; Kamaljit Singh²; German E. Giese¹; Michael Alebich¹; Shubhra Gupta¹. ¹John H Stroger Hospital of Cook County, Chicago, IL; ²Rush University Medical Center, Chicago, IL. (Tracking ID #1939546)

BACKGROUND: The advent of the Apple iPad, has brought about a revolution in the delivery of medical care. Though clothing, stethoscope and phone contamination has been well documented, to date, no study has examined the bacterial contamination of iPads. Our study aims to establish the rates of bacterial contamination of iPads used by residents in two urban teaching hospitals. We present the findings from one of these institutions.

METHODS: The study was funded by a collaborative research grant from the parent institutions. Twenty four residents from John H. Stroger Jr., Hospital of Cook County who used an iPad for patient care were enrolled into the study. They provided their iPad to the investigators during work hours for culture at random times. A pre-moistened swab was obtained from two 5cm2 areas of the screen, as well as the “home” button. These swabs were delivered to the microbiology lab promptly and inoculated onto three agar plates for aerobic and anaerobic culture, isolate identification and antibiotic sensitivity. An anonymous questionnaire with demographic information was provided to the residents at time of swab.

RESULTS: All 24 devices were either iPads or iPad mini. There were five PGY-1, eight PGY-2, and eleven PGY-3 residents. Ninety-two percent residents took their device regularly inside patients’ rooms and 67% placed them on the bed, table or elsewhere. Sixty-three percent carried their device in their hands as opposed to a case or coat pocket. Ninety-six percent used the same device for work and personal reasons and 50% did not clean them at all. All devices grew multiple skin flora including coagulase negative staphylococcus. One device grew MRSA, three devices grew MSSA, one Acinetobacter and six grew alpha-hemolytic streptococcus. There was no significant statistical difference in contamination between those who cleaned their devices versus those who did not (OR 2.0, 95% CI 0.3843 to 10.4096, p value = 0.4102).

CONCLUSIONS: The iPad is changing how medicine is practiced; however, no study has examined potential harms afforded by these devices. Our study revealed that the iPad does harbor potentially dangerous pathogens and little regulations guide the decontamination of these devices at the bedside. The relatively low prevalence of organisms is reassuring but further studies in large numbers are needed to establish the role of the iPad in increasing hospital-acquired infections. Based on the behavior of residents in this study, standard precautions for the use of these devices around patients is recommended.

BAR AND NIGHTCLUB INTERVENTION TO DECREASE YOUNG ADULT SMOKING IN NEW MEXICOSara Kalkhoran¹; Pamela M. Ling^1,2; Jeffrey W. Jordan³. ¹University of California, San Francisco, San Francisco, CA; ²University of California, San Francisco, San Francisco, CA; ³Rescue Social Change Group, San Diego, CA. (Tracking ID #1937993)

BACKGROUND: Over 20% of young adults in New Mexico are current smokers, which exceeds current national estimates. We evaluated the effect of an anti-tobacco Social Branding intervention on daily and non-daily smoking rates among young adult bar patrons.

METHODS: A series of cross-sectional surveys were collected from adults aged 18–26 in bars and nightclubs popular among young adults in Albuquerque, New Mexico from 2009 to 2013 using randomized time-location sampling. A Social Branding intervention was implemented in Albuquerque during this time. Social Branding is a variant of social marketing using commercial marketing strategies to compete with tobacco marketing by creating a smoke-free brand to promote smoke-free lifestyles within a specific bar subculture. This intervention focused on the “Partier” culture, which was characterized by attendance at large nightclubs, fashion consciousness, valuing physical attractiveness, and achieving high social status by exuding an image of confidence and financial success. Multivariate multinomial regression was used to evaluate differences in rates of daily (smoked on 25–30 of past 30days) and non-daily (smoked 1–24 of past 30days) cigarette use between baseline and each of 3years of follow up, controlling for demographic characteristics, other risk behaviors, and tobacco-related attitudes.

RESULTS: Data was collected from 1066 individuals at baseline, and 767, 1242, and 1231 participants at follow-up years 1, 2, and 3, respectively. Current smoking rates decreased from 47.4% at baseline to 37.4% at follow-up year 3 (p < 0.001). In multivariate multinomial regressions, those surveyed at year 3 had significantly lower odds of being either a current daily (OR = 0.41, 95%CI [0.26–0.64]) or non-daily (0.57, [0.41–0.80]) smoker compared to baseline. Factors associated with increased odds of both daily and non-daily smoking included lower education, LGBT sexual orientation, advertising receptivity, binge drinking, and higher trend sensitivity (a measure of social personality traits and frequency/intensity of socializing in bars) (Table 1). Other factors associated with daily and non-daily smoking are shown in Table 1.

CONCLUSIONS: Social Branding interventions aimed at young adult bar patrons may help reduce smoking rates in this population. Bar-based interventions tailored to specific subculture values and lifestyles are an efficient way to address risky behaviors in this hard-to-reach population that continues to have high rates of cigarette smoking.

Predictors of Being a Non-daily or Daily Smoker Among Young Adult Bar Patrons in Albuquerque, New Mexico

*Assessed by agreement with questions regarding support for action against the tobacco industry **Determined by pictures of people with whom the participants were more likely to associated and bars which the participants were more likely to frequent

BARRIERS AND FACILITATORS TO PREVENTIVE CANCER SCREENING AMONG HMONG- AND SPANISH-SPEAKING PATIENTSElizabeth A. Jacobs^1,3; Rebecca J. Schwei¹; Maichou Lor². ¹School of Medicine and Public Health, University of Wisconsin, Madison, WI; ²School of Nursing, University of Wisconsin, Madison, WI; ³School of Medicine and Public Health, University of Wisconsin, Madison, WI. (Tracking ID #1938836)

BACKGROUND: Language barriers in health care are a large and growing problem in the U.S. Having limited English proficiency (LEP) is a well-recognized risk factor for disparities in health care quality and outcomes. Despite access to high quality interpreter services, LEP patients still experience disparities in preventive cancer screening. We conducted a qualitative study to better understand what factors might contribute to the disparities of cancer screening rates and their reduction among LEP patients seen in primary care.

METHODS: We conducted 28 semi-structured interviews with LEP patients who were eligible for preventive cancer screenings (n = 11 Hmong, n = 10 Spanish) and providers (n = 7 Physician) in a large academic medical system. Patient interviews were conducted in Hmong or Spanish. We interviewed general internal and family practice physicians with varying percentages of LEP patients in their practice. All interviews were audio taped, transcribed verbatim, translated and coded for interpretation. Hmong, Spanish, and provider interviews were coded separately. Themes were developed using content analysis.

RESULTS: Physician recommendation, doctors who care about and respect their patients, gender-concordant doctors, and familial experience with cancer facilitated cancer screening in both Hmong- and Spanish-speaking participants. Physicians identified high quality interpreters, electronic medical record reminders, and educational materials in appropriate language as facilitators to delivering preventive cancer screening to their LEP patients. Two common barrier themes for Hmong- and Spanish-speaking patients were: (1) previous negative health care experiences, which created distrust of health care, and (2) lack of knowledge about cancer, what cancer prevention can do, and how to access community resources. Additional barriers for Hmong patients included poor quality interpreters and the need for health care decisions to be made by the nuclear family. Spanish-speaking patients uniquely identified lack of money or insurance, not enough time, and machismo in men as additional barriers to receipt of preventive cancer screening. Physicians reported that competing social, economic, and health demands of patients, visit time constraints, cultural beliefs of patients, and the nature of the screening tests (e.g. embarrassing) as barriers for LEP patients to seek screening.

CONCLUSIONS: We found that cancer screening is influenced by many different personal, family, interpreter, provider, health care system, and community factors. This study provides insights for addressing cancer health disparities for these two ethnic minority populations in our community. Culturally-tailored community cancer education about prevention and cancer in general, and preventive cancer screenings specifically, is a first step to increasing knowledge and acceptability of preventive cancer screening in Hmong- and Spanish-speaking patients. Additionally, helping providers navigate the barriers that they may face when suggesting preventive cancer services to LEP patients may also increase the rates of preventive cancer screenings in Hmong- and Spanish-speaking patients.

BEARING WITNESS: IMPACT OF A BRIEF INTERVENTION ON RESIDENT ADVERSE EVENT REPORTINGJennifer B. Cowart^1,2; Cynthia Peacock²; John Lin²; Addison A. Taylor^1,2. ¹Houston MEDVAMC, Houston, TX; ²Baylor College of Medicine, Houston, TX. (Tracking ID #1923141)

BACKGROUND: Recent ACGME guidelines require training in quality improvement and patient safety for resident physicians. Prior to the release of these guidelines, it was not clear how much training residents received in incident reporting. We sought to evaluate resident knowledge of hospital incident reporting systems, before and after a brief intervention aimed at increasing residents’ submissions of incident reports.

METHODS: Internal medicine residents at a large academic program were asked to participate in an anonymous survey before and after a brief educational intervention. Invitations for the pre-test survey were sent by email in August 2012. Questions were given in the form of a statement with agreement rated on a 7-point Likert scale (from “strongly disagree” to “strongly agree”). Residents were also asked to enter the number of adverse event reports they had ever submitted. After survey completion, monthly lectures were given during regularly scheduled education time (15min per month) at each hospital site. Reference guides for the incident reporting systems at each facility were added to the departmental website. Invitations for the post-test survey were sent in May 2013.

RESULTS: 189 residents were invited to participate in the 9-question survey on two occasions (378 total). 138 (73%) completed all questions in 2012 and 84 (44%) completed all questions in 2013. There were significant increases in residents’ stated knowledge of mechanisms to report adverse events at two training institutions, with the median response changing from “disagree” to “agree” on a 7-point Likert scale (p = 0.000 for both). Before specific instruction in incident reporting, only 24.1% of residents stated they had ever reported an adverse event. At the post-survey, 45.3% of the responding residents stated they had reported one or more adverse events (p = 0.001). A majority of respondents agreed that they worked in a “culture of safety” (76.6% before, 78.6% after). There was no significant difference in the proportion of respondents who disagreed with this statement from before and after the intervention (10.2% vs 9.5%). The same held true of the proportion of residents who agreed that they were worried about being punished for reporting adverse events both before and after the intervention (27.5% vs 28.6%).

CONCLUSIONS: Brief education in adverse event reporting led to an increase in resident awareness of both reporting mechanisms and in self-reported submissions of adverse event reports. The increase in reporting did not correspond with an increase in residents perceiving themselves to be safe from reprisal. Areas for further investigation include identifying barriers to residents’ reporting witnessed events and whether residents’ concerns about reprisal for reporting of adverse events are a significant deterrent.

After education on adverse event reporting, residents were more likely to say they had reported at least one adverse event compared to before education on adverse event reporting.

BEDSIDE INTERPROFESSIONAL ROUNDS: MEDICINE PROVIDERS’ PERCEPTIONS OF BENEFITS AND BARRIERSJed Gonzalo¹; Ethan Kuperman²; Erik B. Lehman¹; Paul Haidet¹. ¹Penn State College of Medicine, Hershey, PA; ²University of Iowa, Carver College of Medicine, Iowa City, IA. (Tracking ID #1937728)

BACKGROUND: Interprofessional collaborative care models improve the quality of medical care. On inpatient medicine services, similar conceptualization amongst nurses and physicians regarding the process and goals of interprofessional rounds are required for successful implementation. However, little work has investigated front-line providers’ perceptions regarding bedside interprofessional rounds. We evaluated the perceptions of nursing staff, resident, and attending physicians regarding the benefits and barriers to bedside interprofessional rounds.

METHODS: In June 2013, a cross-sectional survey was sent via email to all medicine nursing staff, resident and attending physicians in our hospital to assess their perceptions regarding the benefits and barriers to bedside interprofessional rounds, defined as “encounters that include the team of providers, at least two physicians plus a nurse or other care provider, discussing the case at the patient’s bedside.” The survey was developed from prior work on bedside rounds and a literature review; 16 items were related to “benefits” and 18 items were related to “barriers” to bedside interprofessional rounds. Descriptive statistics, non-parametric Kruskal-Wallis, and non-parametric correlation statistics were used.

RESULTS: Of 171 surveys sent, data were collected from 149 respondents (87% response). Highest-ranked benefits to bedside interprofessional rounds were “improves communication between nurses and physicians,” “improves awareness of clinical issues need to be addressed,” and “improves team-building between nurses and physicians;” lowest-ranked benefits were “decreases patients’ hospital length-of-stay” and “improves timeliness of consultations.” Nursing staff reported more favorable ratings compared to both attending and resident physicians for all 16 items (all p-values <0.05); the rank order between respondent groups showed a high degree of correlation (r = 0.92–0.96, p < 0.001). Highest-ranked barriers were “nursing staff have limited time,” “time required for encounters,” and “coordinating the start time of encounters;” lowest-ranked barriers were “patient lack of comfort” and “physicians” lack of bedside skills.” The rank order between respondent groups showed variable correlation (nurse-attending 0.62, nurse-resident 0.76, attending-resident 0.82).

CONCLUSIONS: Nursing staff were more likely to report positive perceptions about the benefits of bedside interprofessional rounds than resident and attending physicians. Overall, the greatest perceived benefits related to improved communication. Standardization of the rounding process and patient inclusion in conversations may allow knowledge assimilation regarding care plans most important to patient care and disposition. Reasons for nurses’ more positive perceptions than physicians are unclear but may reflect their experience with benefits or professional culture of this patient-centered activity. Time was the common theme in the highest-rated barriers, prompting consideration of addressing modifiable systems factors to increase this activity; cost-neutral measures such as tailoring nursing and physician schedules, modifying care coordination rounds, pre-scheduling patient rounding times, and geographic localization of patients and providers may attenuate this barrier. Further investigation into the impact of bedside interprofessional rounds on patient satisfaction and outcomes are necessary to validate widespread dissemination.

BEDSIDE ROUNDS IN THE ERA OF COMPETENCY-BASED EDUCATION: A WORKPLACE-BASED STRATEGY FOR FEEDBACK AND REFLECTIONJed Gonzalo^1,2; Brian S. Heist²; Briar Duffy^3,2; Liselotte Dyrbye⁴; Mark J. Fagan⁵; Gary S. Ferenchick⁶; Heather Harrell⁷; Paul Hemmer⁸; Walter N. Kernan⁹; Jennifer R. Kogan¹⁰; Colleen M. Rafferty¹; Raymond Y. Wong¹¹; Michael Elnicki². ¹Penn State College of Medicine, Hershey, PA; ²University of Pittsburgh School of Medicine, Pittsburgh, PA; ³University of Minnesota School of Medicine, Minneapolis, MN; ⁴Mayo Clinic College of Medicine, Rochester, MN; ⁵Alpert Medical School of Brown University, Providence, RI; ⁶College of Human Medicine, Michigan State University, East Lansing, MI; ⁷University of Florida College of Medicine, Gainesville, FL; ⁸Uniformed Services University of the Health Sciences, Bethesda, MD; ⁹Yale University School of Medicine, New Haven, CT; ¹⁰Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA; ¹¹Loma Linda University School of Medicine, Loma Linda, CA. (Tracking ID #1937769)

BACKGROUND: Feedback and reflection are two strategies integral to competency-based medical education and trainee development. These methods are best employed in authentic workplace-based settings, opportunities which are not well described in the literature. Bedside rounds have been highlighted as an ideal venue for observation, feedback, and reflection, and the literature suggests methods for implementing these strategies, but actual strategies used to deliver feedback and team-based reflection by current-day bedside teachers has not been described. We sought to understand the content and timing of feedback and team-based reflection provided by bedside teachers in the context of patient-centered bedside rounds.

METHODS: In 2010, a qualitative thematic analysis of transcripts from digitally recorded, one-on-one telephone interviews with experienced bedside teachers (n = 34) was performed. Co-investigators from 10 U.S. institutions were recruited and asked to identify participants who had: 1) served as inpatient attending physician at least 2weeks in the prior 2years, and 2) performed “bedside rounds” while inpatient attending, defined as: “The team of providers, including at least one resident and attending physician, presenting the patient’s history or reviewing one physical exam component, in addition to discussing the diagnosis/management at the bedside in the patient’s presence.” Interviews consisted of closed- and open-ended questions; questions related to feedback were designed from frameworks identified in the literature. For questions related to reflection, defined as “…analyzing an experience for the purposes of learning or improve practice,” no studies were identified, therefore an inductive approach was used. Two investigators analyzed transcripts independently, with regular adjudication sessions to modify the codebook, followed by theme extraction. The study was exempt from review by all Institutional Review Boards.

RESULTS: Participants averaged 14years of academic experience and 50% were associate/full professor. Positive feedback about physical exam skills or clinical decision-making occurred during bedside encounters, while positive or constructive team-based feedback occurred immediately following encounters; individualized constructive feedback occurred following rounding sessions in private locations. Team-based reflection occurred immediately following bedside encounters and related to commonplace “teachable moments” more so than less frequent, emotionally-charged events. Three content areas related to team-based reflection were identified: (1) patient characteristics/emotions, (2) trainee actions/emotions, and, (3) attending physician role modeling.

CONCLUSIONS: Bedside teachers use bedside rounds as a workplace-based opportunity to stimulate feedback and team-based reflection with medical trainees. Bedside rounds provide authenticity as bedside activities allow assessments of trainees’ performance and skills in the context of real-time patient encounters. As demonstrated by participants, reflection can be used to promote cognitive and humanistic growth of trainees. Although reflection has been identified in the literature, no prior work has identified a taxonomy of events triggering reflection. Amidst current-day pressures of inpatient medicine, these content and timing-related strategies for feedback and reflection could assist in faculty development efforts geared toward competency-based education.

BEHAVIORAL ECONOMICS-INFORMED INTERVENTIONS TO REDUCE INAPPROPRIATE ANTIBIOTIC PRESCRIBING: A PILOT CLUSTER RANDOMIZED TRIALStephen D. Persell¹; Jeffrey A. Linder²; Mark W. Friedberg^3,2; Daniella Meeker⁴; Elisha M. Friesema¹; Andrew J. Cooper¹; Craig R. Fox⁵; Noah J. Goldstein⁵; Jason N. Doctor⁶. ¹Northwestern University, Chicago, IL; ²Brigham and Women’s Hospital, Boston, MA; ³RAND, Boston, MA; ⁴RAND, Santa Monica, CA; ⁵UCLA, Los Angeles, CA; ⁶University of Southern California, Los Angeles, CA. (Tracking ID #1938251)

BACKGROUND: Physicians often use antibiotics for acute respiratory infections (ARIs) when they are not indicated. Interventions to reduce inappropriate prescribing have been only modestly successful.

METHODS: In preparation for a larger, multi-site trial, we performed a pilot cluster-randomized controlled clinical trial with physician as the unit of randomization and office visit as the unit of analysis to evaluate different behavioral economic and social psychological approaches to reduce inappropriate antibiotic prescribing. Physicians and a nurse practitioner at a single large urban academic internal medicine practice were enrolled and randomized within a 2 × 2 × 2 factorial experiment. Qualifying ARI visits triggered each intervention. The interventions were: 1) Accountable Justifications: when prescribing an antibiotic for an ARI, clinicians are prompted to record an explicit justification that appears in the patient electronic health record (EHR); 2) Suggested Alternatives: through computerized clinical decision support, clinicians prescribing an antibiotic for an ARI receive a list of non-antibiotic treatment choices prior to completing the antibiotic prescription; and 3) Peer Comparison: each provider’s rate of inappropriate antibiotic prescribing relative to top-performing peers is provided periodically by email. All providers completed an on-line educational module about ARI treatment at the start of the intervention period. The primary outcome was the rate of antibiotic prescribing for visits with an ARI diagnosis for which antibiotics are not indicated (e.g. acute bronchitis, unspecified upper respiratory infection). Secondary endpoints were prescribing rates for acute sinusitis or pharyngitis, and prescribing rates for all other respiratory infections or symptoms of respiratory infection. We excluded patients with major comorbidities (e.g., COPD, cirrhosis) or a concomitant diagnosis that suggested a non-respiratory-tract bacterial infection (e.g., cystitis). EHR-data was used to measure outcomes during the year prior to and during a 1-year intervention period. We used mixed models to assess intervention effects accounting for fixed effects for patient age and providers’ prior prescription rate and random provider effects.

RESULTS: 28 of 38 providers participated. Antibiotic prescribing fell in the intervention year for non-antibiotic-appropriate ARI diagnoses (from 24.7 to 4.6%), for sinusitis/pharyngitis (from 45.1 to 35.1%), and for all other respiratory diagnoses or symptoms of infection (43.5 to 31.5%; all p < 0.001). There were no significant differences in the primary outcome for any of the interventions. Suggested Alternatives led to a lower odds of antibiotic prescribing for sinusitis/pharyngitis, odd ratio (OR) 0.35, (95% confidence interval [CI] 0.22–0.58); and for all respiratory visits combined (OR 0.72 [95% CI 0.54–0.96]). Peer comparisons led to lower odds of antibiotic prescribing for all respiratory visits combined (OR 0.73 [95% CI 0.53–0.995]). There was no increase in the proportion of total ARI visits within the potentially antibiotic appropriate category or other ARI diagnosis/symptom category.

CONCLUSIONS: The conduct of this study had a large impact on antibiotic prescribing behavior. Given the moderately low baseline rate and the possible Hawthorne effect, the prescribing rate for the primary outcome may have reached a floor below which adding additional interventions could not reduce it further. Since this was a single site, this widespread decline in prescribing may have been due communication between intervention and control group providers that led to attitudinal or cultural changes more broadly within the practice. Evidence of gaming the system (e.g., changing diagnosis codes to make antibiotic prescribing appear more appropriate) was not observed. An ongoing, larger, practice-randomized trial will help elucidate the effects of these interventions on provider behavior.

BETTER INFORMED PATIENTS ARE LESS LIKELY TO CHOOSE PERCUTANEOUS CORONARY INTERVENTION FOR STABLE ANGINAMichael B. Rothberg¹; Senthil K. Sivalingam²; Reva Kleppel²; Bo Hu¹; Marc Schweiger². ¹Cleveland Clinic, Cleveland, OH; ²Baystate Medical Center, Springfield, MA. (Tracking ID #1937449)

BACKGROUND: Patients with chronic stable angina often believe that percutaneous coronary intervention (PCI) will prevent a myocardial infarction or extend life, even though their cardiologists do not believe this. The extent to which cardiologists contribute to this misperception is unknown. One possibility is that patients do not recieve complete information about the procedure. We reviewed consent conversations to assess the level of informed decision making and its association with choosing to have PCI.

METHODS: Using the Verilogue Point-of-Practice Database, which includes visits with >600 physicians in nine geographic regions throughout the U.S., we searched outpatient/non-acute visit transcripts audio-taped between August 2008 and August 2012 for mention of PCI, cardiac catheterization, angiogram or stent placement. We limited transcripts selected to those that included consent discussions with a cardiologist. Two investigators reviewed each transcript to identify the 7 elements of decision making required for a complex medical decision, using the framework developed by Braddock, et. al. The 7 elements include discussion of the 1) patient’s role in decision making, 2) nature of the decision 3) the alternatives 4) pros and cons of the alternatives 5) uncertainties associated with the decision 6) assessment of the patient’s understanding and 7) exploration of patient preference. We also assessed the cardiologists’ recommendation and the patient’s decision regarding catheterization and PCI (cath/PCI). Because these procedures are usually done at the same time, patients give consent for both procedures at once. We assessed the association between the decision to undergo cath/PCI and specific elements of informed decision making, angina severity, and physician recommendation. Finally, we assessed the association between completeness of informed decision making, defined as the number of elements of informed decision making present in the conversation, and the patient’s decision to undergo cath/PCI in a logistic regression model.

RESULTS: The final dataset included 27 cardiologists in conversations with 60 patients (median two patients per cardiologist, range 1 to 6). Cardiologists were all male, had been in practice for an average of 20.5years (range 8 to 34), and 2/3 belonged to a private office-based group practice. Mean patient age was 66years, 70% were male and 75% were white. Of the 60 discussions, 4 (7%) included all 7 elements of informed decision making; 28% met a more restricted definition of procedure, alternatives and risks. In just over half (52%) there was some discussion of the patient’s role in decision making. Discussion of the clinical issue was present in 97% of discussions, but discussion of alternatives (27%) and uncertainties (10%) were uncommon. Assessment of patient understanding and exploration of preferences each occurred in 65% and 73% cases, respectively. At the end of the visit, 44 (73%) patients chose to undergo PCI. Physicians made recommendations in 49 cases and patients followed the recommendations in 90% of cases. In univariate analysis, the proportion of patients choosing cath/PCI was lower when there was discussion of 1) uncertainty associated with the decision (17% vs. 80%, p < 0.01), 2) the patient’s role in decision making (61% vs. 86%, p = 0.03), and 3) discussion of the alternatives (31% vs. 89%, p = <0.01). The proportion choosing cath/PCI was lower, but not statistically so, when the physician assessed patient understanding (67% vs. 86%, p = 0.11) and explored patient preference (68% vs. 88%, p = 0.13). Anginal symptoms were present in 81% of patients, but limited activity in only 28% of patients. Presence of symptoms was not statistically associated with choosing cath/PCI (p = 0.44), but patients whose symptoms limited activity were more likely to choose cath/PCI (94% vs. 71%, p = 0.06). In multivariable analysis, better informed patients were more likely to choose not to undergo cath/PCI (OR = 2.7 for each additional element of informed decision making, 95% CI 1.2 to 6.2, p = 0.02).

CONCLUSIONS: In consent conversations between cardiologists and patients with stable angina, informed decision making is often incomplete. More complete discussions are associated with patients choosing not to undergo cath/PCI. Efforts to involve patients in shared decision making for cath/PCI might also reduce resource utilization.

BLOOD PRESSURE CONTROL IN DIABETIC PATIENTS AT FEDERALLY QUALIFIED HEALTH CENTERS: THE IMPACT OF OFFICE-GUIDELINES APPLIED TO PRACTICE (OFFICE-GAP) PROGRAMAde B. Olomu¹; Bikki Gautam¹; Bethany Buda¹; Wei-Wen Hsu³; Gurpreet Chahal¹; David Todem²; Alexis Therman¹; Esha R. Kumar¹; Margaret Holmes-Rovner⁴. ¹Michigan State University, East Lansing, MI; ²Michigan State University, East Lansing, MI; ³Michigan State University, East Lansing, MI; ⁴Michigan State University, East Lansing, MI. (Tracking ID #1932917)

BACKGROUND: The most important intervention for preventing death and adverse cardiovascular (CVD) outcomes is adequate control of blood pressure (BP) in diabetic (DM) patients. Unfortunately, diabetic patients with hypertension (HTN) are often not adequately controlled and HTN management is less intensive in patients with DM. HTN is particularly burdensome in low-income groups, where the prevalence of uncontrolled HTN is higher than the general population. Federally Qualified Health Centers (FQHCs) provide care for low income and medically underserved populations. Objectives: 1) to determine the rate and predictors of Blood Pressure (BP) control in patients with DM and HTN in FQHCs. 2) determine the impact of a CVD secondary prevention intervention- Office Guidelines Applied to Practice (Office-GAP) Program on BP control in FQHCs.

METHODS: The Office-GAP study is a two-center intervention and control study designed to improve cardiovascular care for minority and low-income populations in outpatient clinical settings. Intervention included provider training, patient education in a group visit, and use of Office-GAP tools/checklist and patient decision aids during office visits. After a group visit, patients followed up with physician visits using GAP tools. We performed chart abstraction of all patients with HTN, CVD and/or DM from September 2010 to Dec 2012 in 2 FQHCs. Hypertension was defined as Systolic BP > 140mmHg (>130mmHg in DM patients) and diastolic BP > 90mmHg (>80mmHg in DM patients). Multivariable logistic Regression was used to identify potential predictors of BP control.

RESULTS: Of 242 patients identified, 188 had DM, and 191 had HTN, 146 had HTN and Diabetes. One hundred patients were in the intervention (Office- GAP) arm and 142 in the control arm. At baseline, 32.79% (60/183) and 26.76% (38/142) of all patients with HTN without diabetes and HTN patients with DM had their BP controlled respectively. We found that diabetic patients enrolled in the Office GAP Intervention were more likely to have their BP controlled at 6months (OR = 2.70, p = 0.048) compared to control patients after adjusting for relevant clinical variables. In addition, the logistic regression model revealed that whites (OR 2.6, p = <0.001) and other races (Hispanics, Nepalese, Somali) (OR = 1.685, p = 0.031) were more likely to have their BP controlled compared to black patients. Furthermore, older age (OR 0.98, p = 0.014) and patients with higher Charlson Index (OR 0.85, p = 0.002) were found to be marginally less likely to have their BP controlled.

CONCLUSIONS: We found that the majority of patients attending FQHCs do not have their BP controlled. BP control among diabetics was substantially less frequent than in nondiabetic patients. Black patients were less likely to have their BP controlled compared to other races. The Office-GAP Intervention improved rates of BP control for diabetic patients in FQHCs. The Office-GAP Program could serve as a model for improving care for patients in outpatient clinical settings. Further studies are needed to determine the effectiveness and cost-effectiveness of this approach.

BUILDING COMMUNITY TRUST THROUGH QUALITY ASSURANCE OF MALARIA DIAGNOSIS AND MANAGEMENT AT A RURAL CLINIC IN UGANDASae-Rom Chae¹; Toni Biskup²; Sara Heinert²; Janet Lin^2,3. ¹University of Illinois at Chicago College of Medicine, Chicago, IL; ²University of Illinois at Chicago College of Medicine, Chicago, IL; ³University of Illinois at Chicago School of Public Health, Chicago, IL. (Tracking ID #1933920)

BACKGROUND: At a rural primary care clinic in Uganda, a patient survey indicated that the community members feel it provides a high, trustworthy quality of care. However, no quality assurance studies have been conducted to characterize the quality of care provided. Malaria remains a leading cause of morbidity and mortality in Uganda. Widespread efforts have been made by the public sector to build capacity in diagnosis and treatment of the disease. This study aims to evaluate the quality of care provided by the clinic by examining malaria diagnosis and treatment patterns and assessing the degree of adherence by the clinic staff to national malaria treatment guidelines. Study objectives: 1. Determine how malaria is being diagnosed and consequently managed among a sample of a Ugandan rural clinic patient population. 2. Assess diagnostic methods and appropriate classification of severity of disease. 3. Evaluate malaria treatment regimens as appropriate based on 2012 Ugandan treatment guidelines. 4. Consider the quality of malaria care provided and whether the clinic is providing trustworthy care to the community.

METHODS: A retrospective chart review of patients seen at a rural clinic in Uganda in a 12-month period from June 2012–June 2013 was conducted to determine the rates of appropriate diagnosis and treatment of malaria. All patients with blood smears done on a Monday, Wednesday, or Friday were included. De-identified data on demographics, presenting symptoms, given diagnoses, medications, blood smear results, and notes on management decisions was recorded in a database. Variables were then coded numerically for each record and a second individual re-coded 10% of the charts for verification. After excluding charts with missing data points, a final sample of 1001 patients was analyzed using SAS and Microsoft Excel.

RESULTS: Of the total 390 positive blood smears, only three patients (0.8%) were not started on malaria medication. Of patients documented to have a negative smear (572 or 57.1%), 83 (14.5%) were started on malaria treatment. 198 (50.8%) of the positive smears were among adult patients above the age of 17years, three of which showed hyperparasitemia. Of these, 1 was referred for higher level care, 1 was started on second-line quinine treatment, and the third was started on alternative first-line treatment. Five patients that were documented to have a diagnosis of malaria and a presenting symptom of seizures, qualifying them as severely ill, were not started on parenteral treatment or referred for higher-level care. Of the total 21 patients started on second-line quinine-based treatment, 11 (52.4%) had a documented and appropriate indication. Of all patients documented to have a positive smear and placed on appropriate malaria treatment, 7 (1.8%) were concurrently started on an antibiotic (i.e. doxycycline, clindamycin), documented as malaria treatment.

CONCLUSIONS: Overall, clinic staff demonstrate a heightened index of suspicion for malaria infection, appropriate for a highly endemic area. Nearly all patients with positive diagnostic tests were started on treatment. A small number were started on empiric treatment despite a negative test. Investigating the reasons for this would be worthwhile in order to prevent over-treatment. Most of the patients experiencing serious, complicated malarial disease, including those with hyperparasitemia, were not treated according to national guidelines with intravenous quinine. Further training may be needed in identifying and documenting cases of severe malarial illness and the appropriate management sequence. This may significantly impact mortality and morbidity. A knowledge assessment of the staff informed by detailed chart reviews and process mapping in the future may be helpful in delineating where intervention might be the most suitable. Additionally, second-line treatment is being over-prescribed or indications are not adequately documented. If the former is occurring, this may affect drug resistance patterns in an unfavorable manner. Lastly, antibiotic regimens are being prescribed to supplement malaria treatments which warrants additional investigation. Without appropriate indication, this practice may contribute to antimicrobial resistance and unnecessary medication side effects. The national guidelines do not discuss the indications for concurrent antibiotic treatment. In summary, improvement is needed in the areas of classifying severity of malarial illness and consequent management, indications for the various types of treatment, and possible bacterial co-infection with indication for additional treatment.

BUILDING EQUITY IMPROVEMENT INTO QUALITY IMPROVEMENT: REDUCING SOCIOECONOMIC DISPARITIES IN COLORECTAL CANCER SCREENING AS PART OF POPULATION HEALTH MANAGEMENTSeth A. Berkowitz¹; Sanja Percac-Lima¹; Jeffrey M. Ashburner¹; Yuchiao Chang¹; Adrian Zai¹; Wei He¹; Richard W. Grant²; Steven J. Atlas¹. ¹MGH, Boston, MA; ²Kaiser Permanente, Oakland, CA. (Tracking ID #1921347)

BACKGROUND: Improving colorectal cancer (CRC) screening rates for patients from socioeconomically disadvantaged backgrounds is a recognized public health priority. We tested the hypothesis that implementation of a system-wide screening intervention could increase overall screening rates while also decreasing screening disparities.

METHODS: In June 2011, we implemented the Technology for Optimizing Population Care (TopCare) intervention to improve CRC screening for all eligible patients in our 18-practice primary care network. The TopCare intervention system electronically identifies patients overdue for screening and contacts them by letter or telephone scheduler, with or without physician involvement. Patients identified by automated algorithm as high risk for non-completion are entered into a more intensive program of patient navigation. Using ≤high school (HS) diploma as our marker for socioeconomic disadvantage, we investigated whether baseline differences in CRC screening completion rates between ≤HS diploma vs. >HS diploma patients narrowed after implementation of TopCare. We used an interrupted time series analysis to compare monthly CRC screening completion rate disparities in the 2years prior to implementation (June 15, 2009–June 14, 2011) and 1year after implementation (June 15, 2011–June 14, 2012).

RESULTS: At the beginning of the study period, 71% of 52,011 eligible patients had completed screening, while by the end 74% of 54,572 eligible patients had completed screening (p < .001). Although CRC screening completion remained higher in >HS diploma vs. ≤HS diploma patients in June 2009 (74% vs. 66%, p < .001) and June 2012 (77% vs. 70%, p < .001), the difference in screening rates decreased from 8.6 to 7.3% over this 3-year period (Figure). In our interrupted time series analysis accounting for secular trends, the introduction of TopCare was associated with a significant decrease in the CRC screening disparity (p < .001). The decrease represents approximately 150 more ≤HS diploma patients with complete CRC screening each month than without the intervention, an improvement that would have taken approximately 27 additional months to occur had pre-intervention trends continued.

CONCLUSIONS: A multifaceted population management intervention sensitive to the needs of vulnerable patients narrowed disparities in CRC screening while also increasing overall screening rates. Embedding interventions for vulnerable patients within larger population-based systems represents an effective approach to increasing overall care quality while simultaneously decreasing disparities.

BUILDING THE INFRASTRUCTURE TO SUPPORT ACCOUNTABLE CARE PAYMENT MODELS: LESSONS FROM THE FIELDMaria Han¹; M. Susan Ridgely²; Cheryl Damberg². ¹UCLA, Los Angeles, CA; ²RAND, Santa Monica, CA. (Tracking ID #1940916)

BACKGROUND: The accountable care organization (ACO) is a novel payment and delivery system innovation aimed at improving value in health care. As hundreds of healthcare organizations have entered into ACO contracts, they have had to make significant structural and organizational changes, including expanding capabilities to deliver population-based care and forging integrated provider relationships. Despite the scale and complexity of the changes required in transitioning to the ACO model, there is little known or reported about how to optimally design an ACO. Given the substantial investment and risk associated with ACO formation, the future success of the ACO movement will require effectively capturing, disseminating, and replicating the lessons learned by early ACOs. To that end, the objective of this study is to delineate the major types of innovations implemented by provider organizations in their endeavors to become successful ACOs.

METHODS: The accountable care organization (ACO) is a novel payment and delivery system innovation aimed at improving value in health care. As hundreds of healthcare organizations have entered into ACO contracts, they have had to make significant structural and organizational changes, including expanding capabilities to deliver population-based care and forging integrated provider relationships. Despite the scale and complexity of the changes required in transitioning to the ACO model, there is little known or reported about how to optimally design an ACO. Given the substantial investment and risk associated with ACO formation, the future success of the ACO movement will require effectively capturing, disseminating, and replicating the lessons learned by early ACOs. To that end, the objective of this study is to delineate the major types of innovations implemented by provider organizations in their endeavors to become successful ACOs.

RESULTS: One-year following grant funding, eight of the 18 organizations had implemented interventions aimed at redesigning clinical care (the remaining grantees focused on building IT infrastructure). While organizations executed a diversity of interventions in care redesign, most of them focused on delivering care that was increasingly coordinated, team-based, and patient-centered. Almost all (7/8) of the grantees instituted programs to improve care coordination, particularly across patient transitions. Six of the grantees implemented team-based care approaches, including utilization of physician extenders, development of disease-oriented teams, and implementation of the patient-centered medical home model. Finally, all of the organizations emphasized patient-centered care through the promotion of patient education initiatives, group appointments, extended office hours, and social needs assessments.

CONCLUSIONS: Many health systems currently face the challenge of moving the ACO from a theoretical concept to an “on the ground” reality. While “nascent” ACOs have experimented with hundreds of unique approaches to ACO development, this study is one of few reports on their experiences and common themes in implementation. As such, the findings of this study may prove useful in understanding what infrastructure investments will create the right capabilities for delivering accountable care.

BUYING TIME: QUALITY IMPROVEMENT PROJECT TO DECREASE PATIENT CALLS FOR MEDICATION REFILLS IN A RESIDENT RUN CONTINUITY CLINICAmanda Lerner^1,2; Shveta Gandhi^1,2; Diane Resnick¹; John T. Connor¹; Rowena Garcia¹; Laurie Hudson-Bonner¹; Joann Sabados-Carolina¹; Andrea Palisi¹; Michael P. Carson^1,2. ¹Jersey Shore University Medical Center, Neptune, NJ; ²Rutgers-Robert Wood Johnson Medical School, Piscataway, NJ. (Tracking ID #1913297)

BACKGROUND: The resident run continuity clinic noted a large volume of calls/faxes for medication refill requests, and we estimated that each required a minimum of 15min of staff time to complete. In addition, running out of medication poses a potential risk to patient well-being. Objectives: utilize the Plan-Do-Check-Act quality improvement process to quantify and identify Refill Request patterns, develop an education plan, and assess the efficacy of the plan to reduce these requests/improve patient safety.

METHODS: Pre-Education Data: Nurses logged each Refill Request over 3months noting time since the last office visit, and method of request (patient walking in, patient/pharmacy call to the prescription line vs. main clinic number, pharmacy fax to one of two machines). An action plan was developed and disseminated to the practitioners. Four months later, a second assessment was completed (Post-Education Reassessment).

RESULTS: Pre-Education Data: 17.8 Refill Requests/day (765/43days). Patients typically return after 12–24weeks, but 61% of the requests were received within 8weeks of a patient’s last visit suggesting physicians were not providing enough refills. Twenty-five percent were due to patients missing visits, and 7% were to clarify/correct a prescription. Pharmacies generated eight messages per day, and sent 278 faxes. Action Plan: Caregivers were instructed to objectively confirm remaining refills (view pill bottle, or review photocopy of the prior prescription in the chart), and provide enough refills until the next visit plus one extra month (“Month Plus 1” policy). Patients were advised to bring all bottles to each visit, and pharmacies were asked to fax all Requests to a single machine located in our clinic. Reassessment: 3.3 requests/day (110/33days), an 81% decrease in the averge volume of daily requests; only 10% due to physician error and 64% due to patient issues. Pharmacy calls dropped by 90% to 33 (~1/day). Patient calls dropped from 3/day to 2/day. As we had 33 observations from the Post-Education phase, a matched analysis required us to use only 33 of the 43 Pre-Education observations. To avoid overestimating the effect, we first compared the 33 lowest Pre-Education data points to the Post-Education values yielding an average decrease of 6.2 requests/day [95% CI (4,8)]. The analysis using the 33 highest Pre-Education values yielded a decrease of 19 requests/day [95% CI (12,26)]. Estimating 15min per Request, this the daily staff workload decreased by at least 1.5, and up to 4.8h a day.

CONCLUSIONS: A simple caregiver and patient education program facilitated an 81% decrease in the volume of calls/faxes for medication refills in this resident-run continuity clinic. Patients often told resident physicians they had refills remaining, however not enough to get them to the next visit. Therefore, objectively determining the number of refills and the “Month Plus 1” policy are likely responsible for the majority of the decrease, and they improve patient care by decreasing the chance that a patient will be without their medication. On average, the 14.5 fewer requests/day decreased the time burden on an already limited clinical staff by 3.6h per day. The staff is satisfied, and a third survey confirmed patients missing visits as the primary reason for Refill Requests. Future efforts will focus on communicating with patients to improve attendance at scheduled visits.

CALLING “DR. GOOGLE”…: DOES TECHNOLOGY FILL THE GAP CREATED BY THE PRENATAL CARE VISIT STRUCTURE?Jennifer Kraschnewski¹; Cynthia H. Chuang¹; Erika Poole¹; Tamara Peyton¹; Ian Blubaugh¹; Alyssa Feher²; Madhu Reddy¹. ¹Penn State Hershey Medical Center, Hershey, PA; ²Cumberland/Perry Tapestry of Health, Carlisle, PA. (Tracking ID #1936600)

BACKGROUND: The structure of prenatal care has changed little over the past several decades despite the rapid evolution of technology. Little is known about how pregnant women engage with technologies (e.g. websites, smartphone apps) and the interface between these tools and medical care, especially for women of lower socioeconomic status. We sought to understand the role of technology in pregnancy through a qualitative study with women enrolled in the Women, Infants and Children (WIC) program.

METHODS: We recruited pregnant women ages 18 and older who owned a smartphone at a WIC clinic in central Pennsylvania. Four focus groups were conducted with a total of 25 women. The focus group guide included questions about the positive experiences and challenges women encountered during their current pregnancy, sources of information, and whether they used their smartphones or other technology for pregnancy-related information. Sessions were audiotaped and transcribed. Two members of the research team independently analyzed each transcript, using a thematic analysis approach. Themes related to the topics discussed were identified, for which there was full agreement.

RESULTS: Three major themes emerged. First, the prenatal visit structure was not patient-centered. Women commented that the first visit occurs too late, with too few visits early in pregnancy, when women had the most questions for their doctors: ”…They expected me to wait 13weeks until I had a conversation with my doctor.” “I know we can’t change the healthcare system… but on your first appointment, they say, ‘We’ll see you in 8weeks…’ and uh… That’s 2months! What am I going to do?” Even when women did see their doctor, the materials they received were viewed as unhelpful: “Today they gave me a whole bag of pamphlets and flyers and didn’t explain or go over them with me… and now I have to go home and try to go through them, while I have a kid running around… and when you’re a new mom, that’s overwhelming.” Second, women turned to technology to fill gaps, including using Google and smartphone applications (i.e. Babycenter): “I found myself using Dr. Google, because I couldn’t get an appointment until I was 14weeks. They just wouldn’t see me.” Turning to technology was viewed to be a generational approach: “It’s interesting cause I think that the generation before us counted a lot on our mom and our sisters to give advice. And now it’s technology…“ Lastly, women reported that technology, although frequently used, was not without limitations: ”… so sometimes you can Google something, like when I was having my round ligament pain for the first time… some things that came up were terrifying… you want to go to the ER right away! You definitely have to be careful and smart about your Google searches.”

CONCLUSIONS: The current prenatal care visit structure is not patient-centered in that it does not allow women to seek advice when they want it the most. A generational shift has occurred, resulting in pregnant women turning to the Internet to fill this gap, which requires significant skills to navigate for useful information. Until systems-based changes are made to the prenatal care visit structure, primary care providers may be called on to assist our patients early in pregnancy seek the information they want and to become better consumers of pregnancy-related information on the Internet.

CAN AN IN-HOSPITAL CARE TRANSITION INTERVENTION IMPROVE PATIENT EXPERIENCE IN THE SAFETY NET? A RANDOMIZED CONTROLLED TRIALBrian Chan¹; L. E. Goldman¹; Urmimala Sarkar¹; Michelle Schneidermann²; Jeffrey Critchfield²; Margot Kushel¹. ¹UCSF/SFGH, San Francisco, CA; ²UCSF/SFGH, San Francisco, CA. (Tracking ID #1937181)

BACKGROUND: Patient experience measures are important markers of quality, as evidenced by use of the Hospital Consumer Assessment of Healthcare Provider Scores (HCAHPS) and Care Transitions Measures (CTM-3) for hospital value-based purchasing reimbursement. Patient experience scores are lower in safety-net hospitals than non safety-net hospitals; this gap is widening. We assessed the effect of a nurse-led, language concordant, hospital-based care transition intervention on patient experience in older adults hospitalized at a safety-net hospital.

METHODS: We randomized, and stratified by language (English, Spanish or Chinese), 700 inpatients age 55 and older who were inpatients at an academic urban safety-net hospital. The intervention consisted of usual care and 1) inpatient visits by a language concordant nurse that provided individualized post-hospitalization education and 2) post-discharge phone calls by a nurse practitioner to reinforce the care plan and to address acute complaints. Language concordant research assistants, blinded to randomization group, measured HCAHPS nursing, physician, medication, and discharge information domains and CTM-3 scores at 30days after hospital discharge.

RESULTS: Of 685 participants who survived to 30days, 90% (n = 617) completed follow-up research interviews. Mean age was 66years; 24% were African American; 19% were White; 32% were Asian and 19% Latino. Over half (61%) of interviews were conducted in English, 24% in Chinese and 15% in Spanish. Over a third of participants noted limited English proficiency, and 51% met criteria for low health literacy. Study nurses spent an average of 157min with the intervention participants. There were no statistically significant differences in mean HCAHP domain scores and CTM-3 scores between the intervention and control groups (Table 1); no significant differences were seen between groups when results were stratified by language.

CONCLUSIONS: The failure of a language concordant nurse-led intervention to improve patient experience scores suggests that these measures are difficult to improve using hospital-based interventions. Thus, caution should be exercised when linking hospital-based reimbursement to improved HCAHPS and CTM scores. Further work is necessary to identify interventions that can improve patient experience and care transitions measures.

Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) Mean Domain and Care Transitions Measure (CTM-3) Scores assessed at 30days

* wilcoxan rank sum statistic use to generate p-value ** overall CTM-3 score out of 100 † individual CTM-3 question score out of a total of 4

CAN PARTICIPATION IN COMMUNITY ENGAGEMENT ACTIVITIES REDUCE DISTRUST IN MEDICAL RESEARCHERS AMONG DIVERSE COMMUNITIES: RESULTS FROM THE COMMUNITY ENGAGEMENT AND RESEARCH PROGRAM OF THE UCLA CTSINazleen Bharmal¹; Li-Jung Liang¹; Sitaram Vangala¹; Stefanie D. Vassar¹; Ibrahima Sankare¹; Loretta Jones²; Keith C. Norris¹; Martin F. Shapiro¹; Arleen Brown¹. ¹UCLA, Los Angeles, CA; ²Healthy African American Families II, Los Angeles, CA. (Tracking ID #1938942)

BACKGROUND: Racial/ethnic minorities are not adequately represented in medical research studies. One explanation, supported by scientific literature, is that distrust in research and the medical community may impede recruitment of racial/ethnic minorities into medical research studies. Among the goals of the UCLA Clinical and Translational Science Institute’s Community Engagement and Research Program (CERP) are to enhance the community’s trust in health research by promoting knowledge sharing between community and academia and strengthening both community and academic infrastructure for sustainable partnered research. To achieve these goals, the CERP has sponsored or co-sponsored a series of large symposia and smaller workshops on clinical research, public health topics, and community engagement. To assess levels of trust among community members who participate in CERP events, we surveyed attendees at every event held over a 14month period.

METHODS: Between June 2012 and August 2013, participants at CERP events were asked to respond to four survey items: 1) “If a medical researcher asked you to participate in a medical research study, you trust that he/she would provide you with a full explanation of the study,” 2) “A medical researcher would not ask you to participate in a research study if he/she thought it would harm you,” 3) “If you participate in a medical research study, you trust that the medical researcher will make sure that you fully understand the consent form before you sign it,” and 4) “If you, or people in your community, choose to participate in medical research, do you feel that the medical researcher will always try to protect you from unnecessary risk?” The written items were self-administered in English or Spanish at the end of each event. The response choices, distributed across a 4 point Likert scale, were dichotomized as “strongly agree/agree” vs. “disagree/strongly disagree/don’t know.” We developed a distrust index with a score ranging from 0 to 4 as our main outcome measure. Only participants who responded to at least three of the four questions were assigned a distrust index score. Data from events was aggregated by the quarter of the year during which the event occurred. We compared distrust scores across all quarters using linear mixed-effects regression model, adjusted for participant’s demographic characteristics and accounting for heterogeneity of variance across types of community event. We present the estimated difference between the distrust index scores using model contrast for 2012 Q2 vs. 2013 Q3.

RESULTS: Over 14 Months, we administered 531 surveys; 468 respondents answered at least three of the four questions (response rate 88%) at nine CTSI/CERP community conferences/workshops. The majority of participants were female (76%); 49% were ages 50years or older; 13% had a high school education or less; and 53% were African American, 20% Latino, 12% White, 11% Asian/Pacific Islander, and 4% other race/ethnicity. The mean distrust score over the timeframe of CTSI/CERP workshops was 1.22 (standard deviation 1.29). The adjusted estimated mean distrust index score (± standard error) decreased from 2.19 ± 0.30 in 2012 Q2 to 1.05 ± 0.26 in 2013 Q3 (p = 0.0035). The estimated change in the distrust index score for African Americans from 2012 Q2 (mean ± SE: 2.34 ± 0.32) to 2013 Q3 (0.82 ± 0.26) was significant (p = 0.0002), but did not reach statistical significance for Latinos (1.56 ± 0.77 vs. 0.69 ± 0.30, p = 0.30), Whites (2.54 ± 0.56 vs. 2.93 ± 0.54, p = 0.62), and Asians (0.90 ± 0.62 vs. 1.23 ± 0.37, p = 0.65).

CONCLUSIONS: The multivariable analyses suggest that CTSI/CERP efforts to increase community engagement in research have been associated with a decrease in distrust in health research, especially among African American participants. These findings are observational and represent ecologic associations; thus our results may be biased by secular trends. Additional qualitative work is needed to better understand reasons for change in levels of distrust over time in these diverse communities as a result of the CTSI/CERP.

CAN THYROID BREAK YOUR HEART? ROLE OF THYROID IN TAKOTSUBO CARDIOMYOPATHY: A SINGLE CENTER RETROSPECTIVE STUDYSourabh Aggarwal; Vishal Gupta. Western Michigan University School of Medicine, Kalamazoo, MI. (Tracking ID #1940273)

BACKGROUND: Takotsubo cardiomyopathy (TC) is a transient systolic dysfunction of the apical and/or mid segments of the left ventricle that mimics myocardial infarction (MI) but in the absence of obstructive coronary artery disease. The exact etiology of TC is unclear. Few isolated case reports have described association of TC with hyperthyroid state. The exact association of thyroid status and TC has never been studied so far.

METHODS: This was a single center retrospective study. All the patients diagnosed with TC between January 2006 and December 2012 at our hospital were identified retrospectively. Mayo’s revised criteria for TC was used to confirm the diagnosis. Baseline paramenters including demographic profile, cardiac markers, thyroid studies, angiographic and echocardiographic findings were extracted and analysed using SPSSv19.0

RESULTS: Seventy eight patient were identified as diagnosed with TC with 72 females (92.3%) and six males (7.6%) with mean age of 66.33 ± 13.37years (range 23–91). Stressful event precipitating TC was identified only in 19 patients (24.36%) at the time of admission. Twenty seven patients (34.61%) had history of hypothyroidism and 25 patients (31.05%) were on levothyroxine replacement. Thyroid profile was available for 44 patients at the time of admission and five patients (11.36%) were found to be in a hyperthyroid state based on low TSH and/or high freeT4. During a mean follow up of 2.8 ± 1.5years, five patients (5.13%) had episodes of recurrence. Mean length of stay (mLOS) was 4.4 ± 3.9days (Median 3days, Range 1–23days). Both TSH level and ejection fraction had significant independent negative correlation with mLOS (p < 0.05).

CONCLUSIONS: Thyroid status/thyroid hormone replacement may have a significant role in the pathogenesis of TC. In this first retrospective study describing the association of TC with thyroid status, precipitating stress was identifiable in only 24% patients whereas 35% of the patients had a history of hypothyroidism with majority on levothyroxine replacement. Additionally, TSH level had an independent negative correlation with mLOS.

CARE FRAGMENTATION AND SURVIVAL FOR PATIENTS WITH STAGE III COLON CANCERTanvir Hussain^1,2; Hsien-Yen Chang²; Christine M. Veenstra³; Craig E. Pollack^1,2. ¹Johns Hopkins University School of Medicine, Baltimore, MD; ²Johns Hopkins Bloomberg School of Public Health, Baltimore, MD; ³University of Michigan Health Systems, Ann Arbor, MI. (Tracking ID #1939909)

BACKGROUND: Cancer care can be complex and fragmented, spanning many settings and providers. Though the Institute of Medicine cites fragmentation as a priority for improving cancer care, little is known about aspects of fragmentation that affect cancer outcomes. We examine two features which may have different implications for improving care delivery: receiving surgical and medical oncologic care from providers affiliated with different hospitals; and receiving care from a surgeon-oncologist pair who infrequently shares patients with one another. We focus on stage III colon cancer patients as guidelines recommend both surgery and chemotherapy.

METHODS: Patients with stage III colon cancer diagnosed between 2000 and 2005 were identified from SEER-Medicare data. Patients were assigned to their operative surgeon and the medical oncologist who billed for the plurality of their visits in the year following diagnosis. Surgeons and oncologists were linked to the hospital where they billed most for inpatient care. Patients were classified as experiencing “hospital fragmentation” if their surgeon and oncologist were assigned to different hospitals (versus the same hospital). We determined the number of patients each surgeon-oncologist pair shared; patients were classified as having “high patient-sharing” physicians if their oncologist and surgeon shared many patients (top quartile of the shared patients distribution versus lower three). Patient-sharing has been validated as a measure of collaboration and information exchange between physicians. Our primary outcome was all cause mortality (censor date 12/31/2007). Secondary outcomes included timely receipt of chemotherapy (9months of diagnosis) and cost of care [total claims using MEDPAR (Part A), NCH (Part B), and Outsaf files)] in the 12months following diagnosis. We used Cox proportional hazard and regression models, adjusted for patient demographics, socioeconomic status, comorbidities, SEER site; surgeon’s yearly procedure volume, oncologist patient panel size; and hospital characteristics (volume; NCI, for-profit, and academic status). Generalized estimating equations and robust standard errors were used to account for hierarchical data and clustering.

RESULTS: Our sample included 7443 patients. Median survival was 3.04years. One-third (N = 2471) received care from surgeons and oncologists associated with different hospitals. No difference in morality was associated with hospital fragmentation (adjusted HR = 1.00, 95% CI: 0.93–1.07). We observed an increased risk of death among patients whose surgeons and medical oncologists shared few patients (lower three quartiles of shared patients) compared to those whose doctors were in the top quartile (HR = 1.15, 1.06–1.25). No statistical interaction between the two predictors was noted in the final model (Wald’s test, p = 0.324). Neither hospital fragmentation nor patient-sharing predicted timely receipt of chemotherapy or 12month costs of care.

CONCLUSIONS: Receiving care from physicians associated with the same hospital did not improve survival for stage III colon cancer patients, whereas receiving care from physicians sharing many patients with another did. These results suggest that efforts to improve care fragmentation need to examine the informal relationships between physicians that may be reflected by patient-sharing. Further, cancer survival, quality, and costs may not be improved by delivery redesign which addresses fragmentation solely by consolidating options for care to one institution.

CAUSES OF SHORT-TERM HOSPITAL MEDICINE READMISSIONS AT AN ACADEMIC MEDICAL CENTERKenton Smitherman; Brian Clay; Leslie Martin-Armstrong. UC San Diego Medical Center, San Diego, CA. (Tracking ID #1927644)

BACKGROUND: Health care continues to undergo significant changes. With the recent creation of the Hospital Readmission Reduction Program (HRRP) by the Centers for Medicare and Medicaid Services (CMS), financial penalties are being assessed to hospitals with high 30-day readmission rates for certain patient groups. Investigating reasons for readmissions could be important in identifying possible patterns and areas for improvement. The objective of this study was to identify the primary contributing cause for each of 341 short-term readmissions (within 7days of discharge) to ten hospital medicine services at an academic medical center over the ten-month period from 4/1/2011 through 1/31/2012.

METHODS: The UC San Diego Health System includes two hospitals with a combined capacity of 552 beds. Use of a full-spectrum inpatient electronic medical record (EMR) began on February 27, 2011 (a system-wide outpatient EMR was already in place). During the period of April 2011 through January 2012, there were 5,789 admissions to the hospital medicine services. From this total, 341 were readmissions within 7days. These readmission cases were retrospectively reviewed by two members of the Division of Hospital Medicine to determine the primary factor causing each patient’s readmission. Patient, health system, and physician factors were considered. Patient factors included: (a) exacerbation of chronic illness, (b) new/separate diagnosis, (c) noncompliance with medication or other therapy, (d) noncompliance with outpatient follow up, and (e) patient not responding to reasonable medication or other therapeutic plan. Health system factors were: (a) lack of timely available outpatient follow up appointment, and (b) inability to provide medication or other therapy in the outpatient setting or at a lower level of care. Physician factors included: (a) discharge with inappropriate medications or therapy, (b) incorrect or incomplete diagnosis (excluding cases when further workup in the outpatient setting was planned), and (c) complication of a procedure or therapy instituted during the preceding hospitalization.

RESULTS: Case review showed that most (95%) short-term readmissions resulted from patient factors: 42% of cases were due to exacerbations of chronic illness, 29% were due to new or separate diagnoses, none were shown to be due to noncompliance with follow up (these readmissions were within 7days of discharge), and 10% were due to patients not responding to reasonable medication or other therapy. The health system factor of inability to provide medication or other therapy in the outpatient setting or at a lower level of care was deemed the primary cause of readmission in 1% of cases. Importantly, physician factors were the primary cause of readmission in 13, or 4%, of the cases: 2 readmissions were due to discharge with inappropriate medications or therapy, 5 readmissions were due to incorrect or incomplete diagnosis, and 6 readmissions were the result of a complication of a procedure or therapy instituted during the preceding hospitalization. Additionally, in this 341 case cohort, 16% of the readmission cases involved patients with chronic liver disease, and 8% of the cases involved patients who were homeless.

CONCLUSIONS: Most short-term readmissions to our hospital medicine services are due to patient factors. Health system and physician factors were identified as causes of readmissions in only 5% of cases. Penalizing hospitals with higher readmission rates may not be a proper tool for decreasing hospital readmissions within 7days of discharge.

CHALLENGES TO INCREASING ACCESS AND CONTINUITY IN A LARGE ACADEMIC MEDICAL CENTER IMPLEMENTING PCMHClaire Robinson¹; Molly Harrod¹; Jane Forman¹; Ann-Marie Rosland^1,2; Adam Tremblay^1,2; Eve A. Kerr^1,2. ¹VA Ann Arbor Health Care System, Ann Arbor, MI; ²University of Michigan, Ann Arbor, MI. (Tracking ID #1937578)

BACKGROUND: A core goal of the patient-centered medical home (PCMH) model is to increase timely access to primary care while maintaining patient-provider continuity. While definitions of access vary, it is often operationalized as receiving a same-day appointment with the patient’s usual provider. For example, the Department of Veterans Affairs (VA) has implemented a national performance measure examining receipt of same-day appointments with the patient’s usual primary care provider (PCP). Large academic primary care clinics face unique challenges in providing prompt access to a patient’s usual provider, particularly because many of these clinics are staffed by providers and residents who are in the clinic only a few hours each week (‘part-time providers’). We examined factors affecting efforts to increase prompt access and provider continuity in a large VA academic medical center that was in the process of implementing PCMH. Among the medical center’s 78 PCPs, 59 were available in clinic half-time or less. We have previously shown that part-time providers at this medical center were less likely to meet the national access measure.

METHODS: We conducted 20 semi-structured interviews with primary care staff (physicians, residents, registered nurses, licensed practical nurses, and clerks), roughly one and a half years into PCMH implementation. We also observed more than 25h of coaching sessions designed to help newly formed primary care teams redesign their delivery processes to improve their access and continuity measures. We coded interview transcripts and coaching observation field notes with descriptive, non-hierarchical, grounded codes that emerged from the data and developed findings via group consensus.

RESULTS: We conducted 20 semi-structured interviews with primary care staff (physicians, residents, registered nurses, licensed practical nurses, and clerks), roughly one and a half years into PCMH implementation. We also observed more than 25h of coaching sessions designed to help newly formed primary care teams redesign their delivery processes to improve their access and continuity measures. We coded interview transcripts and coaching observation field notes with descriptive, non-hierarchical, grounded codes that emerged from the data and developed findings via group consensus.

CONCLUSIONS: The large number of part-time primary care providers and residents combined with the large volume of patients made same-day access with the patient’s usual PCP challenging. Increased flexibility to schedule non face-to-face visits during part-time PCP’s non-clinical duties could help improve access. However, strategies that promote a team approach to access are likely the only long-term solution if we are to improve access while continuing to support providers who choose or need to work part-time. More information is needed on patient needs and preferences for prompt access and continuity in academic environments, so these new strategies can be tailored to patient-centered goals.

CHARACTERISTICS OF PERSISTENT EMERGENCY DEPARTMENT HIGH UTILIZERS IN A LARGE PRIMARY CARE PRACTICE-BASED RESEARCH NETWORKAndrew S. Hwang¹; Shan W. Liu²; Carine Yelibi³; Jeffrey M. Ashburner³; Wei He³; Brandon Auerbach³; Steven J. Atlas³; Clemens S. Hong³. ¹Tufts University School of Medicine, Boston, MA; ²Massachusetts General Hospital, Boston, MA; ³Massachusetts General Hospital, Boston, MA. (Tracking ID #1934654)

BACKGROUND: A small percentage of high-utilizing patients account for a disproportionate share of emergency department (ED) visits. Understanding the characteristics of persistent high-utilizers may allow for more effective design of interventions to reduce ED utilization and costs.

METHODS: We performed a cohort analysis to study ED high-utilizers among 161,707 patients receiving care in a large primary care practice-based research network (PBRN) between 2005 and 2007. We identified all PBRN patients who visited the ED ≥4 times in 2007, and divided them into 2 groups: 1) patients who also made ≥4 ED visits between 2005 and 2006 (persistent users) and 2) patients who made <4 ED visits between 2005 and 2006 (non-persistent users). We compared the following characteristics in the two groups: demographic characteristics, chronic conditions, behavioral health problems, count of electronic medication prescriptions (e-prescriptions), and past healthcare utilization (between 2005 and 2007). We used Poisson regression models to assess the association between persistent ED high-utilization and subsequent ED visits and hospitalizations over 5years (2008–2012) using a “change-in-estimate” strategy, where we included potential covariates (marriage status, insurance, language, chronic conditions, behavioral health problems, count of e-prescriptions, and past healthcare utilization) in the models if they altered the incidence rate ratio (RR) by >5%.

RESULTS: Among 697 patients with ≥4 ED visits in 2007, 276 (40%) were persistent users. Compared to non-persistent users, persistent users were significantly (p < 0.01 for all) more likely to be unmarried (80% vs. 70%) and English speaking (93% vs. 87%), have non-commercial insurance (81% vs. 68%), or have a diagnosis of coronary artery disease (26% vs. 17%), chronic obstructive pulmonary disease (49% vs. 38%), congestive heart failure (50% vs. 40%), chronic pain (18% vs. 10%), depression (16% vs. 8%), bipolar disorder (18% vs. 11%), psychosis (44% vs. 27%), or alcohol- (34% vs. 17%) or drug-related problems (37% vs. 21%). Persistent users also received a higher total number of e-prescriptions over 3years (53 vs. 40) and were more likely to be e-prescribed narcotics (88% vs. 77%), benzodiazepines (83% vs. 69%), anti-psychotics (55% vs. 36%), and medications for depression/anxiety (64% vs. 50%) or smoking cessation (24% vs. 16%). Compared to non-persistent users, persistent users had higher rates of missed primary care appointments (19% vs. 14%), and higher unadjusted rates of acute care utilization (2,685 vs. 1,134 ED visits/1,000 patients/year and 1,822 vs. 934 admissions/1,000 patients/year) in the prior 3years (2005–2007). Compared to non-persistent users, persistent users also had higher unadjusted rates of acute care utilization (2,030 vs. 1,483 ED visits/1,000 patients/year and 892 vs. 440 admissions/1,000 patients/year) over the subsequent 5years. Adjusting for prior ED visits and hospitalizations, compared to non-persistent users, persistent users had a higher rate of subsequent hospitalization (adjusted RR 1.33 [1.08–1.65]), but not a higher rate of subsequent ED visits (adjusted RR 1.31 [0.93–1.84]). Including all covariates in the models changed the overall effects by less than 4% for both outcomes without changing effect direction or significance.

CONCLUSIONS: Compared to non-persistent ED high-utilizing patients, persistent ED high-utilizing patients are more medically and psychosocially complex and have a higher risk of future hospitalization. Persistent ED high-utilizers represent an important target for intervention to improve quality and reduce costs. Such efforts will likely require intensive, multidisciplinary, care management approaches.

CHARACTERIZATION AND LACK OF PLANNING FOR ADVANCED LIFE EVENTS AMONG OLDER ADULTSLee Lindquist¹; Priya Sunkara¹; Vanessa Ramirez-Zohfeld¹; Chris Forcucci²; Megan Huisingh-Scheetz³; Kenzie A. Cameron¹. ¹Northwestern University Feinberg School of Medicine, Chicago, IL; ²Aging and In-Home Services, Fort Wayne, IN; ³University of Chicago, Chicago, IL. (Tracking ID #1939204)

BACKGROUND: Despite older adults’ wish to remain independent in their own homes, critical health and life events occur that impede their ability to do so. A lack of information exists on what these advanced life events (ALE) entail and the advanced planning older adults perceive is necessary.

METHODS: We conducted focus groups with older adults living in community and independent living facilities in rural, suburban, and urban locales. Subjects were asked to discuss their future living plans and events/experiences that might impact their ability to remain in their own home. Focus group questions also elicited if subjects had planned for, discussed with others (e.g. spouse, offspring), or acted on these plans if one of these advanced life events occurred to them. Content analysis and constant comparative analysis was used to analyze the results.

RESULTS: Across all sites, 68 older adults participated in the focus groups with a mean age of 73.9years (SD = 6.53); 50 (73.5%) were female. Analysis revealed older adults perceived several major advanced life events impacted a senior’s ability to remain in their own home: 1) Hospitalization or Serious Illness, 2) Functional Loss 3) Cognitive loss, 4) Spousal death. Subjects voiced experiences with each of these events in multiple forms (e.g. self, spouse, friend, relative). However, very few perceived these events would occur to them, “I’ve decided I’m not going to get Alzheimer’s!” Planning for these events were almost non-existent as many older adult subjects felt that they were too sick or too healthy to look forward. “I’m on borrowed time. So I’m not going to worry about where I’m going to go for my old age because I don’t think I’m going to have one.”,“The new 70 is the old 58!” A number of seniors also felt that offspring would handle it for them although, many subjects stated that they had not discussed current or future needs with their offspring. While many subjects voiced that they had completed end-of-life documentation, including powers of attorney, living wills, obituary/funeral requests, there was a lack of advanced planning for life events that have a high likelihood of occurring to seniors.

CONCLUSIONS: Helping seniors to create plans for Advanced Life Events (e.g. hospitalization, cognitive loss, functional loss) is important as it provides seniors a voice in their future while including key stakeholders (e.g. offspring, spouse). It is well known that these ALEs have a high propensity for occurring with age. Ongoing research will create and test a planning tool that will assist older adults in being prepared for their needs during these events. By planning for—instead of reacting to—ALEs, older adults may be able to remain in their own homes safer and longer.

CHILDREN: A BARRIER TO MATERNAL HEALTH CARE UTILIZATION?Christina Pham²; Johannes R. Kratz¹; Danny McCormick². ¹Massachusetts General Hospital, Cambridge, MA; ²Harvard Medical School/Cambridge Health Alliance, Cambridge, MA. (Tracking ID #1936573)

BACKGROUND: Prior studies have demonstrated that women with children are at higher risk for obesity, metabolic syndrome, cardiovascular disease, depression, and all-cause mortality than women without children. No prior studies have examined the impact of having a young child on a mother’s utilization of primary, specialty and preventive care that could mediate these disparities.

METHODS: We analyzed cumulative data from 5years (2008–2012) of the National Health Interview Survey (NHIS). The NHIS is a representative nationwide sample of households conducted annually by the Centers for Disease Control and Prevention (CDC) though face-to-face in home interviews. The study sample included women age 60 or younger who answered the Family, Person, and Sample Adult surveys (57,746). The NHIS collects data on the number of children under the age of 18 at home, age, income, race/ethnicity, education level, employment status, and health insurance. The primary outcome of this study was having a visit to a primary care physician in the preceding year. Secondary outcomes included visits to a medical specialist or Ob/Gyn in the past year as well as basic health preventative care such as blood pressure checks, cholesterol checks, tetanus shots, mammograms, and Pap smears in the past year. The primary predictor variable was having one or more children under the age of 18 in the home. A total of 27,918 (48.4%) patients had children under the age of 18; 29,828 (51.6%) did not. We used multivariable logistic regression models to test associations between predictor and outcome variables while controlling for age, income, race/ethnicity, education level, employment status, and health insurance.

RESULTS: The mean age of the patients was 36.3 and 42.3 for patients with and without children. When adjusted for age, income, race, educational level, employment status, and health insurance, patients who had children under the age of 18 had a 12% lower odds of having a PCP visit in the preceding year (OR = 0.88 [95% CI, 0.85–0.92]; P < 0.001) and a 27% lower odds of having seen a medical specialist (OR = 0.73 [95% CI, 0.70–0.76]; P < 0.001) in the past year. Patients with children under the age of 18 also had a lower odds of receiving basic health screening such as blood pressure checks (OR = 0.90 [95% CI, 0.83–0.97]; P = 0.009), cholesterol checks (OR = 0.91 [95% CI, 0.86–0.96]; P = 0.003), tetanus shots (OR = 0.94 [95% CI, 0.90–0.97]; P = 0.001), and mammograms (OR = 0.85 [95% CI, 0.79–0.92]; P < 0.001). In contrast, women with children had a higher odds of visiting an Ob/Gyn (OR = 1.22 [95% CI, 1.18–1.27]; P < 0.001) and of having received a Pap smear (OR = 1.29 [95% CI, 1.22–1.37]; P < 0.001) in the past year. Decreased health care utilization was not a function of cost of care: patients with children had a lower odds of reporting needing medical care but not receiving it because of cost (OR = 0.72 [95% CI, 0.68–0.77]; P < 0.001).

CONCLUSIONS: Women with children younger than 18years old are less likely to engage in their own primary care and receive important preventive care than women without young children. Current health care delivery system models may not be optimally structured to meet the needs of women with children.

Odds of PCP visit within the previous yearOdds of PCP visit within the previous year

*Baseline comparison group

CHOOSING WISELY: DO PHYSICIANS AND PATIENTS AGREE ON WHAT CONSTITUTES HIGH-VALUE MEDICAL CARE?Ana Sofia Warner¹; Neel Shah^1,3; Abraham Nick Morse¹; Eliyahu Lehmann²; Rie Maurer²; Zoe Moyer²; Lisa Lehmann^1,2. ¹Harvard Medical School, Boston, MA; ²Brigham and Women’s Hospital, Boston, MA; ³Beth Israel Deaconess Medical Center, Boston, MA. (Tracking ID #1938226)

BACKGROUND: Healthcare delivery systems focus on the concept of value delivery as a way of optimizing patient experience and outcomes while containing costs. Value, patient health outcomes per dollar spent, is a concept that providers, patients and policymakers embrace. Although the Choosing Wisely campaign targets physicians and patients to promote high-value care, there is little data on whether there is concordance between physicians’ and patients’ interpretation of value, particularly for care that has been identified as wasteful, such an unindicated imaging and antibiotic use.

METHODS: To assess physicians’ and patients’ perceptions of the value of care we conducted a cross-sectional survey of 201 primary care patients and 150 primary care physicians at three academic medical centers. The first vignette described a patient with symptoms of a tension headache who is concerned about a brain tumor, but initially denied imaging. The second vignette described a patient with 3days of symptoms consistent with a viral infection who requests and is denied antibiotics. Respondents rated the care provided in each vignette on a 5-point Likert scale (see Table 1). Descriptive statistics were used to describe frequencies. Chi-square and Fisher’s exact tests were used to compare proportions between patients and physicians. Univariate regression analysis was used to assess predictors of patients’ rating care as high value. Multivariable regression analysis is ongoing.

RESULTS: The response rate was 63% (201/319) for patients and 53% (150/283) for physicians. Among patients, 37% were men, 63% were white. Physicians more often reported that imaging for a tension headache represented low-value care than patients (77% vs. 38%, p < 0.001), and more often reported that denying antibiotics for a viral infection represented high-value care (95%, vs. 66%, p < 0.001). Patients improved their rating of care after they were told about the potential harms of radiation, and after receiving information about potential side effects and reasons for not providing antibiotics or imaging (15% increase in positive ratings for both scenarios). Patients who were concerned about receiving too few tests, treatments and medications were less likely to give a high rating to Part 1 of the headache vignette (OR 0.42 (0.23–0.78) p < 0.006), while patients who rated their own health higher were more likely to rate it favorably (p < 0.01). Patients who had seen their physicians more often were more likely to rate Part 1 of the URI vignette highly (p < 0.03). More educated and white patients were more likely to rate Part 1 of both vignettes highly (p < 0.01).

CONCLUSIONS: There is significant disagreement between physicians’ and patients’ perceptions of the value of care. Patients are more likely to favorably rate care withholding antibiotics than care withholding imaging. Patients are likely to improve their rating of care if they are informed about potential side effects, reasons for not providing antibiotics or imaging, or national guidelines consistent with their doctor’s care. Addressing these topics in clinical settings will be critical to our ability to contain healthcare costs and deliver high value care.

Table 1: Patients and Physicians Ratings of Clinical Vignettes as ‘Very Good’ (4) or ‘Excellent' (5)

CLASSIFICATION OF THEMES HIGHLIGHTS BOTH VARIABILITY AND CONSISTENCY IN PHYSICIAN APPROACHES TO PROFESSIONAL DILEMMASElizabeth C. Bernabeo¹; Shiphra Ginsburg²; Eric Holmboe¹. ¹American Board of Internal Medicine, Philadelphia, PA; ²University of Toronto, Toronto, ON, Canada. (Tracking ID #1940171)

BACKGROUND: Professionalism is grounded in a dynamic system comprised of elements at the individual, social, and societal level. In this view, a physician’s professional behavior is dependent on their personal characteristics, yet also influenced by situational and contextual phenomena arising during learning and practice. Identifying and reflecting on situational context may be difficult for physicians, particularly for physicians in real time professional dilemmas or for whom certain responses have become a personal style or habit.

METHODS: As part of an initiative to develop a self-directed approach to assessment of professionalism, we asked a random sample of general internists and specialists in outpatient and inpatient settings (N = 366) to complete an anonymous web-based survey addressing physician approaches to typical professional challenges. The survey prompted physicians to rate the likelihood of their responding in stated ways, and to rate the importance of a set of contextual factors in their decision making. Physicians also responded to open ended questions prompting alternative responses, additional factors that guided those responses, and what kind of new learning or impact the activity had on themselves personally.

RESULTS: The overall response rate was 132/366 (36.1%). Qualitative analysis of the open ended comments revealed several themes around which physicians approach professional dilemmas. Physicians were guided by issues of patient welfare (e.g. “I tend to come down on the side of what is best for the patient”), how much they like the patient (e.g. “Decisions are influenced by the level of the relationship with the patient”), policy, ethics, or legal issues (e.g. “I am worried about litigation”), and principles, values or rules (e.g. “I follow my own rules)”. We further identified several personal styles or habits of physicians in responding to professional dilemmas: lenient (e.g. “I am too eager to please”), avoidant (e.g. “I have a great concern about the hassle factor”), and inconsistent (e.g. “I am not consistent with certain patients or behaviors”).

CONCLUSIONS: Physicians were able to reflect on personal approaches and factors listed on the survey, as well as those that were not. The classification of themes that guide physician decisions in professional dilemmas is important, and suggests that despite variability in responses, physicians think about professionalism in systematic and complex ways. Helping physicians explore and identify with their personal reactions to professional dilemmas in a deliberate way can help promote mindfulness and renew interest and commitment to professionalism in practice. Self awareness and reflection are infrequent in medical practice, yet may be powerful agents of change for physicians.

CLASSIFYING PATIENT STATEMENTS ABOUT UNCONTROLLED PAIN IN PRIMARY CAREStephen G. Henry¹; Richard L. Kravitz¹; Robert A. Bell². ¹University of California Davis, Sacramento, CA; ²University of California Davis, Davis, CA. (Tracking ID #1938853)

BACKGROUND: Physicians frequently cite patient requests for opioids as one reason that visits involving patients on opioids are often frustrating. Past research in this area has focused on “difficult” visits or patients in specialty pain clinics. Little is known about how patients on opioids actually discuss uncontrolled pain during routine primary care visits.

METHODS: Data were derived by coding transcripts of routine primary care visits that were recorded as part of a clinical trial that compared two interventions designed to facilitate patient discussion of depression symptoms. First, we read transcripts to identify patients on chronic opioids (i.e. taking at least 1 dose/day for >90days). Second, we iteratively reviewed transcripts to develop and apply a coding system to identify strategies patients used to express poorly controlled pain. Our final coding system was adapted from Street’s system for coding patient participation and comprised four types of patient statements: concerns (statements expressing negative emotions about pain), questions, requests for action (e.g. referral), and assertive statements.

RESULTS: We identified 26 eligible transcripts that included patients on opioids for chronic pain. Patient concerns were the most common statement type (n = 84) followed by questions (n = 32), requests for action (n = 22) and assertive statements (n = 16). Approximately 28% (n = 43) of statements related to pain and suffering in general, 20% (n = 31) pertained to pain-related functional impairment, 29% (n = 44) were about non-opioid pain treatments, 19% (n = 29) related to opioids, and 5% (n = 7) concerned the underlying pain diagnosis. One patient requested a change from methadone to oxycodone after the physician suggested this change. All other requests for opioid prescriptions involved refills of patients’ existing opioid prescriptions. No direct requests for higher opioid doses were observed.

CONCLUSIONS: Patient were more likely to mention uncontrolled pain indirectly (i.e., through statements of concern and/or questions) rather than through direct requests for action or assertive statements. Statements about opioids comprised a minority of patient statements about uncontrolled pain. Contrary to conventional wisdom, direct patient requests for higher opioid doses appear uncommon in routine primary care visits involving patients on opioids for chronic pain.

CLINICAL CHARACTERISTICS OF PATIENTS ADMITTED FOR ACUTE EXACERBATION OF COPD THAT PREDICT SHORTER LENGTH OF STAYMatthew Cerasale¹; Khalil Mroue²; Juan Cesar Fernandez Castillo¹; Jennifer Swiderek³; David Paje¹. ¹Henry Ford Hospital, Detroit, MI; ²Wayne State University School of Medicine, Detroit, MI; ³Henry Ford Hospital, Detroit, MI. (Tracking ID #1940337)

BACKGROUND: Chronic Obstructive Pulmonary Disease (COPD) exacerbation is a common clinical condition that may lead to frequent hospitalization. However, identifying which patients would be appropriate for a short stay in an observation unit rather than an inpatient admission has not been well established. The aim of this study was to determine which patient characteristics and clinical findings are associated with hospitalizations of <48h, and thus can be used to guide referral to the observation unit.

METHODS: The study is based on a retrospective review of the medical records of patients who were primarily hospitalized for acute exacerbation of COPD to an 805-bed teaching hospital. Patients were excluded if they were admitted directly to the intensive care unit. Demographic data, baseline history and clinical findings in the emergency department and upon arrival to a hospital unit were abstracted. The occurrence of all-cause death, readmission and revisit to the emergency department within 30days were also recorded. The clinical characteristics of patients who stayed <48h were compared to those who stayed longer using independent T-tests for continuous variables and Chi-squared test for categorical variables.

RESULTS: The medical records of 181 patients who were primarily hospitalized for acute exacerbation of COPD between 2009 and 2013 were reviewed. There were 68 (38%) patients who stayed <48h (mean length of stay [LOS], 1.63 ± 0.62days) and 113 (62%) patients who stayed longer (mean LOS, 3.82 ± 1.99days). Both groups were similar in terms of age, sex, and racial distribution. They were also similar with regards to smoking status, prior use of inhaled steroids and anticholinergics, intubation within the past year, and hospital admission in the past year. Compared to patients who stayed longer, patients who stayed <48h had faster initial respiratory rate in the emergency department (23.82 ± 6.05/min vs. 21.42 ± 4.00/min, p = 0.002) and they had higher initial hemoglobin (13.26 ± 1.44g/dL vs. 12.53 ± 1.89g/dL, p = 0.007). The rest of the initial clinical findings in the emergency department and upon admission were similar in both groups. The rates of 30-day outcomes were also comparable: revisit to the emergency department within 30days (22% vs. 15%, p = 0.3197), inpatient readmission (18% vs. 14%, p = 0.6801) and death from all causes (0% vs. 2%, p = 0.7083).

CONCLUSIONS: A significant proportion of patients who are admitted for acute exacerbations of COPD stay less than 48h. There are no clinically significant differences in baseline characteristics and in initial clinical findings between patients who stay longer and those who stay less than 48h. Clinical outcomes were also similar in both groups.

CLINICAL EXCELLENCE IN HOSPITAL MEDICINE: INSIGHT GLEANED THROUGH OBSERVATIONSusrutha Kotwal; Eric Howell; Haruka Torok; Regina Landis; Scott Wright. Johns Hopkins Bayview Medical Center, Baltimore, MD. (Tracking ID #1925957)

BACKGROUND: As of July 2010, there were more than 30,000 hospitalists in the United States. The quality of care rendered by hospitalists is not uniform. Although patient satisfaction metrics like Press Ganey attempt to measure service excellence, no empiric research has been performed to characterize or define clinical excellence in hospital medicine.

METHODS: The Chiefs of Hospital Medicine at five different hospitals were emailed and asked to identify the “clinically excellent” hospitalists within their groups. The named hospitalists were invited to participate in the study; those agreeable would be shadowed during routine clinical care of patients. A data extraction sheet was developed and pilot tested to focus the observer’s attention on elements believed to be associated with clinical excellence. Detailed field notes, both quantitative and qualitative, were collected during the observation.

RESULTS: A total of 26 hospitalists were shadowed as they took care of patients. The average age of the physicians was 38years and their average experience in hospital medicine was 6years. Each hospitalist was observed for an average of 5h and they were observed with a mean of seven patients (total number of patient encounters observed = 197). On average, the physicians spent 11min with each patient; patients were examined for a mean of 100s. In terms of balance of dialogue, these respected providers spoke on average 71% of the time during the discourse. The frequency of certain select desirable behaviors observed were as follows: starting encounters with an “open ended” question (76%); integrating non-medical conversation to connect with patient (30%); displaying empathy during the encounters (27%). Each encounter was given a global rating and 14% were judged to be “clinically excellent”. Several reasons that encounters were judged not to be “clinically excellent” were the lack of non-medical conversation used to connect with patients (70% of the encounters), doctor neglected to wash hands prior to entering room (24% of the encounters), doctor did not uncover/disrobe the body part being examined (39%), and failing to ask permission prior to examining the patient (36%).

CONCLUSIONS: This study represents a first step in trying to characterize clinical excellence in Hospital Medicine, using both qualitative and quantitative data. Because hospitalists spend only a small amount of time with patients, relative to the time spent in indirect patient care, it is imperative that we make the most of this caring time and strive for clinical excellence consistently during every encounter.

CLINICAL IMPACT OF MEDICAL ASSISTANT HEALTH COACHING IN LOW-INCOME PATIENTS WITH UNCONTROLLED DIABETES, HYPERTENSION, AND HYPERLIPIDEMIA: A RANDOMIZED CONTROLLED TRIALDavid Thom; Rachel Willard-Grace; Danielle Hessler; Denise DeVore; Camille Prado; Thomas Bodenheimer; Ellen Chen. University of California, San Francisco, San Francisco, CA. (Tracking ID #1925738)

BACKGROUND: Medical assistants, part of one of the fastest growing allied health professions, are more likely than clinicians to be culturally and linguistically concordant with patients. Medical assistants trained in health coaching could provide an important resource as members of the primary care team. Previous studies of medical assistant health coaching programs found positive trends in clinical outcomes such as hemoglobin A1c but were not designed as randomized trials, lacked power to find statistically significant differences, and focused on a single condition such as diabetes or hypertension. We conducted a randomized controlled trial to test the effectiveness of clinic-based medical assistant health coaching versus usual care to improve clinical outcomes among low-income patients with uncontrolled type 2 diabetes, hypertension, and/or hyperlipidemia over 12months.

METHODS: The Health Coaching in Primary Care enrolled low-income patients with uncontrolled diabetes (hemoglobin A1C (HbA1C) > 8.0%), hyperlipidemia (LDL > 130mg/dL or >100mg/dL if diabetic), and/or hypertension (SBP > 140mmHg) from two safety-net clinics. Patients were randomized to receive usual care or health coaching from a medical assistant who completed health coach training. The primary outcome was a composite measure of being at goal (defined asHbA1c < 8.0%; LDL < 130mg/DL or <100mg/dL if diabetic; and SBP < 140mmHg) for at least one of the conditions uncontrolled at enrollment. Secondary outcomes were meeting all possible goals and meeting each goal separately.

RESULTS: Of 654 patients determined to be eligible, 441 (67%) were enrolled; approximately three-quarters from one clinic. Clinical outcome measures at 12months were available for 94% (132/144) of patients enrolled with uncontrolled diabetes, 95% (156/165) of patients enrolled with uncontrolled hypertension, and 83% (169/203) for patients enrolled with uncontrolled hyperlipidemia. There were no significant differences in participant characteristics by study arm (Table 1). Participants in the health coaching arm were more likely than participants in the usual care arm to achieve both the primary composite measure of one of their clinical goals and the secondary composite measure of reaching all of their clinical goals (Table 2). Almost twice the number of participants in the health coaching arm achieved the goal of HbA1c ≤ 8.0%. At the largest study site, participants in the health coaching arm were more likely to achieve the specific LDL cholesterol goal compared to usual care (41.8% vs. 25.4%, difference 16.4%, CI 1%–32%). There was no significant difference in the proportion of patients who met the treatment goal for hypertension.

CONCLUSIONS: Medical assistants serving as in-clinic health coaches can improve glycemic and cardiovascular health outcomes over usual care for low-income patients with uncontrolled diabetes, hypertension and/or hyperlipidemia

Table 1. Participant characteristics (% or mean (sd))

* For the 328 participants born outside the United States.

Table 2. Percent of participants reaching primary and secondary outcomes at 12months by study arm

* Achieved target for control of one or more of specific conditions for which enrolled. † Achieved target for control of all conditions for which enrolled.

CMS PAYMENT REFORM AND THE INCIDENCE OF HOSPITAL-ACQUIRED PULMONARY EMBOLISM OR DEEP VEIN THROMBOSISRisha Gidwani^1,2; Jay Bhattacharya³. ¹Veterans Health Administration, Menlo Park, CA; ²Stanford University, Stanford, CA; ³Stanford University, Stanford, CA. (Tracking ID #1938121)

BACKGROUND: The Centers for Medicare & Medicaid Services (CMS) has historically provided larger reimbursement to hospitals for inpatient stays in which a patient complication developed post-admission (a hospital-acquired condition). This manner of reimbursement financially rewarded hospitals that may have provided poor-quality care to their patients. In October 2008, CMS stopped reimbursing hospitals for the marginal cost of treating certain preventable hospital-acquired conditions. We evaluated whether CMS’s refusal to pay for hospital-acquired pulmonary embolism (PE) or deep vein thrombosis (DVT) resulted in a lower incidence of these conditions.

METHODS: This analysis employed difference-in-difference regression modeling. Difference-in-difference modeling is a quasi-experimental approach used when it is infeasible to conduct a randomized controlled trial, and when there is one group that is exposed to the intervention, another group that is unexposed to the intervention, and a single start date to the intervention. Comparing pre-post data in the exposed group to pre-post data in the unexposed group serves to remove biases that may be due to systematic differences across the unexposed and exposed groups and well as biases that may be due to larger trends in the environment. In this analysis, the intervention is CMS payment reform, the exposed group is Medicare patients receiving hip or knee surgery, the unexposed group is non-Medicare patients receiving hip or knee surgery, and the outcome of interest is the incidence proportion of PE or DVT. We limited the cohort to Medicare patients aged 65–69 and non-Medicare patients aged 60–64 to minimize the impact of age on developing PE or DVT. A hierarchical regression model was used to account for correlation between discharges within a single hospital. The model was populated using 2007–2009 data from the Nationwide Inpatient Sample, an all-payer database of inpatient discharges in the United States.

RESULTS: There were 136,634 encounters for hip or knee replacement surgery from 2007 to 2009 in patients aged 60–69. At baseline, PE/DVT occurred in 0.81% of all hip or knee replacement surgeries for Medicare patients aged 65–69. After adjusting for race, sex, hospital region, teaching hospital status, elective admission, rural hospital status, and median houseful income quartile for patient zip code, the CMS policy change was associated with a 32% reduction, or a 0.26 percentage point reduction, in the incidence proportion of hip or knee encounters with PE or DVT. The incidence of PE or DVT in these orthopedic encounters increased from the pre-intervention to the post-intervention period for non-Medicare patients aged 60–64, while it declined in that same period for Medicare patients aged 65–69. Results were robust to changes in model specification.

CONCLUSIONS: As the U.S. healthcare system increasingly moves from fee-for-service to fee-for-value reimbursement, it becomes important to evaluate whether fee-for-value-based reimbursement has a desired effect. Our administrative-data based analysis of CMS hospital-payment reform indicates payment modification had the desired effect of reducing hospital-acquired PE or DVT amongst patients with hip or knee replacement surgeries. After controlling for other variables, the policy change was independently associated with a 32% reduction in the incidence of hospital-acquired PE or DVT. While regression models indicate the 0.26 percentage point reduction was statistically significant, it remains the purview of policy makers and clinicians to decide whether a reduction in the incidence of these conditions from 0.81% to 0.55% is clinically significant. To our knowledge, this represents the first analysis of the effect of CMS payment reform on the incidence of hospital-acquired pulmonary embolism or deep vein thrombosis.

COGNITIVE REFLECTION AND ANTIBIOTIC PRESCRIBING FOR ACUTE RESPIRATORY INFECTIONSDwan Pineros¹; Jason N. Doctor⁶; Mark W. Friedberg⁵; Daniella Meeker⁴; Yelena Kleyner³; Jeffrey A. Linder^1,2. ¹Brigham and Women’s Hospital, Boston, MA; ²Harvard Medical School, Boston, MA; ³Partners HealthCare System, Boston, MA; ⁴RAND, Santa Monica, CA; ⁵RAND, Boston, MA; ⁶University of Southern California, Los Angeles, CA. (Tracking ID #1938790)

BACKGROUND: Clinicians frequently prescribe inappropriate antibiotics for acute respiratory infections (ARIs). One contributor to inappropriate antibiotic prescribing may be a tendency toward quick, automatic thinking by clinicians.

METHODS: The Cognitive Reflection Test (CRT) has three questions with intuitive but incorrect answers that respondents tend to reach if they do not consider the question carefully (i.e., incorrect answers that quick, automatic thinking produces). The CRT is scored from 0 to 3, representing the number of correct answers. We administered the CRT to 139 primary care clinicians in June 2013. From billing data and the electronic health record between June 2012 and June 2013, we calculated clinician-level antibiotic prescribing rates for ARIs in 3 categories: all ARIs, antibiotic-appropriate ARIs (e.g., pneumonia, streptococcal pharyngitis, or sinusitis), and non-antibiotic-appropriate ARIs (e.g., colds or acute bronchitis). We used ANOVA to compare CRT scores with antibiotic prescribing rates.

RESULTS: Thirty-three clinicians (24%) scored 0 points on the CRT; 27 (19%) scored 1; 45 (32%) scored 2; and 34 (24%) scored 3. Antibiotic prescribing rates were 28% for all ARIs, 50% for antibiotic-appropriate ARIs, and 17% for non-antibiotic-appropriate ARIs. For all ARIs, the antibiotic prescribing rates for CRT scores of 0, 1, 2, and 3 were 33%, 25%, 26%, and 29%, respectively (p = 0.23). For antibiotic-appropriate ARIs, the antibiotic prescribing rates for CRT scores of 0, 1, 2, and 3 were 51%, 50%, 50%, and 50%, respectively (p = 0.99). For non-antibiotic-appropriate diagnoses, the antibiotic prescribing rates for CRT scores of 0, 1, 2, and 3 were 23%, 14%, 14%, and 17% (p = 0.01).

CONCLUSIONS: Clinicians who exhibit less reflective thinking may have higher antibiotic prescribing rates for non-antibiotic-appropriate diagnoses, compared to those who are more likely to think reflectively. It is possible that differences in clinicians’ cognitive reflection could contribute to other variations in care.

COLORECTAL CANCER SCREENING RATES 5 YEARS AFTER SWITCHING TO A HIGH-DEDUCTIBLE HEALTH PLAN, OVERALL AND AMONG LOWER-INCOME MEMBERSJames F. Wharam¹; Bruce E. Landon²; Robert LeCates¹; Fang Zhang¹; Steve Soumerai¹; Dennis Ross-Degnan¹. ¹Harvard Medical School, Boston, MA; ²Harvard Medical School, Boston, MA. (Tracking ID #1940079)

BACKGROUND: The ACA is expected to cause a “seismic shift” in high-deductible health plan (HDHP) enrollment. Colorectal cancer (CRC) screening improves mortality but the increasing prevalence of HDHPs could reduce screening rates, especially among the poor. To protect against this, the ACA requires that health insurers cover CRC screening with no out-of-pocket costs. We studied HDHPs at a large national insurer that similarly has provided generous coverage of CRC screening tests over the last 8years. No published studies have examined long-term CRC screening patterns after transition to HDHPs or long-term effects on the poor.

METHODS: We studied administrative claims of 28,877 HDHP members age 50–64 insured through a large national insurer between 2004 and 2011. We analyzed screening rates for 1year before and up to 5years after their employers mandated a switch from traditional plans to HDHPs, compared with rates among 57,748 contemporaneous propensity score matched controls whose employers chose to remain in traditional plans. Our measure was screening with any of FOBT, flexible sigmoidoscopy, double-contrast barium enema, or colonoscopy. We considered members who received these tests screened for the next 1, 5, 5, and 10years, respectively. We subtracted monthly screening rates of the control group from those of the HDHP group to generate a before-after differenced trend, then used Joinpoint analysis to determine whether there was a statistically significant inflection point after the HDHP transition. We stratified plots and analyses by geocoded poverty status and fit a linear trend line to the baseline differenced trend to generate a predicted 5-year screening rate. We fit a second-order polynomial trend line to the follow-up rates to estimate an observed 5-year screening rate.

RESULTS: CRC screening rates in the overall HDHP group relative to the control group trended down beginning in the fourth month of the first follow-up year, an estimated −0.176% per month trend change (p < 0.0001). Both lower- and higher-income HDHP members followed a similar pattern (Figures 1 and 2). HDHP members from lower-income neighborhoods experienced an estimated 2.9% absolute decrease in CRC screening rates after 5years (Figure 1), while members from higher-income neighborhoods experienced an estimated 4.3% absolute reduction (Figure 2).

CONCLUSIONS: HDHPs were associated with reduced CRC screening over 5years, but the reduction was not greater among members from lower-income neighborhoods. Fully covering CRC screening under HDHPs might protect lower-income members to some degree from substantial reductions in screening. However, health plans and policymakers should consider strategies to broadly promote CRC screening among HDHP members.

Figure 1. Difference in monthly colorectal cancer screening status between higher-income high-deductible and traditional plan members, showing counterfactual trend predicted by the baseline trend and the fitted follow-up trend.

Figure 2. Difference in monthly colorectal cancer screening status between lower-income high-deductible and traditional plan members, showing counterfactual trend predicted by the baseline trend and the fitted follow-up trend.

COMMERCIAL WEIGHT LOSS PROGRAMS - WHICH ONES WORK? A SYSTEMATIC REVIEWRuchi Srivastava¹; Kimberly Gudzune¹; Zoobia Chaudhry¹; Ambereen K. Mehta¹; David Jacobs². ¹Johns Hopkins, Baltimore, MD; ²University of Maryland, Baltimore, MD. (Tracking ID #1935829)

BACKGROUND: Commercial weight loss programs are popular among overweight adults, yet a 2005 review found little evidence to support their effectiveness. New studies have since tested these programs. We aimed to determine the weight loss benefits among popular commercial weight loss programs.

METHODS: We selected five programs based on expert recommendations and Internet popularity. We searched MEDLINE from 1/2002 to 6/2013. We included randomized controlled trials (RCTs) among adults that were ≥12weeks and compared a commercial program to usual care (UC) or lifestyle counseling. We also included trials from the prior review that met these criteria. Paired investigators screened results to assess eligibility, then abstracted data on study design, population characteristics, and weight. We synthesized data qualitatively by program.

RESULTS: Overall, we included 27 RCTs. Weight Watchers (WW): As compared to UC, WW groups had 2.2–10% greater mean percent weight loss at all time points (8 RCTs). There were inconsistent effects when comparing WW to counseling (2 RCTs). Atkins: As compared to counseling, Atkins groups had 0.8–6.2% greater mean percent weight loss at 3–6months (6 RCTs); however, there were no significant between group differences at 12months or beyond. Jenny Craig (JC): One RCT compared JC to UC, where JC resulted in 7.5% and 6.4% greater mean percent weight loss at 6 and 12months, respectively. Slim-Fast (SF): As compared to UC, SF groups had 6.3–9.7% greater mean percent weight loss at 3–6months (4 RCTs). There were inconsistent effects when comparing SF to counseling (3 RCTs). Nutrisystem: One RCT compared Nutrisystem to UC, where Nutrisystem resulted in 6.8% greater mean percent weight loss at 6months.

CONCLUSIONS: While commercial weight loss programs can help patients lose weight, these programs may not be superior to counseling. Clinicians may consider these programs when clinic-based options are unavailable.

COMMUNICATION FROM URGENT CARE CENTERS TO EMERGENCY DEPARTMENTSRebekah Gardner^1,2; Esther Choo¹; Stefan Gravenstein^4,2; Rosa Baier^2,3. ¹Alpert Medical School of Brown University, Providence, RI; ²Healthcentric Advisors, Providence, RI; ³Brown University School of Public Health, Providence, RI; ⁴Case Western Reserve University, Cleveland, OH. (Tracking ID #1942530)

BACKGROUND: Care transitions occur when patients move from one healthcare setting or provider to another (e.g., when a hospital patient is discharged). Although successful transitions require timely, accurate and sufficient communication of clinical information, such communication is inconsistent and there are few established guidelines outside of the hospital setting. Consistent, high-quality communication is particularly important during and after visits to urgent care centers (UCCs). However, our review of the medical literature revealed no prior investigation on the quality and consistency of communication from UCCs to emergency departments (EDs). Our aim was to assess the quality, consistency and effectiveness of healthcare communication between UCCs and EDs during care transitions.

METHODS: We administered a short anonymous survey to a convenience sample of emergency medicine physicians in Rhode Island. We contacted department chairs at the state’s 13 acute-care hospitals and asked them to distribute the electronic survey to their emergency medicine physicians (MDs or DOs in active practice). The survey asked practitioners to characterize the frequency and completeness of the information they receive from UCCs. We used descriptive statistics to characterize survey respondents, their perceptions about UCC to ED information transfer and their preferences for future communication. We also conducted a qualitative analysis of open-ended questions.

RESULTS: Eleven (84.6%) ED chairs agreed to distribute the survey. Approximately half (48.7%) of the 199 emergency medicine physicians at these 11 hospitals responded. Most were attending physicians (62.1%) and had ≥10years of experience in the ED (55.7%). Overall, respondents’ perceptions of the consistency, relevancy and usefulness of the information they receive from UCCs varies widely. For example, despite expressing strong preferences to receive this type of information, respondents report receiving reason for referral 72.0% of cases and a copy of the chart in 60.6%of cases. Receipt of contact information (52.3%) and phone calls to the ED (42.0%) are lower. Conversely, respondents report rarely communicating ED findings back to UCCs (16.7%). Comments reveal frustration with poor information transfer, particularly when the UCC clinicians are unavailable, and a desire to standardize or formalize communication expectations.

CONCLUSIONS: These data highlight a very real disconnect between the communication and information that emergency medicine physicians believe is necessary for providing care and the information that they receive from UCCs. Additionally, the ED providers’ nearly universal lack of communication back to the urgent care center following treatment of the admitted patient suggests the absence of a closed communication loop between these care settings. Our findings speak to the importance of establishing communication standards and of including UCCs in ongoing initiatives to improve patient care transitions and experiences.

COMMUNITY COLLEGE PATHWAYS TO MEDICAL SCHOOL: EXPLORING THE EXPERIENCES OF LATINO STUDENTSEfrain Talamantes¹; Carol Mangione^2,3; Karla Gonzalez³; Gery Ryan⁴; Alejandro Jimenez⁵; Fabio R. Gonzalez⁶; Seira Greenwood⁷; Gerardo Moreno^2,8. ¹Department of Veteran Affairs and the University of California, Los Angeles, Los Angeles, CA; ²University of California, Los Angeles, Los Angeles, CA; ³University of California, UCLA, Los Angeles, CA; ⁴RAND Health, Santa Monica, CA; ⁵University of California, San Francisco, San Francisco, CA; ⁶San Jose City College, San Jose, CA; ⁷University of California, Los Angeles, Los Angeles, CA; ⁸University of California, Los Angeles, Los Angeles, CA. (Tracking ID #1940189)

BACKGROUND: The community college is an essential pathway to medical school for Latino and low-income students, and contributes to a diverse physician workforce. There has been an increase in community college enrollment for the last several decades. The objective of this study was to explore the facilitators and barriers Latino pre-medical students face when they attend a community college.

METHODS: Participants included 48 Latino pre-medical students who were enrolled at or previously attended a California community college, and 20 stakeholders who directly advise and support these students. We conducted five focus groups (4 of undergraduate students who were currently enrolled or previously attended a community college, and 1 of medical school students who attended a community college) and 20 semi-structured interviews with counselors advisors, and teachers based at the community college, and medical school admissions staff and faculty from several California medical schools. Focus groups and interviews were digitally audio-taped, transcribed, and reviewed independently by two coders. Standard qualitative content-analysis methods were used to identify major themes and a third investigator adjudicated disagreements.

RESULTS: Results revealed the following major facilitator themes 1) pre-medical enrichment programs 2) pre-medical specific advising 3) Latino physician mentors and 4) family support for pursuing a pre-medical education. Major barrier themes included: 1) a perception that college was not affordable 2) poor high school academic preparation 3) lack of family support and responsibilities that compete with time to study and 4) limited counseling or advising. The financial burden of attending a community college required many students to take low-paying jobs to pay for tuition, books, housing, and transportation, while taking rigorous science coursework. Enrichment programs such as Math Engineering Science Achievement (MESA) play an important role in providing community college students with academic tutorials and advising, but are not readily accessible to students because of capacity constraints and stringent financial verification requirements. Stakeholders also indicate the need for a more comprehensive and sustainable pre-medical enrichment programs that include financial support and pre-medical counseling for low-income students, many of whom were the first in their family to attend college.

CONCLUSIONS: This is the first qualitative study that explores the role the community college as a pathway to medical school for Latino pre-medical students. We identified a number of potentially mutable barriers for premedical students who attend community college. Enrichment programs and policies that decrease the financial burden of pursuing higher education potentially can mitigate many of these barriers. The significant growth in the number of Latinos at the community college pursuing premedical education represents an important part of the pipeline to increase the diversity of the physician workforce.

COMMUNITY COLLEGE PATHWAYS: IMPROVING THE U.S. PHYSICIAN WORKFORCE PIPELINEEfrain Talamantes¹; Carol Mangione^2,3; Karla Gonzalez²; Alejandro Jimenez⁴; Fabio R. Gonzalez⁵; Gerardo Moreno^2,6. ¹Department of Veteran Affairs and the University of California, Los Angeles, Los Angeles, CA; ²University of California, Los Angeles, Los Angeles, CA; ³University of California, UCLA, Los Angeles, CA; ⁴University of California, San Francisco, San Francisco, CA; ⁵San Jose City College, San Jose, CA; ⁶University of California, Los Angeles, Los Angeles, CA. (Tracking ID #1939846)

BACKGROUND: Diversity in the physician workforce increases healthcare access for underserved populations, improves culturally and linguistically appropriate care, and better prepares all physicians to serve the needs of a diverse population. How we can achieve that diversity, however, is not well understood. One answer may lie in where students attain the undergraduate training they need before entering medical school. There are no studies that clarify how or if the undergraduate pathway contributes to a diverse physician workforce that is interested in serving underserved communities. Furthermore, the Community College (CC) as a pathway to medical school has not been explored at all.

METHODS: We performed a cross-sectional analyses of the 2012 Association of American Medical Colleges matriculant and applicant files, and the Matriculating School Questionnaire to assess bivariate associations between student characteristics and participation in a CC pathway. We also estimate the association between participation in a CC pathway and acceptance to medical school, intention to practice in underserved areas or work primarily with minority populations using multivariate logistic regression to adjust for confounders.

RESULTS: One third of applicants (N = 40,491) and matriculants (N = 17,518) to U.S. allopathic medical schools used a CC. A higher proportion of Latino matriculants used the CC pathways (34% vs. 27%, P < 0.001), compared to Whites. Matriculants who attended a CC after graduating from high school and before transferring to a 4-year university were more likely to have parents without a college education (34% vs. 13.4%, P < 0.001), had lower odds of acceptance into medical school (adjusted odds ratio [AOR] = 0.68, 95% CI 0.61 to 0.75, P < 0.05) and higher odds (AOR = 1.26, 95% CI 1.04 to 1.53, P < 0.05) of intentions to practice in underserved communities, compared to medical students who never attended a CC, after adjusting for covariates.

CONCLUSIONS: The CC pathways are essential for underrepresented minorities and students who have intentions to practice in underserved communities. While many recruitment and outreach efforts are strictly targeted at 4-year universities, there may be ample opportunities to support students pursuing careers in medicine that use the community college pathways and nurture their desire to ultimately practice medicine in an underserved community.

COMMUNITY HEALTH CENTER PATIENTS’ ATTITUDES AND PREFERENCES REGARDING RECEIVING PREVENTIVE SERVICE OUTREACH MESSAGESTiffany Brown; Shira N. Goldman; Ji Young Lee; Muriel Jean-Jacques; Shreya Shah; Stephen D. Persell. Northwestern University, Feinberg School of Medicine, Chicago, IL. (Tracking ID #1938151)

BACKGROUND: As community health centers (CHC) work to improve their quality of care for clinical preventive services, one possible strategy is outreach to patients who are due for certain services. Multiple studies have evaluated outreach for immunizations and cancer screenings among vulnerable populations. This proactive outreach may be particularly beneficial for CHC patients who may be less likely to have regular preventive service visits and for whom preventive services may be a lower priority if limited clinical encounter time is focused on acute and chronic health issues. However, little is known about CHC patients’ attitudes towards this type of outreach and how they prefer to be contacted with these messages. Our objective was to understand CHC patients’ attitudes and preferences regarding receiving clinical preventive services outreach messages from their CHC. We report interim results from a survey of CHC patients.

METHODS: As part of a larger project evaluating the effectiveness of an outreach intervention to promote the prevention of cardiovascular disease (CVD), we conducted telephone surveys with patients within three CHC systems; two in Chicago, IL and one in Flagstaff, AZ. Patients were eligible if they had been randomized to intervention arm of a randomized controlled trial of mailed and telephone outreach to individuals at increased risk for the development of CVD. Eligibility criteria were: men ≥35 and women ≥45years old, a 10-year risk of coronary death or myocardial infarction (based on Framingham Risk Score) of at least 10%, and a visit to CHC within 6months prior to randomization. The survey was performed 1year after the intervention was delivered. The survey assessed patient attitudes towards receiving messages about three preventive services: (1) influenza vaccination (2) cancer screenings (3) and primary prevention of CVD. In addition, participants were asked how they preferred to be contacted with these messages. Participants responded to Likert items ranging from 1 (strongly disagree) to 5 (strongly agree). Demographic items were also included.

RESULTS: Among eligible subjects with the recruitment protocol for the survey completed, n = 60 completed interviews (56% response rate). The majority of participants were male (91.7%), 48.3% were black and 38.3% were white, 68.9% had a high school degree or higher. Participants expressed agreement with statement that it was a good idea for CHC to let them know when they were due for flu shot (mean 4.3, standard deviation [SD] 1.3), due for cancer screenings (mean 4.7, SD 0.8), and if there were things they could due to lower their risk of developing CVD (mean 4.7, SD 0.7). Patients expressed agreement with wanting to be called by health center staff member (mean 4.1, SD 1.4), and receiving information by mail (mean 4.3, SD 1.3), however they expressed disagreement when asked about receiving a text message (mean 2.5, SD 1.6) or an email message (mean 2.2, SD 1.6). Participants reported strong agreement with preference to come in to discuss recommended services with their provider in person (mean 4.4, SD 1.0).

CONCLUSIONS: CHC patients in our sample report positive attitudes towards receiving outreach messages about their preventive service needs from their clinic. Participants prefer to receive these messages by telephone call or mail as opposed to text message or email. These results may have limited generalizability given specific eligibility criteria and our small, primarily male sample. These data will help inform future outreach efforts to CHC patients to promote important clinical preventive services.

COMPARATIVE EFFECTIVENESS OF A PRACTICE-BASED TRIAL OF BLOOD PRESSURE CONTROL IN BLACKS: IS LESS MORE?Antoinette Schoenthaler¹; Jeanne Teresi²; Leanne Luerassi¹; Stephanie Silver²; Jian Kong²; Taiye Odedosu¹; Gbenga Ogedegbe¹. ¹New York University School of Medicine, New York, NY; ²The Hebrew Home at Riverdale, Riverdale, NY. (Tracking ID #1929818)

BACKGROUND: The efficacy of interventions targeted at comprehensive therapeutic lifestyle changes (TLC) is well proven in reducing blood pressure (BP) among patients with hypertension (HTN). However their translation to primary care practices is limited, particularly among black patients with HTN, who share a disproportionately greater burden of HTN-related outcomes. More importantly, the comparative effectiveness of single session TLC interventions on BP reduction in primary care practices is unproven. The aim of this vanguard trial was to evaluate the comparative effectiveness of a practice-based comprehensive lifestyle intervention targeted at recommended TLC, delivered through group-based counseling and motivational interviewing (MINT), versus a single session counseling on lifestyle modification (SSC) in reducing BP at 6months among low-income, blacks with uncontrolled HTN.

METHODS: A total of 194 black patients were randomized to either the MINT-TLC or SSC group. The comprehensive lifestyle intervention (MINT-TLC) was based on established clinical practice guidelines for prevention and treatment of HTN, which recommends weight loss (if overweight), regular physical activity, limiting and/or reducing sodium and alcohol intake, and eating a low-fat diet that is rich in fruit and vegetables. Patients in the MINT-TLC group attended 10 weekly group classes (intensive phase) focused on TLC; followed by 3 monthly individual MINT sessions (maintenance phase) delivered by trained Health Educators. Patients in the SSC condition received a single 30-minute individual counseling session on therapeutic lifestyle changes at the baseline visit by trained study staff. To match the MINT-TLC group for content of intervention material, those in the SSC group also received print versions of the intervention curriculum that was distributed in the group classes. The primary outcome was change in systolic BP and diastolic BP at 6months assessed with an automated BP monitor (WatchBP). The average of three BP readings was used as the primary outcome measurement, following standard American Heart Association guidelines. The primary analyses examined systolic and diastolic BP separately based on intent-to-treat. A repeated measures mixed model approach was used to account for continuous primary outcomes collected at three waves (baseline, 3 and 6months), and clustering within primary care providers. The post-treatment values of continuous outcomes were modeled as functions of baseline values, treatment and the interaction of baseline and treatment.

RESULTS: The mean age of all patients was 57years, 50% were women, 69% had income <$20,000/year with mean baseline BP 147.4/89.3mmHg. Average attendance at the group-based classes in the MINT-TLC condition was 50%; 35% of patients completed all three individual MINT sessions. There was non-significant reduction in systolic and diastolic BP for the MINT-TLC compared to the SSC group. The net adjusted reduction in systolic BP by 6months was 12.9mmHg for the SSC vs. 9.5mmHg for the MINT-TLC group. The reduction in diastolic BP was 7.6mmHg for the SSC vs. 7.2mmHg for the MINT-TLC group.

CONCLUSIONS: Despite the non-significant between-group difference in BP reduction in the intensive group counseling with multiple office visits versus the single counseling session plus educational material both groups exhibited comparable and clinically meaningful BP reduction. Evaluation of the effectiveness of the interventions in a large Phase 3 trial is warranted in this patient population. Trial Registration: Clinicaltrials.gov NCT01070056

COMPARING CLINICIANS’ PERCEPTION OF THEIR OWN AND THEIR PEERS’ ANTIBIOTIC PRESCRIBING TO ACTUAL ANTIBIOTIC PRESCRIBING FOR ACUTE RESPIRATORY INFECTIONS IN PRIMARY CAREHarry Reyes Nieva^1,2; Jason N. Doctor³; Mark W. Friedberg^1,4; Caroline Birks^2,5; Yelena Kleyner⁶; Julie Fiskio¹; Lynn A. Volk⁶; Jeffrey A. Linder^1,2. ¹Brigham and Women, Boston, MA; ²Harvard Medical School, Boston, MA; ³University of Southern California, Los Angeles, CA; ⁴RAND Corporation, Boston, MA; ⁵Massachusetts General Hospital, Boston, MA; ⁶Partners Healthcare System, Boston, MA. (Tracking ID #1941741)

BACKGROUND: Primary care clinicians often prescribe inappropriate antibiotics for acute respiratory infections (ARIs). Inappropriate prescribing may be a result of differences in clinicians’ perceptions of their own antibiotic prescribing rates, perceptions of their peers’ antibiotic prescribing rates, and actual antibiotic prescribing rates.

METHODS: Using a mixed methods approach, we first surveyed 269 primary care clinicians in the Partners Primary Care PBRN about their agreement with antibiotic prescribing guidelines for non-specific upper respiratory infections (URIs), pharyngitis, acute bronchitis, and sinusitis. For each of these diagnoses, the survey asked clinicians to estimate ARI antibiotic prescribing rates for themselves and their peers in five categories: 0%–20%, 21%–40%, 41%–60%, 61%–80%, or 81%–100%. We then calculated respondents’ actual ARI antibiotic prescribing rates between May 2011–April 2012 using billing and electronic health record data and grouped them using the same five antibiotic prescribing rate categories as the survey. We used the Kruskal-Wallis test to compare respondents’ estimates of their own antibiotic prescribing rates to that of their peers.

RESULTS: 169 of 269 (63%) primary care clinicians in 23 practices responded to our survey. Mean respondent age was 49years, 60% were women, and 85% were physicians. We linked respondents to 20,025 ARI visits. Clinicians overwhelmingly agreed with antibiotic treatment guidelines for non-specific upper respiratory infections (100% of clinicians), pharyngitis (95%), acute bronchitis (92%), and sinusitis (98%). For non-specific URIs, the proportion estimating antibiotic prescribing rates in the five categories (i.e., 0%–20%, 21%–40%, 41%–60%, 61%–80%, or 81%–100%) for themselves was 72%, 19%, 7%, 2%, and 0%, respectively, and for their peers was 37%, 41%, 17%, 5%, and 0%, respectively (p < 0.001). The actual non-specific URI antibiotic prescribing rate was 34%; 75% underestimated and 3% overestimated their own rate. For pharyngitis, the proportion estimating antibiotic prescribing rates in the five categories for themselves was 81%, 13%, 2%, 2%, and 1%, respectively, and for their peers was 63%, 26%, 8%, 3%, 0%, respectively (p < 0.001). The actual pharyngitis antibiotic prescribing rate was 59%; 71% underestimated and 0% overestimated their own rate. For acute bronchitis, the proportion estimating antibiotic prescribing rates in the five categories for themselves was 48%, 31%, 14%, 6%, and 1%, respectively, and for their peers was 21%, 40%, 27%, 10%, and 2%, respectively (p < 0.001). The actual acute bronchitis antibiotic prescribing rate was 76%; 76% underestimated and 1% overestimated their own rate. For sinusitis, the proportion self-reporting antibiotic prescribing rates in the five categories for themselves was 35%, 41%, 19%, 3%, and 2%, respectively, and for their peers was 13%, 40%, 33%, 13%, 1%, respectively (p < 0.001). The actual sinusitis antibiotic prescribing rate was 88%; 86% underestimated and 2% overestimated their own rate.

CONCLUSIONS: Clinicians agree with ARI antibiotic treatment guidelines. Clinicians think their peers prescribe antibiotics at higher rates than themselves. Clinicians’ estimates of their own and their peers’ prescribing rates are lower than actual prescribing rates. To reduce inappropriate antibiotic prescriptions for ARIs, future interventions might correct clinicians’ misperceptions of their own and their peers’ antibiotic prescribing rates.

COMPARING WARFARIN ANTICOAGULATION MEASUREMENTSZayd A. Razouki^1,2; James Burgess³; Shibei Zhao²; Adam Rose^1,2. ¹Boston Medical Center, Boston, MA; ²Bedford VA Medical Center, BEdford, MA; ³Boston University School of Public Health, Boston, MA. (Tracking ID #1932592)

BACKGROUND: Percent time in therapeutic range (TTR) is a measure of anticoagulation intensity, while INR variability is a measure of anticoagulation stability. Both are valid intermediate measures for warfarin anticoagulation control that predict warfarin associated complications (WAC, i.e. ischemic stroke, major bleeding including fatal bleeding) in patients with atrial fibrillation. However, less is understood about how these two measures complement each other, when used together. This study examines how TTR and INR variability predict WAC, and compares them to a composite measure that combines both of them.

METHODS: We included 40,404 patients anticoagulated for atrial fibrillation, age 65+, within the Veterans Health Administration. We used ICD 9 codes to define WAC as our outcome. For each patient, we calculated TTR using Rosendaal’s method, INR variability using Fihn’s method, and constructed a composite measure (CM) using an equally weighted method, adding standardized TTR to the standardized log-transformation of INR variability. High TTR and low log INR variability represent good anticoagulation. Therefore, we converted the direction of log INR variability to make it consistent with TTR. We divided our sample into quintiles for each measure based on their level of anticoagulation control (i.e. very poor, poor, moderate, good and excellent control). We used Cox regression models to predict WAC; hazard ratios were compared across quintiles of TTR, log INR variability and the intermediate composite measure. The reference group was the quintile of “excellent control”

RESULTS: The composite measure predicted ischemic stroke better than TTR and log INR variability in all quintiles, especially in the first quintile of “very poor control” (HR = 2.43 vs. HR = 2.10 and HR = 1.74, respectively). CM and TTR predicted major bleeding similarly except in the second quintile of “poor anticoagulation” (CM HR = 1.57 vs. TTR HR = 1.35); but both measures were generally better than log INR variability in all quintiles. CM was superior to both TTR and log INR variability when predicting fatal bleeding across all quintiles, especially in the “very poor control” quintile (HR = 3.04 vs.HR = 2.49 vs.HR = 2.18, respectively) and the moderate control quintile (HR = 1.45 vs. HR = 1.19 vs.HR = 1.19 respectively). We were not able to calculate p values to compare whether differences in hazard ratios were statistically significant because individual patients were not distinctly separated across the three studied measures.

CONCLUSIONS: Our composite measure encompasses two aspects of anticoagulation control, namely intensity and stability. Our findings suggest its usefulness in measuring anticoagulation control. It has an advantage in predicting ischemic stroke and fatal bleeding over TTR and log INR variability but it is only better than the log INR variability in predicting major bleeding.

COMPARISON OF EPIGENETIC VERSUS INTENSIVE CHEMOTHERAPY FOR NEWLY DIAGNOSED ELDERLY ACUTE MYELOID LEUKEMIA PATIENTS ≥60 YEARS OLDNeha Gupta¹; Shipra Gandhi¹; Austin Miller²; Laurie A. Ford³; Elizabeth A. Griffiths³; James E. Thompson³; Meir Wetzler³; Eunice Wang³. ¹University at Buffalo, Buffalo, NY; ²Roswell Park Cancer Center, Buffalo, NY; ³Roswell Park Cancer Center, Buffalo, NY. (Tracking ID #1938243)

BACKGROUND: Epigenetic therapy (Epi) with the hypomethylating agents (eg. azacitidine and decitabine) is increasingly being utilized for induction treatment of elderly Acute myeloid leukemia (AML) patients based on studies demonstrating both tolerability and prolonged survival. By contrast, standard intensive chemotherapy (IC) with cytarabine and daunorubicin effectively induces remission in many individuals but is associated with significant toxicity and higher mortality. We compared our institute’s experience with Epi vs. IC for the upfront treatment of newly diagnosed AML patients ≥60years old.

METHODS: We performed a retrospective chart review of 164 patients ≥60year old with newly diagnosed AML at Roswell Park Cancer Center between 3/2008 and 2/2013. N = 84 patients received IC, and n = 82 patients received Epi. Kaplan Meier method, log rank test, and multivariate analysis with cox proportional hazard models were used to assess overall survival (OS) and correlation with covariates.

RESULTS: Baseline patient characteristics demonstrated a difference in median age of patients in each group (IC 67years vs. Epi 75years; p < 0.01). All other factors were comparable. At our center, older AML patients receiving IC had superior complete response at any time point (CRatp) (43% vs. 21%; p < 0.01). IC also resulted in a longer median OS compared to Epi (10.6 vs. 7.9months; p = 0.01). Thirty-day mortality and leukemia-free survival (LFS) were similar across the two groups (IC 10% vs. Epi 11%; p = 0.8; 11.2 vs. 9.3months; p = 0.47 respectively). Choice of Epi drug (Azacitidine vs. Decitabine) did not impact results. In multivariate Cox regression analysis, older age, higher ECOG score, increased peripheral blasts, and poor-risk cytogenetics were independently associated with inferior survival.

CONCLUSIONS: Our results suggest that IC and Epi represent clinically equivalent approaches for the upfront treatment of older AML patients. In spite of significantly higher complete response (CR) and overall response rate (ORR) in the IC group, our finding of improved OS following IC vs. Epi was not substantiated in multivariate analysis. This finding suggests that difference in response of two agents in univariate analysis may be explained by the comparatively younger age of patients in the IC group. Leukemia-free survival and 30-day mortality were the same for IC vs. Epi-treated patients, as were all response and survival outcomes in the poor-risk cytogenetics subgroup. These data highlight the growing need for prospective clinical trials to conclusively determine the respective roles of IC vs. Epi therapy in older AML patients.

COMPARISON OF READABILITY OF PATIENT DISCHARGE INSTRUCTIONS WITH AND WITHOUT USE OF DISEASE-SPECIFIC TEMPLATESStephanie Mueller; Kyla Giannelli; Robert Boxer; Jeffrey L. Schnipper. Brigham and Women, Boston, MA. (Tracking ID #1937804)

BACKGROUND: In the United States, 21–23% of the adult population have low literacy, and an additional 27% have low health literacy (difficulty understanding written health education materials). Written materials are commonly used to communicate with patients at time of discharge, but can vary in readability level. We evaluated the association of the use of templated discharge instructions with readability scores of discharge instructions provided to patients at hospital discharge.

METHODS: We performed a retrospective cohort study of patients discharged from an academic medical center. Data were obtained from electronic medical records. Patients were eligible for inclusion if discharged from any inpatient service to home between October, 2011 and September, 2012, following initiation of a web-based “discharge module,” a quality improvement initiative to improve written communication at discharge, that included optional use of disease-specific templated discharge instructions. We examined the frequency of use of the templated discharge instructions. We then evaluated readability using two common measures: Flesch-Kincaid Grade Level (FKGL) and Flesch Reading Ease Level (FREL) in discharges that provided templated discharge instructions (with or without modification) vs. discharges that provided clinician-derived discharge instructions (where templated instructions were available and not used or were unavailable for the specific discharge diagnosis).

RESULTS: Of the 233 randomly selected patient discharges, 105 (45.1%) provided templated discharge instructions and 128 (54.9%) provided clinician-derived discharge instructions (Figure 1). The mean FKGL of templated discharge instructions was grade 5.6 vs. 7.6 for clinician-derived discharge instructions (p < 0.001). The mean FREL of templated discharge instructions was 71 vs. 57.4 for clinician-derived discharge instructions (p < 0.001). In subgroup analysis, FKGL of discharge instructions did not differ significantly when templated instructions were modified versus used without modification (5.8 vs. 5.5, p = 0.15). FREL of modified templated discharge instructions was 69.1 vs. 72.9 for templated instructions without modification (p = 0.02).

CONCLUSIONS: We found that we still lack many disease-specific templated discharge instructions within our “discharge module,” the major reason for clinicians to create discharge instructions themselves. Compared to discharges that used templated instructions, those that did not had significantly lower readability scores (higher FKGL and lower FREL). Modification of templated instructions, which may lead to more customized messages, did not appreciably detract from readability. Our findings suggest that templated instructions are more literacy friendly than clinician-derived discharge instructions.

CONTINUING MEDICAL EDUCATION (CME) PARTICIPANTS’ USE OF SPECIFIC TYPES OF SOCIAL MEDIA (SM) AND ASSOCIATION WITH ATTITUDES FOR USING SM IN CMEAlexander Vengerovsky; Amy T. Wang; Jayawant N. Mandrekar; Nicole P. Sandhu; Christopher M. Wittich; Karen F. Mauck; Thomas J. Beckman. Mayo Clinic, Rochester, MN. (Tracking ID #1939285)

BACKGROUND: Most research on social media (SM) in medicine pertains to issues of professionalism. One recent study found that continuing medical education (CME) course participants who use SM tend to be younger and favor the use of SM for enhancing CME learning. However, the potential link between the specific types of SM and participants’ views on using SM for CME remains unknown. Therefore, we determined what categories of SM are used by CME participants, along with the relationship between categories of SM use and favorable attitudes on using SM for CME.

METHODS: This was a cross-sectional survey of 539 participants at a Mayo Clinic Internal Medicine CME course in 2011. The Social Media Use and Perception Instrument (SMUPI) consisted of 10 items (5-point scales), and has been shown to have strong content, internal structure and criterion validity. Participants were surveyed regarding their demographic characteristics and individual use of SM. We also grouped types of SM into the following categories: professional (LinkedIn, Doximity, or Sermo), social (Facebook, Google+, Myspace, or Twitter), informational (blogs, wikis, podcasts, or StumbleUpon), and media (YouTube, Skype, Picasa, or Flickr). Comparisons of SMUPI scores for users versus non-users of each type of SM, and within categories of SM (professional, social, informational, and media), were calculated using the Wilcoxon Ranked sum test. The threshold for statistical significance was set at p < 0.05.

RESULTS: Of 539 CME participants, 327 (response rate = 61%) completed the survey and 291 (89%) reported using SM. CME course participants’ had generally favorable attitudes towards SM (Average SMUPI score 3.12). When comparing users and non-users of specific types of SM, higher SMUPI scores existed for users of Facebook (3.42 vs. 3.00, p < 0.0001), Google + (3.50 vs. 3.08, p < 0.0001), Twitter (3.74 vs. 3.19, p = 0.003), LinkedIn, (3.53 vs. 3.18, p = 0.007), Wikis (3.38 vs. 3.12, p = 0.02), and YouTube (3.32 vs. 3.09, p = 0.02). There were no significant SMUPI score differences for users versus non-users of Skype, blogs, podcasts, and Picasa. Finally, higher SMUPI scores were demonstrated for users versus non-users in the SM categories of professional (3.58 vs 3.17; p = 0.001), social (3.37 vs 2.82; p < 0.0001), and informational (3.35 vs 3.06; p = 0.008).

CONCLUSIONS: We are unaware of previous studies on the relationships between types of SM use and favorable attitudes towards utilizing SM for CME learning. Our results suggest that users of Facebook, Google+, LinkedIn, Twitter, and Wikis had the most positive attitudes regarding the use of SM in CME. Interestingly, these particular types of SM incorporate adult-learning principles of real-time interactivity and adaptability in order to optimize communication and community-building. We suggest that as CME directors increasingly utilize SM, they should focus on incorporating interactive types of SM into their CME courses.

CONTINUITY IN A VA PATIENT-CENTERED MEDICAL HOME REDUCES EMERGENCY DEPARTMENT VISITSKrisda Chaiyachati^1,2; Kirsha S. Gordon³; Theodore Long⁴; Ali M. Khan⁶; Forrest Levin^3,5; Emily M. Meyer²; Amy C. Justice^1,2; Rebecca Brienza^1,2. ¹Yale University School of Medicine, New Haven, CT; ²VA Connecticut Healthcare System, West Haven, CT; ³VA Connecticut Healthcare System, West Haven, CT; ⁴Yale University School of Medicine, New Haven, CT; ⁵Evergreen Design, Guilford, CT; ⁶Iora Health, Brooklyn, NY. (Tracking ID #1922254)

BACKGROUND: One major goal of the Patient Aligned Care Team (PACT), the VA’s Patient-Centered Medical Home (PCMH) model, is to improve continuity of care between patient and provider, partly because poor continuity is believed to contribute to the inappropriate use of healthcare services like the emergency department (ED). Hypothesizing that patients with high continuity of care levels would visit the ED less, we sought to model patient and clinical factors associated with varying continuity levels and assess the association between high levels of continuity and the frequency of ED visits in the year after the initial implementation of PACT teams at Connecticut’s West Haven VA (WHVA).

METHODS: We performed a retrospective analysis of patients with established care at the WHVA primary care clinic from March 1, 2011 to February 29, 2012. We defined continuity as a patient seeing their assigned primary care provider (PCP) or trainee—resident physician or nurse practitioner. Using a multivariable regression model we determined the association between continuity and the frequency of ED visits, controlling for demographics, patient assignment to a trainee, comorbidities, having outside insurance, military service period, and distance from the ED. Additional models analyzed individuals who had at least one continuity visit in the study year to determine the impact of levels of continuity on ED utilization.

RESULTS: Veterans meeting inclusion criteria totaled 13,495 individuals with 42,969 total Clinic visits. Among those, 3185 (24%) used the WHVA ED for 15,458 total ED visits. After controlling for covariates and potential confounders, continuity with an assigned PCP was independently associated with a 46% reduction in the odds of using the WHVA ED [adjusted odds ratio (AOR) 0.54; 95% confidence interval (CI): 0.41–0.71]. In the subset of patients having at least one primary care visit with continuity, those with medium (33–50% of visits) and high continuity (>50% of visits) levels were 30% less likely (AOR 0.70; 95% CI: 0.54–0.92) and 41% less likely (AOR 0.59; 95% CI: 0.25–0.76) to use the ED compared to patients with low continuity (<30% of visits). Controlling for low (<3 visits/year) versus high (≥3 visits/year) primary care users did not significantly alter the impact of continuity on ED utilization. Patients assigned to a trainee were associated with lower ED utilization (AOR 0.78; 95% CI: 0.69–0.89).

CONCLUSIONS: Improved continuity of care between patients and providers, including trainees, is associated with decreased ED utilization within the WHVA’s PACT system. To our knowledge, this is the first detailed analysis of continuity within a large PCMH model and assessment of the association between continuity and ED utilization. As outpatient centers and health systems adapt PCMH models to their specific setting, understanding the impact of patient-centered care indicators, like continuity, on healthcare utilization patterns and patient outcomes will be important for understanding PCMH components that may ultimately reduce health system costs.

CONTINUITY VISITS - A MODEL TO IMPROVE RELATIONAL CONTINUITY BETWEEN PATIENTS AND THEIR LONGITUDINAL PHYSICIANS DURING ACUTE HOSPITAL ADMISSIONSRyan Thompson; Chrisanne Sikora; Maria Veo; Keith Jennings; Timothy Ferris. Massachusetts General Hospital, Boston, MA. (Tracking ID #1941793)

BACKGROUND: With the rise of hospital medicine, and professional pressures facing ambulatory-based physicians, fewer patients are seeing the longitudinal physicians they know best when hospitalized. Research has shown that a majority of hospitalized patients feel they receive better care and experience higher levels of trust when cared for by physicians they have known over time. To address this decline in “relational continuity,” we began a pilot program to incentivize physicians identified as having a longitudinal relationship with a patient to make a “Continuity Visit” during the patient’s hospital admission. A secondary aim of the program is facilitate stronger clinical collaboration between hospital and ambulatory physicians based on their unique knowledge of the patient.

METHODS: A Continuity Visit is a brief one-time inpatient visit by a patient’s longitudinal physician with the patient and inpatient care team. Physicians eligible to participate include primary care physicians (PCPs), specialist physicians, and hospitalist physicians (for patients readmitted within 30days). We provide a small reimbursement to physicians who make a Continuity Visit, and document their visit in the medical record. Physicians are ineligible to participate if they are actively involved in the patient’s current admission, as determined by billing data. PCPs at our institution have received automated electronic notifications of their patients’ Emergency Department visits and hospital admissions for many years. We developed a parallel notification system for specialist physicians by using electronic billing data and an algorithm that identifies longitudinal relationships between patients and specialists. We also began sending hospitalists automated notifications of their 30-day patient readmissions. These electronic “continuity notifications” contain the patient’s admitting diagnosis and hospital location, and a brief outline of the Continuity Visit program, including instructions on how to make and bill for a Continuity Visit. We tracked departmental variation in the number of Continuity Visits made since the launch of the program, and administered an anonymous survey to participating physicians regarding the perceived value of Continuity Visits to their patients and colleagues, and regarding their professional satisfaction with Continuity Visits.

RESULTS: Since the program’s launch in January 2012, 4164 hospitalized patients received a Continuity Visit from a physician with whom they have a longitudinal relationship. Specialists from 12 different departments made and billed for 2916 of these visits (70%), PCP’s 1100 visits (26%), and hospitalists 148 visits (4%). Specialties with the highest volume of Continuity Visits included oncology, obstetrics, and cardiology. Comparing the first quarters of 2012 and 2013, the number of Continuity Visits increased by approximately two-fold among specialists, two-fold among PCPs, and four-fold among hospitalists. We surveyed physicians who made a Continuity Visit to one or more patients during a 3month period (July–September 2013; n = 182). Of 106 respondents (58%), 52 were specialists, 43 were PCPs, and 8 were hospitalists. Ninety-seven percent of respondents felt that their Continuity Visit was useful to the patient visited (59% highly useful, 38% moderately useful; see Table 1) Eighty percent felt their Continuity Visit was useful to the patient’s inpatient attending physician (37% highly useful, 43% moderately useful). Lastly, 96% of physicians reported personal and/or clinical satisfaction from making the Continuity Visit (61% highly satisfied, 35% mostly satisfied; see Table 2).

CONCLUSIONS: Continuity Visits to hospitalized patients by their longitudinal physicians can facilitate stronger relational continuity between patients and physicians, improve collaboration between inpatient and ambulatory physicians, and improve information transfers across transitions of care. We found participating physicians to be highly satisfied with the Continuity Visit model. Further analysis is needed to assess whether the involvement of longitudinal physicians through Continuity Visits has impact on clinical and quality outcomes, including hospital length of stay, readmission rates, and standardized patient experience.

Reasons for Physician-Perceived Usefulness of their Continuity Visit to the Patient Visited (n = 106)

* Respondents could check more than one response option

Reasons for Physician-Reported Satisfaction with Continuity Visits (n = 106)

* Respondents could check more than one response option

CONTRACEPTIVE USE IN WOMEN HAVING BARIATRIC SURGERYRachel S. Casas; Iris Tong; Ghada Bourjeily. Warren Alpert Medical School of Brown University, Providence, RI. (Tracking ID #1936953)

BACKGROUND: Contraceptive counseling in women undergoing bariatric surgery is crucial due to the increased risk of adverse pregnancy and fetal outcomes in the 1 to 2years following surgery. Women are typically advised to avoid pregnancy during this time period. Oral contraceptive pills (OCPs) are commonly used in the post surgical population, but their effectiveness may be limited by malabsorption. Intrauterine devices (IUDs) and surgical sterilization are the most effective contraceptive options for women not desiring future pregnancy. The aim of this study is to determine if women undergoing bariatric surgery are using and counseled about appropriate contraceptive options.

METHODS: A 36 question survey was sent electronically to pre- and post-surgical patients at the Center for Bariatric Surgery at the Miriam Hospital in Providence, Rhode Island. The survey consisted of questions about demographic data, contraceptive choices, and contraceptive counseling from healthcare providers. The survey did not include any questions containing patient identifying information. Included patients were aged between 18 and 44years, sexually active with men, pre-menopausal, and lacked a history of hysterectomy or premature ovarian failure.

RESULTS: A total of 35 women completed the survey and met inclusion criteria. Overall, 30 participants (86%) were counseled about contraceptives and 28 participants (80%) were counseled to avoid pregnancy in the 1 to 2years following surgery. Of the 21 women who had bariatric surgery within the last 2years, 14 women (67%) were using contraception, and seven women (33%) were not. Among the seven contraceptive non-users post bariatric surgery, all were counseled to avoid pregnancy 1 to 2years post surgery, six were counseled about contraceptive choices, and five did not desire pregnancy. Reasons stated for contraceptive nonuse in this subgroup included side effects and lack of regular partner. Of the 21 women who had bariatric surgery within the last 2years, 11 (52%) were using OCPs, three (14%) were using male condoms, two (10%) were using IUDs, and many women were using multiple forms of birth control. The majority of women chose their contraceptive methods based upon ease of use, ease of accessibility, and recommendation of healthcare provider. No pregnancy was reported following surgery.

CONCLUSIONS: While the majority of women undergoing bariatric surgery were using and counseled about contraceptives, many post surgical women were using less effective forms of contraception. Continued contraceptive counseling with a focus on the most effective methods, the IUD or sterilization for women not desiring future pregnancy, is needed in this population.

COORDINATING CANCER CARE: WHAT ORGANIZATIONS DO TO DELIVER HIGH-QUALITY BREAST CANCER CARE. A FUZZY SET QUALITATIVE COMPARISON ANALYSISNina A. Bickell¹; Alexandra Moss²; Ann S. McAlearney². ¹Mount Sinai School of Medicine, New York, NY; ²The Ohio State University, Columbus, OH. (Tracking ID #1938688)

BACKGROUND: Underuse of adjuvant breast cancer treatments delivered by different specialists may be worsened by fragmented care and improved by effective coordination. To improve care coordination and adjuvant treatment delivery, we assessed approaches to coordinate cancer care at hospitals serving predominantly minority breast cancer patients, hospitals historically at higher risk for poorer quality.

METHODS: We interviewed 89 key informants (n = 58 clinical; n = 16 administrative; n = 12 clerical; n = 3 other) from 11 inner-city hospitals with high volumes of minority breast cancer patients to better understand how organizational characteristics might impact coordination of care, implementation and success of the T & F innovation. We used standard techniques to code the data from the key informant interviews and then used fuzzy set Qualitative Comparative Analysis (fsQCA) as the analytical method. For these results we focused our analyses on five conditions that had particular impact on successful implementation of the T&F innovation: handoffs; organizational attention to clinic patients; no-shows; patient-centeredness; and integrated EHRs. We then divided the hospitals into three outcome groups (<10% underuse, 10–20% underuse, >20% underuse), and calibrated each of the conditions for six-value fsQCA using fuzzy set QCA methods.

RESULTS: Five organizational factors appeared particularly salient with respect to impacting coordination of adjuvant breast cancer care. First, with respect to tracking no-shows, the better sites had some way of keeping track of which no-shows actually came back and which no-shows needed to be contacted again; further, this tracking process was someone’s designated job. How organizations managed handoffs in care was also important. Handoffs at the better sites included an exchange of clinical information (e.g., doctors speaking directly to other doctors) as well as a clerical transfer of responsibility (clerks making appointments for patients with other specialties). Handoffs at the less good sites included one or none of these elements. Another key organizational factor involved the level of attention organizations paid to clinic patients. While some sites tolerated hectic, understaffed, and ineffective clinics, the better sites did not. A fifth critical factor was whether sites had an integrated EHR. We found that all the better sites had fully implemented and integrated EHRs, where providers could see what is going on with their patient in other specialties. Some sites had multiple EHRs leaving physicians unable to access needed treatment information. Finally, organizations that appeared to have a culture of patient-centeredness also appeared to have better quality of care. In our study, most of the good sites had a clear focus on trying to make things easy for the patient/the attitude that the hospital works for the patient and not the other way around.

CONCLUSIONS: As care coordination across sites and specialties is encouraged by federal law and regulation, specialty care silos and rigid communication systems still pose barriers to change. Our results suggest that organizational factors can have an important role impacting care coordination for breast cancer care, and highlight five mechanisms that may be truly critical with respect to ensuring good coordination and quality of breast cancer care.

CORRELATES OF SEXUAL SATISFACTION IN MIDLIFE WOMEN: COMMUNICATION IS KEYHolly N. Thomas^1,2; Rachel Hess¹. ¹University of Pittsburgh, Pittsburgh, PA; ²VA Pittsburgh Healthcare System, Pittsburgh, PA. (Tracking ID #1935614)

BACKGROUND: Sex is an essential component of quality of life. While most studies of sex in women focus on sexual dysfunction, sexual satisfaction is just as important. Characteristics related to sexual satisfaction in women are not well-defined and likely change with aging. We used a population-based cohort of midlife and older women to define factors related to sexual satisfaction.

METHODS: The National Survey of Midlife Development in the United States (MIDUS) is a nationally representative cohort study of adults ages 28 to 84. Self-administered questionnaires collected information regarding demographics, psychosocial factors, and physical and mental health. The second wave (2004–2006) included questions about sex. Sexual satisfaction was assessed by: “How would you rate the sexual aspect of your life these days?” with a 0 (worst) to 10 (best) visual analog scale. Women who reported any sexual activity in the prior 6months were included in the primary analyses. Descriptive statistics were used to compare sexually active and inactive women. Post-stratification weights based on race, age, and education status were used to adjust percentages to the general population. Univariable ordered logistic regression was used to assess the relationship between (1) sociodemographics, (2) relationship characteristics, and (3) physical and mental health of the woman and her partner, and the outcome (sexual satisfaction). Significant (p < 0.05) variables were entered into a multivariable model.

RESULTS: Of the 2647 women in MIDUS, 1345 (63.6%) were sexually active in the prior 6months and 531 did not answer the questions regarding sex. The proportion of sexually active women decreased with age. However, among those married or living with a partner, 81.8% were sexually active overall, and 56.7% of women 65 and older were sexually active (Table 1). In the multivariable model, better communication with partner, higher frequency of sex, better self-rated mental health, earlier menopausal status, lack of dyspareunia, and no use of antidepressants were related to higher sexual satisfaction at the p < 0.05 level (Table 2). At p ≤ 0.002, only communication and frequency of sex were significant (Bonferroni adjustment for multiple comparisons).

CONCLUSIONS: As long as they have a partner, a large proportion of women are sexually active in midlife and beyond. Neither age nor physical health is significantly related to sexual satisfaction in midlife women. Women who report better communication with their partners have higher sexual satisfaction. Providers caring for women should not assume that sexual satisfaction declines with aging. Improving communication with one’s partner may be a key target in the treatment of sexual problems in women.

Proportion of sexually active women in MIDUS (if married or cohabitating)

Factors associated with higher sexual satisfaction among sexually active women in MIDUS

Adjusted for relationship satisfaction/strain, intimacy, importance of sex, prior sexual satisfaction, age, generation, physical health, chronic diseases, sexual assault, hormone therapy use, BMI, partner status, relationship length, partner disease, partner physical and mental health, difficulty paying bills

COST EFFECTIVENESS COMPARISON OF SCREENING STRATEGIES FOR LATENT TUBERCULOSIS INFECTION AMONG BACILLUS CALMETTE-GUERIN VACCINATED HEALTH CARE WORKERSMehrshid Kiazand^1,2; Noor Khan^1,2; Kenneth J. Smith^2,3. ¹UPMC Mercy, Pittsburgh, PA; ²University of PIttsburgh School of Medicine, Pittsburgh, PA; ³UPMC, Pittsburgh, PA. (Tracking ID #1906127)

BACKGROUND: The interferon- gamma release assay (IGRA) is recommended as an alternative to the tuberculin skin test (TST) to detect latent tuberculosis infection (LTBI). IGRA is the preferred test for Bacillus Calmette- Guerin (BCG) vaccine recipients. Despite recommendations, TST is frequently used in many institutions to screen for LTBI, regardless of previous BCG vaccination. We evaluated the cost-effectiveness of alternative strategies for LTBI detection among BCG vaccinated health care workers.

METHODS: We have designed a decision tree to compare three screening strategies in BCG recipients; (1) TST alone; (2) IGRA alone and (3) TST, if positive, then IGRA. In the base case, we assume that LTBI prevalence (assessed by TST among foreign born U.S. residents) is 18%. In BCG vaccinated populations TST sensitivity and specificity is 71% and 56%, while IGRA sensitivity and specificity is 76% and 96%. Sixty-three percent of health care workers will accept isoniazid (INH) for LTBI treatment. The cost of TST is $9.79 and IGRA is $33.67. The cost of INH treatment for 9months is $508.95 and the average cost of managing INH related hepatitis (4% risk) is $539.74.

RESULTS: In the base case, the IGRA alone strategy cost $23.62 more per person than TST followed by IGRA ($88.87 vs. $62.55), but IGRA alone increased the LTBI absolute detection and treatment rate by 2.5% (8.6% vs. 6.1%). Thus compared to TST then IGRA, IGRA alone cost $1053 per case detected and treated. Compared to IGRA alone, the TST alone strategy is dominated, costing more ($168.89 vs. $88.87, due to more false positives receiving INH) with fewer cases detected and treated. Compared to TST then IGRA, TST alone costs more than $5500 per case. IGRA alone dominates TST alone if IGRA testing costs less than $120. TST alone is no longer dominated and costs less than $1000 per case, if 10% or less of health care workers agree to INH treatment. Model results were robust to variation of other parameter values.

CONCLUSIONS: TST alone to detect LTBI in BCG vaccinated populations will likely cost more than other strategies and detect fewer cases than an IGRA alone strategy. However, IGRA alone costs more than $1000 per additional case detected and treated compared to TST then IGRA. It is unclear if this cost is acceptable to payers. Although not strongly recommended, IGRA after a positive TST may be more economically reasonable than other strategies in BCG recipients and worthy of future study.

COST-EFFECTIVENESS OF DECISION SUPPORT STRATEGIES FOR SAFELY REDUCING ANTIBIOTIC USE IN ACUTE BRONCHITISConstantinos I. Michaelidis¹; Melissa S. Kern²; Kenneth J. Smith³. ¹University of Pittsburgh School of Medicine, Pittsburgh, PA; ²Geisinger Health System, Danville, PA; ³University of Pittsburgh School of Medicine, Pittsburgh, PA. (Tracking ID #1929178)

BACKGROUND: A recent clinical trial suggests that printed (PDS) and computer decision support (CDS) strategies can safely reduce antibiotic use in management of acute bronchitis compared to usual care (UC). Yet, little is known about the cost-effectiveness of these strategies. We evaluate the cost-effectiveness of PDS and CDS strategies in safely reducing antibiotic use in management of acute bronchitis.

METHODS: We developed a clinical trial based cost-effectiveness analysis comparing three strategies for acute bronchitis management: UC, PDS and CDS. We assumed a societal perspective, 5year program duration and 30day time horizon after each office visit. Our cohort was the U.S. population aged 13–64years presenting with acute bronchitis in an ambulatory setting. The primary differences between the three strategies were the added intervention costs and reduced likelihood of antibiotic use in the PDS and CDS strategies relative to the UC strategy. Estimated per patient program implementation costs in year 1 were $1.07 in the PDS strategy and $18.11 in the CDS strategy. We performed a base case cost effectiveness analysis and one-way and probabilistic sensitivity analyses in which the primary outcome was the societal cost per antibiotic safely avoided. We assumed that the societal willingness-to-pay (WTP) threshold was $43 per antibiotic safely avoided based on prior estimates of the downstream hidden societal costs of antibiotic resistance attributable to each ambulatory antibiotic prescription.

RESULTS: In the base case analysis, the PDS strategy dominated both the UC and CDS strategies, with lesser total costs (PDS: $2,614, UC: $2,629, CDS: $2,635) and fewer antibiotic prescriptions (PDS: 3.79, UC: 4.60, CDS: 3.95 antibiotic prescriptions) per patient over 5years. In one-way sensitivity analyses, the PDS strategy dominated the UC strategy across all tested parameter values except when antibiotics reduced the duration of work loss by ≥0.1days (base case: 0.0days) or cost ≤$10 (base case: $30). The PDS strategy dominated the CDS strategy across all tested parameter values except when the adjusted odds of antibiotic use during the intervention period relative to baseline was ≥0.64 in the PDS strategy (base case: 0.57) or ≤0.57 in the CDS strategy (base case: 0.64). The PDS strategy cost less than the societal WTP threshold ($43 per antibiotic safely avoided) across all tested parameter values except when antibiotics reduced the duration of work loss by ≥0.5days (base case: 0.0days). The analysis was not sensitive to variation in the costs of physician education or patient educational brochures in the PDS and CDS strategies, posters in the PDS strategy or electronic medical programming in the CDS strategy. In a probabilistic sensitivity analysis, varying all parameters simultaneously, the probability of the PDS strategy being preferred at a societal WTP threshold of $43 per antibiotic safely avoided was 66.4%. The PDS strategy was preferred across all thresholds from $0–$100 per antibiotic safely avoided.

CONCLUSIONS: A printed decision support strategy to safely reduce antibiotic use in ambulatory management of acute bronchitis is likely to be less costly and more effective than usual care and computer decision support strategies. This analysis suggests that even a modestly effective intervention to safely reduce antibiotic prescribing can be economically reasonable when implementation costs are low.

CT SCAN OVERUSE IN FREQUENTLY ADMITTED MEDICAL PATIENTSMina Owlia; Gregory M. Bump; Franziska Jovin; Christopher Deible; Marion Hughes; Lan Yu. University of Pittsburgh Medical Center-Oakland, Pittsburgh, PA. (Tracking ID #1903461)

BACKGROUND: Computed Tomography (CT) is an overused diagnostic tool. Clinicians may have limited understanding of how often CT scans lack clinically important findings. We describe the frequency of CT scan use in patients with frequent medical hospitalizations as well as the frequency of clinically important findings identified on these examinations.

METHODS: We retrospectively reviewed all CT scans done in 130 patients with greater than 7 admissions to medical services between January 1, 2011 and Dec 31, 2011 within an integrated health care system. We calculated the number of CT scans performed, the anatomic site of imaging, the source of ordering (emergency department, inpatient floor, or intensive care unit) and scored all CT scans using a novel scoring system on a 0–4 scale based on the severity of radiographic findings. Higher scores signified more clinically important findings with a score of 0 representing a normal study, a score of 1 showing minimal abnormalities with no need for follow up, score of 2 showing minor abnormalities which likely require followup, a score of 3 showing major abnormalities likely to affect management, and a score of 4 representing a critical or emergent finding.

RESULTS: A total of 795 CT scans were performed with a mean of 6.7 (±SD 5.8) CT scans per patient. Abdominal/Pelvis (39%), chest (30%), and head (22%) CT scans were the most frequently obtained. The most common site of CT scan ordering (53.7%) was the inpatient floors. Head CTs had the lowest rate of clinically significant findings (4%), compared to 38.3% of chest CTs and 31.5% of abdominal/pelvis CTs (Table 1).

CONCLUSIONS: Patients with frequent medical admissions undergo multiple CT scans that infrequently show findings that change clinical management. In particular, head CTs are rarely associated with significant findings and should be ordered less frequently. Inter-disciplinary measures should be advocated by hospitalists, emergency department physicians, and radiologists to decrease unnecessary imaging in this population.

Distribution and frequency of reconciled scan scores based on type of CT scan

CULTURAL COMPETENCE AND PERCEPTIONS OF COMMUNITY HEALTH WORKERS’ EFFECTIVENESS FOR IMPROVING HYPERTENSION OUTCOMES AND REDUCING HEALTH DISPARITIESMeta Mobula¹; Mekam T. Okoye¹; L. Ebony Boulware²; Kathryn A. Carson^1,3; Jill A. Marsteller³; Lisa A. Cooper^1,3. ¹Johns Hopkins University School of Medicine, Baltimore, MD; ²Duke University School of Medicine, Durham, NC; ³Johns Hopkins Bloomberg School of Public Health, Baltimore, MD. (Tracking ID #1938515)

BACKGROUND: Community Health Worker (CHW) interventions improve hypertension control among ethnic minorities, but health professionals’ perceptions of their effectiveness may impede integration of CHWs into health care delivery systems. Health professionals’ cultural competency has also been identified as a potential strategy to improve health outcomes of patients from diverse social groups. We explored the association between provider and staff self-reported cultural competency and preparedness to deliver cross-cultural care with attitudes towards effectiveness of CHWs in reducing health care disparities and improving clinical outcomes.

METHODS: We administered a questionnaire to providers (physicians and nurse practitioners) and clinical staff (nurses, medical assistants and others) from six urban primary care practices in Maryland to assess their self-reported cultural competence; preparedness to deliver cross-cultural care; and perceptions of a proposed CHW intervention. Cultural competence was assessed in three domains: motivation to learn about other cultures, cultural behavior and power and assimilation. We quantified the associations of cultural competence and preparedness with attitudes towards the effectiveness of CHWs using logistic regression adjusting for respondent age, race, gender, provider/staff status and years at the practice.

RESULTS: We contacted 200 providers and clinical staff, and 119 (60%) participated. Providers and staff reporting more cultural motivation had higher odds of perceiving CHWs as helpful for improving patient outcomes (OR = 2.56, 95% CI = 1.32–4.96) and reducing health disparities (OR = 10.05, 95% CI = 3.55–28.44). Those reporting more frequent culturally competent behaviors also had higher odds of believing CHWs would help improve patient outcomes (OR = 1.96, 95% CI = 1.02–3.76) and reduce health disparities (OR = 3.59, 95% CI = 1.62–7.99). Providers’ attitudes towards power and assimilation for socially disadvantaged groups were not associated with their perceptions of CHWs. Cultural preparedness was associated with providers’ perceived utility of CHWs in reducing health disparities (OR = 2.32, 95% CI = 1.20–4.47), but not significantly associated with improving health outcomes (OR = 1.72, 95% CI = 0.88–3.35).

CONCLUSIONS: Providers and staff with higher self-assessed cultural motivation, culturally competent behaviors, and cultural preparedness have more positive expectations of CHW interventions to reduce health disparities. The group that self-assesses as more culturally competent may know more about specific challenges faced by underserved patients and thus be more likely to see advantages of involving CHWs to overcome those challenges. Cultural competence training for providers and staff may complement the use of CHWs and support their effective integration into primary care clinics that are seeking to improve health outcomes and reduce disparities.

CULTURAL TAILORING OF STROKE RISK FACTOR REDUCTION AND WALKING PROMOTION COMMUNITY INTERVENTION FOR SOUTH ASIAN OLDER ADULT IMMIGRANTSNazleen Bharmal¹; Brian S. Mittman²; Raina Shah³; Catherine Sarkisian^2,1. ¹UCLA, Los Angeles, CA; ²VA Greater Los Angeles Healthcare System, Los Angeles, CA; ³University of California, Irvine, Irvine, CA. (Tracking ID #1933529)

BACKGROUND: South Asians are people with origins from India, Pakistan, Bangladesh, Nepal, and Sri Lanka. They are among the fastest growing U.S. immigrant groups. Epidemiological studies in Western countries have documented disproportionately high prevalence and mortality rates of coronary heart disease and stroke, with much of the excess stroke mortality in South Asians attributed to physical inactivity. Our objectives were to culturally-tailor an existing stroke risk factor reduction/walking promotion intervention and assess the feasibility of implementing the proposed intervention in senior centers for South Asians.

METHODS: We used community-based participatory research (CBPR) and implementation science methods to collect and analyze data from a community advisory board (CAB), focus groups, and stakeholder interviews in an iterative fashion. Community partners included a South Asian community-based organization (CBO) and two senior centers located near ethnic enclaves of low-income, recently immigrated South Asians. The theoretically-grounded intervention curriculum was designed as part of a NIH/NINDS-funded RCT and incorporated elements of moderated small group discussions, scenarios, didactics, goal-setting and reflection over eight sessions. CAB members reviewed the initial curriculum and provided input on the curriculum approach, content, context, and language based on the ecological validity and community-based developmental models to cultural adaptation. Using CAB input on the curriculum, we conducted three focus groups of Hindi-speaking South Asian older adults to identify knowledge and awareness of stroke risk factors and barriers and facilitators to walking. Focus group audio recordings were translated into English and transcribed. Using content analysis, CAB members reviewed the focus group transcripts and generated themes of stroke awareness and walking beliefs, as well as potential curriculum modifications to address these beliefs. We also interviewed key stakeholders to identify general and site-specific factors associated with greater likely effectiveness of the proposed intervention.

RESULTS: During the initial review, CAB members recommended changes to the intervention curriculum to be better received by South Asian seniors (Table 1). The 33 focus group participants had a mean age of 72.8years. Most were male (55%), had at least a college degree from India (65%), had lived in the U.S. for less than 20years (56%), and were mostly or very comfortable speaking English (58%). Participants had been diagnosed with stroke (19%), hypertension (64%), and hyperlipidemia (44%), and engaged in walking (67%) or yoga (33%) for physical activity. Themes identified by CAB members from the focus groups to address in the intervention are presented in Table 2. CAB members recommended having 15–20min of group walking after each small group session. Stakeholder interviews confirmed enthusiasm from both senior centers as intervention sites, CBO interns as potential session moderators, and CBO willingness to provide resources for participant recruitment and follow-up. Two barriers identified were the diversity of South Asian languages spoken by potential participants and participants’ lack of experience in engaging in a long-term intervention, as opposed to a one-time workshop.

CONCLUSIONS: Community-based participatory approaches are effective ways to integrate evidence-based interventions that are systematically and culturally consistent into existing community infrastructure. We found this CBPR approach led to concrete and substantive input on content and delivery, while still allowing for the fidelity of the intervention. Our community-academic partnership resulted in interest about the topic of healthy aging, strong enthusiasm and leadership in shaping the culturally-tailored intervention, and excitement about its implementation by the community partners, CAB, and potential participants. Our next step will be to conduct a pilot intervention in one senior center with the goal of increasing walking steps and beliefs that physical activity is an important part of healthy aging that can reduce stroke risk.

Table 1: Cultural adaptation recommendations by CAB for community-based intervention

Table 2: Themes from focus group of South Asian older adults

CURRENT PRACTICES AND OPPORTUNITIES IN A RESIDENT CLINIC REGARDING THE CARE OF OLDER ADULTS WITH MULTIMORBIDITYNancy Schoenborn; Cynthia M. Boyd; Matthew McNabney; Anushree Ray; Danelle Cayea. Johns Hopkins University School of Medicine, Baltimore, MD. (Tracking ID #1936991)

BACKGROUND: Multimorbidity (2 or more chronic conditions) is a common problem that affects more than half of all older adults. Multimorbidity is associated with higher mortality, disability, institutionalization, decreased quality of life, and higher rates of adverse effects from treatments or interventions. Guiding Principles for the care of older adults with multimorbidity were developed and published by the American Geriatrics Society. Clinicians report inadequate training in caring for older adults with multimorbidity but it is not clear what opportunities arise in clinical encounters to apply the Guiding Principles and how clinicians currently practice regarding the Guiding Principles. Our study explores the current practices and opportunities for improvement in a resident clinic regarding the care of older adults with multimorbidity.

METHODS: In this qualitative study, we audio-recorded clinic visits between residents and patients who were ≥65years old and with ≥2 chronic medical conditions in the General Internal Medicine resident clinic at Johns Hopkins Bayview Medical Center. We explored the resident-patient discussions related to the five domains of the Guiding Principles: Patient Preferences; Interpreting the Evidence; Prognosis; Clinical Feasibility; and Optimizing Therapies and Care Plan. The recordings were transcribed, coded and analyzed using the grounded theory approach, generating themes and sub-themes.

RESULTS: We have recorded visits between 17 residents (6 PGY1s, 6 PGY2s, 5 PGY3s) and 26 patients. Each resident had 1–4 recorded visits. Each patient was recorded once. Patients’ mean age was 74.1years, with on average 3.5 comorbidities, 12.7 medications, and mean self-rated health score of 68 on the EuroQol visual analog scale. All visits included discussions that were thematically related to the content of at least one of the Guiding Principles, most often Patient Preferences, Clinical Feasibility, and Optimizing Therapies and Care Plan; discussions related to the other two Guiding Principles were less frequent with very rare discussion related to Prognosis: a) The discussions related to Patient Preferences often included patients’ opinions about decisions such as whether to pursue medication changes, procedures, or specialist referrals, but were not often associated with discussions about the expected benefits and harms of these decisions. b) Discussions related to Interpreting the Evidence occurred infrequently, and did not often include consideration of the applicability of the evidence in the context of multimorbidity. c) The discussions related to Clinical Feasibility often considered treatment complexity and feasibility in the context of medications compared to other treatment modalities. d) The discussions related to Optimizing Therapies and Care Plan more frequently involved specific diseases than the interactions among multiple conditions or their treatments. e) A number of opportunities to incorporate Guiding Principles as prompted by patients or patients’ medical conditions were missed by residents.

CONCLUSIONS: During clinic visits with older patients with multimorbidity at a resident clinic, discussions related to some of the Guiding Principles occurred more frequently than others. Medical content of the visits and patient prompts provided regular opportunity to incorporate the Guiding Principles but were not always acted upon, suggesting room for improvement. We will implement an innovative curriculum to teach how to better incorporate elements of the Guiding Principles into clinical visits.

DANCE YOUR WAY TO BETTER HEALTH: EFFECTS OF TANGO DANCE ON MILD COGNITIVE IMPAIRMENT IN OLDEST OLD ADULTSTeresa Ingram¹; Madeleine E. Hackney². ¹Emory, Atlanta, GA; ²Emory, Atlanta, GA. (Tracking ID #1941542)

BACKGROUND: With the baby boomer generation in the United States, our elderly population will increase exponentially over the course of the next 30years. Many oldest-old adults have sensory, motor, and cognitive impairments leading to reduced independence and social isolation. Not only does cognitive impairment causes decline in quality of life, but it also increases morbidity and mortality risk and carries a huge burden on the healthcare economics. Our purpose is to determine the effects of a 12week course of Tango-adapted dance classes vs. group educational sessions on the cognitive performance in an oldest-old adult population with Mild Cognitive Impairment (MCI).

METHODS: Participants were assigned to 20, 90-min lessons in Tango or Education that took place at six different senior independent living communities. Only one site offered Education; while tango was offered at the other sites. Participants were assessed on the following measures of cognition: Trail Making Test (TMT), with parts A and B, Montreal Cognitive Assessment (MoCA), Timed Up and Go Cognitive, and Brooks Spatial Memory Test prior to and after intervention. Errors and number correct were evaluated.

RESULTS: 74 participants enrolled in the original study (Tango, n = 52; Education, n = 12). For this analysis, criteria for exclusion included anyone under age 80 and completion of less than 5 education or dance sessions which yielded a study population of 41 (Tango, n = 32, Education, n = 9). Within the tango group, there was improvement in Timed Up and Go Cognitive scores (pre-intervention average 19.4, post-intervention average 18.1, change average score of 1.29; p-value 0.07). Timed Up and Go score alone did not improve significantly within the Dance group (change average of 0.75, p-value of 0.37). There was no significant improvement in Timed Up and Go Cognitive scores in the Education group. Alternatively, Brooks spatial scores improved significantly within the Education group (pre-intervention average 59.5, post-intervention average 70.8; p-value 0.007) compared to Dance group (p- value 0.05). There were no statistically significant differences with in MoCA or TMT scores with either intervention.

CONCLUSIONS: Although the Timed Up and Go Cognitive scores within the Dance therapy group were not statistically significant, an improvement within the dual tasking/cognitive portion was evidenced by non-improvement within the motor aspect of the test (Timed and Go). There may be a role for dance and/or exercise in helping to maintain or improve MCI in oldest old adults. It is unclear why there were no improvements within the MOCA or Trail Making Tests which are tests of executive functioning. Further investigation of benefits of educational classes on the effects of spatial memory will be studied. This study is unique because of the focus on adapted Tango as well as having a population group over the age of 80. Limitations include low power, especially in the education group, and significant dropout rate. Also participant knowledge of class assignment could lead to potential bias. Larger population studies that also look into the effect of co-morbities on test performance is warranted.

DEATH AMONG PATIENTS HOSPITALIZED WITH PNEUMONIA; IMPLICATIONS FOR HOSPITAL OUTCOME MEASURESMihaela S. Stefan^1,3; Michael B. Rothberg²; Randa Jaber¹; Jane Garb¹; Janice Fitzgerald¹; Peter K. Lindenauer^1,3. ¹Baystate Medical Center, Springfield, MA; ²Cleveland Clinic, Clevaland, OH; ³Tufts University School of Medicine, Boston, MA. (Tracking ID #1936240)

BACKGROUND: Risk-standardized mortality rates are used to assess the quality of care for patients hospitalized with pneumonia, and were recently added to Medicare’s Value Based Purchasing. The measure includes patients regardless of advanced directives (unless the patient is already in hospice) and those with terminal illnesses for whom pneumonia may play only a minor role in their death. The objective of this study was to assess the proportion of patients included in the CMS pneumonia mortality measure for whom pneumonia was a major contributor to death. In addition, we compared the intensity of care during hospitalization including admission to intensive care unit (ICU) and ventilation management for patients for whom pneumonia was a major contributor to death and those for whom it was not.

METHODS: Retrospective, chart review of adults who died with a principal diagnosis of pneumonia between January 2008 and December 2012 in three hospitals in Western Massachusetts. Determination of pneumonia as a major or minor contributor in the patient death was assessed by two reviewers who used a standardized abstraction form. Results were compared and discussed until the agreement reached 100% (for a total of 57 charts); thereafter charts were assessed independent by each reviewer. Pneumonia was considered to have been a minor determinant if at admission patient had advanced comorbidities which may have been the major contributor to death (e.g. metastatic cancer). Patients’ characteristics, comorbidities, ventilation management and advance directives were collected and compared between the two groups of patients.

RESULTS: We included 204 pneumonia deaths from three hospitals; 169 from a large academic center and 35 from 2 community hospitals. For the first 30 charts reviewed, inter-rater k agreement coefficient was 0.35. Discrepancies between the reviewers were resolved through discussion; agreement then rose to a k of 0.63 for the next 12 cases and to a k of 1.0 for the next 15 cases. The median age of the study population was 78.5years, 46% were male, 98% had at least one major co-morbidity and majority were white (87%). During hospitalization, 24% were intubated and 25% died in the intensive care unit. Pneumonia was considered to have played a minor role in patients’ death in 167 (80%) cases. Overall, 51% of cases had a do-not-resuscitate order (DNR) at admission and 25% at the time of death. Of those who had their code status changed during hospitalization, the change to DNR or comfort care was made on the day of death in 41% of the cases.. Compared with patients for whom pneumonia was a major reason to death, patients for whom pneumonia was a minor contributor to death were slightly younger, had more advanced comorbidities (100% vs 86%); were less likely to be admitted to ICU (28% vs 42) and to be mechanically ventilated (22% vs. 32%); were less likely to have mechanical ventilator withdrawn before death (78% vs. 92%) and were more likely to expire on the medical ward (76% vs. 53%).

CONCLUSIONS: Only a small fraction of deaths included in the pneumonia mortality measure are the direct result of pneumonia. The majority of patients who die are frail elderly with advanced non-modifiable conditions, most of whom decide to forego aggressive care at some point during their admission. The deaths of these patients do not necessarily reflect the quality of pneumonia care received. Only half of these patients had DNR orders on admission, and many initially received aggressive care, suggesting opportunities for discussion of end of life care before hospitalization.

DECADE-LONG TRENDS IN MORTALITY AMONG PATIENTS WITH AND WITHOUT DIABETES AT AN ACADEMIC MEDICAL CENTER IN AN ERA OF IMPROVED GLYCEMIC CONTROLNeel M. Butala¹; Benjamin K. Johnson³; James F. Dziura^3,1; Janis E. Bozzo²; Silvio Inzucchi¹; Thomas Balcezak^1,2; Leora I. Horwitz^1,2. ¹Yale School of Medicine, New Haven, CT; ²Yale-New Haven Hospital, New Haven, CT; ³Yale Center for Analyltic Sciences, New Haven, CT. (Tracking ID #1937226)

BACKGROUND: Given the increasing prevalence of diabetes, historically worse outcomes for patients with diabetes when hospitalized, and recent trends toward improving glycemic control in both inpatient and outpatient settings, it is important to understand whether the mortality risk of diabetes during hospitalization has changed over the last decade. The objective of this study was to determine 10-year trends in in-hospital mortality for patients with versus without diabetes and to examine the potential impact of inpatient and outpatient glucose control.

METHODS: We conducted a serial cross-sectional observational study of all adult non-obstetric patients with an inpatient admission between Jan 1, 2000 and Dec 31, 2010 at a major urban medical center. The primary exposure of interest was a diagnosis of diabetes mellitus and the main outcome was in-hospital mortality. Additionally, we looked at measures of inpatient glycemic control (percentage of days with glucose below 70, glucose above 299, and glucose between 70 and 179, and standard deviation of glucose measurements) and hemoglobin A1c, which served as a measure of outpatient glucose control and a proxy for severity of diabetes. We constructed multivariate logistic regression models including time from start of study until admission in years, whether an individual had diabetes, and the interaction between time and diabetes status as well as age, sex, race, payer, diagnosis and comorbidities while stratifying by whether the hospitalization included an intensive care unit (ICU) stay. We then restricted our sample to ICU patients with diabetes and determined the relation between the measures of glucose control and changes in mortality over time using logistic regression.

RESULTS: We included 322,938 patient admissions, of which 54,645 (16.9%) were ICU patients. 76,758 patients (23.8%) had diabetes: 26.3% of ICU patients and 23.2% of non-ICU patients. Among ICU patients with diabetes, measures of inpatient control were available for 91.5% and hemoglobin A1c for 34.7%. The overall mortality rate was 2.4%, representing 2.7% of patients with diabetes and 2.2% of patients without diabetes. However, raw mortality for patients with diabetes decreased from 3.6% in 2000 to 2.2% in 2010 while mortality for patients without diabetes decreased from 2.4 to 2.3% during this interval. Among ICU patients, the 7.8% reduction in the odds of mortality in each successive year for patients with diabetes (OR = 0.923; 95% CI: 0.906, 0.940) was significantly greater than the 2.6% yearly reduction for those without diabetes (OR = 0.974; 0.963, 0.985; p < 0.001 for interaction). However, among non-ICU patients, the 9.6% reduction in odds of mortality for patients with diabetes (OR = 0.904; 0.879, 0.929) was not significantly different from the 7.5% reduction for those without diabetes (OR = 0.925; 0.909, 0.940). Among ICU patients with diabetes, a significant decline in mortality persisted after adjustment for measures of inpatient and outpatient glucose control.

CONCLUSIONS: Patients with diabetes in the ICU have experienced a disproportionate reduction in in-hospital mortality that cannot be explained by the quality of inpatient or outpatient glucose control. While improved glycemic control may have other benefits, it does not appear to impact in-hospital mortality. Potential explanations for this interesting trend include improved cardiovascular risk management or advances in therapies for diseases that disproportionately affect diabetics (e.g. coronary stents).

DEEP VEIN THROMBOSIS BELOW THE KNEE: DO WE TREAT AND SHOULD WE TREAT?Anita Pudusseri; Raji Shameem; Robert E. Graham. Lenox Hill Hospital, New York, NY. (Tracking ID #1910852)

BACKGROUND: Below knee DVT is rarely associated with progression to pulmonary embolism (PE). However, patients often receive treatment with novel oral anti-coagulants, with a notable consequence of anti-coagulation being increased risk of bleeding. The current Guidelines for Anti-Thrombotic therapy from Chest 2012 (based upon observational studies and expert opinion) recommend that patients with acute isolated distal DVT of the leg and severe symptoms or risk factors for extension require initial anticoagulation over serial imaging of the deep veins.

METHODS: This was a retrospective review of patients that presented to a community hospital from June 1, 2012 to June 30, 2012 requiring a lower extremity venous doppler for any reason. Information was gathered using the Synapse database. The location of the DVT, which exact vein, the reason for the study, which department, whether the patient was treated and with what form of treatment were all collected. Patients already on anti-coagulation at the time of diagnosis were excluded, including but not limited to atrial fibrillation, stroke prevention, pulmonary embolism, and acute coronary syndrome.

RESULTS: Forty-five patients with lower extremity DVT were included. Of the 45 patients, 28 patients were from the medicine department, 3 from orthopedics, 2 from cardiology and surgery, 1 from obstetrics and 9 from other departments. 80% (36/45) of patients were treated with anti-coagulation for DVT. Most patients were treated with heparin to warfarin bridge (36%; 13/36) or enoxaparin to warfarin bridge (42%; 15/36). Five patients were treated with enoxaparin alone, one patient was treated with heparin to fondaparinux sodium bridge and three patients were treated with warfarin alone. Of the three patients that were not treated with anti-coagulation, one patient had a known DVT and had repeat ultrasound imaging. The other two patients had intramuscular vein thrombosis and thus were not treated. Two of the 45 patients had adverse events of clinically significant bleeding.

CONCLUSIONS: A large majority of our patients were treated with anti-coagulation for a below knee DVT despite the known low risk of progression to PE. Given this data we plan on conducting a prospective study examining the management of a new diagnosis of below knee DVT. One of the outcomes studied will be to determine the extension of DVTs over a period of time using serial imaging.

DEFICITS IN THE TRANSFER OF ADVANCE CARE PLANNING INFORMATION FOR PATIENTS DISCHARGED FROM HOSPITALS TO NURSING HOMESPeggy Leung^2,1; Clarke Low¹; Anne M. Walling^1,2; Neil Wenger^1,2. ¹University of California, Los Angeles, Los Angeles, CA; ²David Geffen School of Medicine at University of California, Los Angeles, Los Angeles, CA. (Tracking ID #1939638)

BACKGROUND: Prognostic discussions and decisions about aggressiveness of care made in the hospital are often lost when patients transition to nursing facilities. For patients with serious illness and particularly for those for whom decisions have been made not to use cardiopulmonary resuscitation (CPR) and other burdensome treatments, Physician Orders for Life-Sustaining Treatment (POLST) is a legal document that translates patient end-of-life treatment preferences into actionable medical orders that are preserved across venues of care. This tool is widely used in hospitals and nursing homes across California. We aimed to investigate how often patients with do not resuscitate (DNR) orders were discharged from the hospital to a nursing facility without adequate transmission of this information across facilities.

METHODS: We identified all adult discharges from two hospitals to a skilled nursing facility over a three month period in 2013. We developed a medical record abstraction instrument that identified a patient’s resuscitation status during hospitalization and at discharge and whether this was communicated to the nursing home in the discharge summary and whether a POLST was completed. In cases where there was discrepancy between the resuscitation status at discharge and the information communicated, we explored the reasons for the difference.

RESULTS: Of 461 patient discharges (449 unique patients), 347 (75.3%) patients were “full code” at time of transfer, 89 (19.3%) patients had a DNR status at the time of transfer, and for 25 (5.4%) patients there was no “code status” documented on transfer. Of the 347 patients who were full code, 65 (18.7%) had a POLST indicating this at discharge. Of the 89 patients who had a DNR order at the time of transfer, 29 (32.7%) had a valid POLST on transfer that contained this information. An additional 23 patients had POLST forms on transfer that were inconsistent with resuscitation decisions at discharge: 13 patients with a “full code” status had a POLST that indicated DNR, five patients with a DNR status on transfer had a POLST that indicated CPR, and five patients had POLST forms with other content inconsistent with decisions made on discharge.

CONCLUSIONS: At two hospitals in one health system, detailed evaluation of medical records for patients discharged to nursing homes demonstrated that even among patients for whom a DNR order was in place at time of transfer, less than one third had POLST documentation of this preference. In the studied hospitals, efforts are needed to improve the transition of resuscitation information from hospital to nursing home.

DEFINING APPROPRIATE USE OF ACID SUPPRESSION AMONG MEDICAL INPATIENTSMatt Pappas¹; Sanjay Jolly²; Sandeep Vijan¹. ¹University of Michigan Health System, Ann Arbor, MI; ²University of Michigan, Ann Arbor, MI. (Tracking ID #1936603)

BACKGROUND: Proton pump inhibitors (PPIs) are commonly used among medical inpatients, with some studies suggesting that as many as half of all inpatients receive PPIs. Routine use of PPIs is common for symptom relief and, in higher risk patients, to reduce the risk of upper GI bleeding (UGIB). Recent evidence has suggested that PPIs increase the risk of nosocomial infections, including healthcare associated pneumonia (HCAP) and Clostridium difficile infection (CDI). We sought to examine the balance between risks and benefits of PPI use among medical inpatients in various categories of risks of UGIB, HCAP, and CDI.

METHODS: We created a microsimulation model to estimate the risks of UGIB, HCAP, and CDI among medical inpatients. Risks of each of these conditions were generated using published risk models. We then modeled changes in risk associated with PPI use for each of these outcomes, basing our estimates on meta-analyses and observational studies. We examined the overall impact of PPI use on inpatient mortality, including both population level risk and individual thresholds at which PPI use would be beneficial (through reduction in UGIB risk) or harmful (through increases in HCAP and CDI). We examined several scenarios to account for heterogeneity in the strength of the evidence for the causal relationship between PPIs and our modeled outcomes, and conducted both one-way and multivariate sensitivity analyses (using second order Monte Carlo simulation) across all parameters in the model.

RESULTS: In our first scenario, where PPI use was causally linked with changes in risks of all three outcomes (UGIB, HCAP, and CDI), we found that PPI use led to an increase in mortality in over 95% of hospitalized patients. In our second scenario, with no causal association between PPI use and CDI, we found that PPI use led to an increase in mortality in 86% of hospitalized patients. Sensitivity analyses showed that even with a low odds ratio for HCAP, PPI use led to increased mortality in 91% of cases (assuming a causal link with CDI) or 68% of cases (no causal link with CDI). We then examined percentiles of probabilities for HCAP, CDI, and UGIB to see if we could identify particular cases where benefit from PPI use was likely. However, only at extremes of the distribution were we able to identify patients who may benefit. For example, we modeled a patient at the 10th percentile for probability of HCAP, case fatality of HCAP, risk of CDI, and case fatality of CDI, but at the 90th percentile risk of UGIB risk and fatality. In such a patient, PPIs were neutral, with benefit in about half of such patients and harm in the other half.

CONCLUSIONS: Our study suggests that for the vast majority of medical inpatients, use of PPIs likely leads to an increase in short-term mortality. Even in patients at particularly high risk of UGIB, only those at the very lowest risk of HCAP and CDI should be considered for prophylactic PPI use. Indeed, our estimates suggest that withholding PPI therapy should be considered for most patients upon admission, with the exception of those who are at very high risk of or are hospitalized for UGIB. There are several limitations to our analysis. First, we are reliant on published estimates of the effect of PPI on HCAP, CDI, and UGIB, and for some outcomes, particularly CDI, there remains some controversy on the causality of the observed association. Second, we did not model additional outcomes such as length of stay; however, given the balance of risks and benefits we found, this is more likely to accentuate than alter our finding that PPIs generally lead to harm. Finally, we did not model specific settings such as the ICU or surgical patients. Overall, our findings suggest that PPI use should be avoided in the vast majority of medical inpatients.

Density plot of change in mortality for 100,000 simulations of 100-patient cohorts, assuming increased risk of CDI with acid suppression (red) and no increased risk of CDI (blue).

DELAYS IN ENTERING ABNORMAL TEMPERATURES INTO THE ELECTRONIC MEDICAL RECORDAlvin Rajkomar; S. Ryan R. Greysen. University of California, San Francisco, San Francisco, CA. (Tracking ID #1939063)

BACKGROUND: Newly abnormal vital signs in hospitalized patients often spur additional diagnostic testing and treatment, like evaluation for sepsis in a patient with a fever. Delays between measuring and recording temperatures in the electronic medical record (EMR), functions often performed by nursing staff, can delay timely recognition of conditions like sepsis, especially as clinicians and electronic surveillance tools use EMR data for monitoring. We sought to characterize the frequency and duration of these delays of entering temperature data into the EMR.

METHODS: We identified all temperature readings from adult inpatients admitted to UCSF Medical Center from June 1, 2012 to Dec 5, 2013. For each temperature reading, the location of the patient and the identity of the nurse who took the temperature were obtained. The delay was calculated by the difference between the computer-generated timestamp when the temperature was entered into the EMR and the nurse-generated time of when the vital sign was physically taken. Fever was defined as a temperature greater or equal to 100.4° Fahrenheit. We used descriptive statistics and t-tests to analyze trends in entering temperature values.

RESULTS: There were over 1.3 million temperature readings throughout the study period. The mean delay was 42.6min for all readings (range: 0 to 720min, standard deviation: 75.8min). There were 66,434 readings with a fever, and the mean delay was statistically lower at 36.6min (p < 0.001). Interestingly, the delay was higher for patients with a new versus a persistent fever: of the 19,619 readings of a new fever, defined as the prior recorded temperature being below 100.4, the mean entry time was 42.3min. Of the 46,815 readings of a persistent fever, defined as the prior temperature also indicating a fever, the mean delay was lower at 34.2min. The delay was not only different between different nursing units but also between different nurses within a unit (Figure 1).

CONCLUSIONS: The average delay between obtaining a temperature and entering it into the EMR is greater than 40min for newly febrile patients, and there is significant variation in delays between nursing units and individual nurses. Hospitals must consider new ways to promote timely entry of data into the EMR of they hope to capitalize on the advent of new sensors and the promise of using big data for real-time electronic surveillance of hospitalized patients.

Figure 1: Median Delay in Entering Temperature into the EMR Each large, numbered rectangle represents a nursing unit, and every colored box represents an individual nurse within that unit. The color of the box is proportional to the median delay in entering temperature data into the EMR by that individual nurse. White indicates a median delay of zero minutes and red indicates a median delay of 60min or more.

DELIVERING VIDEO PATIENT EDUCATION FROM THE EHR: PROMISES AND PITFALLSShira N. Goldman¹; Tiffany Brown¹; Stephen D. Persell¹; Alpa Patel²; Crystal T. Doan¹; Kenzie A. Cameron¹. ¹Northwestern University, Chicago, IL; ²Northwestern Medical Faculty Foundation, Chicago, IL. (Tracking ID #1926289)

BACKGROUND: Alerts triggered by content in patients’ electronic health records (EHR) are used regularly to prompt providers to improve the quality of care. These alerts can be leveraged to deliver educational content to specific patients at the point of care. We implemented an alert for licensed practical nurses (LPNs) and medical assistants (MAs), prompting them to show an educational video to patients eligible for pneumococcal vaccination (PnVx) while patients waited for their primary care provider (PCP) to enter the exam room. We report on lessons learned from this implementation.

METHODS: The PnVx educational video Best Practice Alert (BPA) was implemented in a large general internal medicine practice at an academic medical center using the Epic EHR. We designed and built the alert using a team of computer programmers, technical support specialists, and clinical and communication experts. We validated the decision rules to ensure the alert only triggered for patients eligible for PnVx who had no documentation of vaccination receipt or refusal in the EHR. PCPs provided consent for their patient panels to be included in this feasibility study. We trained LPNs and MAs to launch the video from within the EHR when the alert appeared, and the PnVx BPA was then activated practice-wide. After 3months, nurses were invited to participate in individual interviews and PCPs were sent a web-based survey. Patients were asked to participate in a telephone survey to assess their perceptions toward being shown a video at the point-of-care, as well as their thoughts on pneumonia, pneumococcal vaccination (on a Likert scale where 1 = Strongly Disagree and 5 = Strongly Agree).

RESULTS: We encountered technical barriers to successful implementation. These included hardware problems such as the need for uniform monitor resolution and lack of sound capability, and software problems, including failure of the video to function across multiple operating systems. Additionally, to retain the security of Epic and account for the fact that LPNs and MAs had varied workflows within the EHR, the BPA required several decision points, which corresponded to mouse clicks. Upon completion of patient intake, the LPN/MA would secure the patient’s record and launch the video in an external browser. Despite these challenges, overall feedback to the alert was positive. Patients (n = 32) perceived the video as easy to understand (M = 4.71, sd = 0.81) and informative (M = 4.46, sd = 0.58). PCPs (n = 30) did not perceive a negative impact on their workflow, nor did they believe the addition of the alert and video delayed the rooming process. LPNs and MAs (n = 10) had varying responses to the addition of this activity to their workflow; some perceived it to be a minor task benefitting patients, while others reported that it was overly complicated and bothersome.

CONCLUSIONS: At a site using multiple provider-directed alerts, implementing a clinical decision support-triggered patient education video identified unanticipated obstacles. While many of the technical challenges were surmountable, encouraging nurses to act on the alert remains challenging. Challenges include the fact that numerous steps are required to launch the video and the absence of a clinic-wide incentive or policy supporting this effort. Although many nurses responded positively, others did not perceive point-of-care patient education as essential for improving patient care. Further research is needed to determine the best ways to involve the entire medical team in patient care and assess the impact of this point-of-care intervention on uptake of vaccination and other preventive health services.

DENTAL REFERRALS IN A FEDERALLY QUALIFIED HEALTH CENTER: THE LINK BETWEEN ORAL AND SYSTEMIC HEALTHLanna Little; James F. Hanley; Garry Souffrant. University of Texas Health Science Center San Antonio, San Antonio, TX. (Tracking ID #1939712)

BACKGROUND: Oral health and systemic health are intricately connected. The Federally Qualified Health Centers (FQHCs) are uniquely positioned to treat both oral and systemic health as they provide comprehensive medical and dental care. The path for most patients to see the dentist is a referral from a primary care physician. Although the dental referral may be an important link between medical and dental health care, most physicians have little insight on how to optimize this process. Little research has been conducted to characterize the referral process and referral population.

METHODS: Chart review was performed, identifying 420 adult patients who have been seen in the past two years at the FQHC and have received a referral to the dentist. Comorbidities and demographics were recorded for each subject, as was information regarding the dental referral. Data was analyzed using SPSS.

RESULTS: The most common reason for referral was pain, which was cited in 58% of referrals. In about 40% of dental referral patients, some sort of dental emergency was reported (pain, abscessed tooth or loose tooth). About one third of patients indicated that seeking dental referral was their primary reason for coming to the clinic. Most patients received only the referral but no treatment by the physician. Only half saw the dentist and half of those were diagnosed with some degree of periodontal disease. Patients with acute complaints such as dental pain or abscess were more likely to see the dentist than those referred for diabetes mellitus. Out of everyone who sought dental referral, 10% were subsequently screened for and diagnosed with a new condition (diabetes or dyslipidemia).

CONCLUSIONS: Although dental referrals were common in our FQHC and dental pathology was frequent, most physicians provided little treatment prior to the dental visit. Most physicians have both limited experiences and limited knowledge of the management of dental urgencies/emergencies. Implementation of inter-professional education in medical school or primary care residencies could equip physicians to treat the patient properly until they see a dentist, and could also facilitate the referral process. It may be appropriate for FQHCs to develop inter-professional guidelines for physicians and dentists concerning the management of these patients. While dental care is often considered discretionary, urgent dental problems significantly impact patients’ quality of life and poor dental hygiene may impact management of many chronic diseases. Having periodontal disease treated may positively affect these patients’ systemic health. This is especially important for patients with diabetes mellitus. Lastly, it is important to take dental pain patients seriously and use their visit as an opportunity to perform the indicated health screenings.

DEPRESSION IS ASSOCIATED WITH CLINICAL INERTIA IN MANAGEMENT OF HYPERTENSION IN THE PRIMARY CARE SETTINGNathalie Moise¹; Karina Davison^1,2; William Chaplin^1,3; Steven Shea^1,4; Ian M. Kronish¹. ¹Columbia University Medical Center, New York, NY; ²Columbia University Medical Center, New York, NY; ³St. John’s University, Queens, NY; ⁴Columbia University, New York, NY. (Tracking ID #1926816)

BACKGROUND: Depression is a known risk factor for poor prognosis among patients with cardiovascular disease. Numerous biological and behavioral mechanisms have been proposed. However, few studies have investigated the association between depression and ‘clinical inertia’, or lack of treatment intensification in individuals not at evidence-based goals for care. To address this gap, we assessed whether a diagnosis of depression is associated with clinical inertia in patients with uncontrolled hypertension.

METHODS: A convenience sample of 28 attending primary care providers (PCP) and 158 patients with uncontrolled hypertension was enrolled from two inner-city, academic-hospital based clinics. Eligibility criteria were age ≥18years, prescribed ≥1 blood pressure (BP) medications, and BP ≥140/90mmHg (or ≥130/80mmHg if diabetic or with chronic kidney disease) on at least two consecutive scheduled visits with one’s PCP. Clinical inertia was defined as lack of (1) medication intensification (2) specialist referral or (3) secondary hypertension work-up. Depression diagnosis was based on PCP documentation in the electronic chart. Multilevel modeling accounting for clustering of participants within PCP was conducted to assess the risk of clinical inertia in depressed patients while adjusting for known predictors of clinical inertia. Sensitivity analyses were performed that (1) excluded visits with documentation of at least one controlled clinic or home BP (2) accounted for medication adherence counseling by PCP and (3) excluded diabetics with systolic BP (SBP) ≤140mmHg.

RESULTS: The mean age (SD) of patients was 64.5 (8.8) years; 74.1% were women, 79.1% were Hispanic, 44.9% was diagnosed with depression and 61.2% had diabetes. On average, participants had a prior visit SBP of 158.7 (15.7) mmHg, current visit SBP of 154.6 (16.7) mmHg, were taking 2.5 (1.1) BP medications and had 5.3 (2.3) problems addressed during the visit. Clinical inertia was more common amongst depressed than non-depressed patients (70% vs. 51%; p = 0.015). Depression diagnosis was associated with clinical inertia in both the adjusted and unadjusted multilevel analyses (RR = 1.40; 95%CI, 1.11–1.74; p = 0.004; adjusted relative risk [ARR] = 1.49; 95%CI, 1.06–2.10; p = 0.021). The relationship remained after excluding those with at least one documented home or clinic visit SBP below goal (ARR = 1.74; 95%CI, 1.07–2.83; p = 0.025), adjusting for adherence counseling (ARR = 1.49; 95% CI, 1.10–2.04; p = 0.010) and excluding diabetics with SBP ≤ 140mmHg (ARR = 1.49; 95% CI, 0.99–2.23 p = 0.057).

CONCLUSIONS: Amongst patients with uncontrolled hypertension, clinical inertia was more likely in those with a diagnosis of depression; this may be one mechanism by which depressed patients have worse cardiovascular outcomes. Research has shown that patients with mental illness receive less intensive medical care, such as coronary revascularization; our study extends this literature by demonstrating differences in clinician behavior with respect to cardiovascular risk factor management in this population. Future studies should explore underlying processes affecting clinician treatment practices. In the meantime, PCPs should be cautious about undertreating cardiovascular risk factors among patients identified as having depression.

DERIVATION OF ICD-10 CODE LISTS FOR NOVEL PATIENT SAFETY INDICATORSWilliam A. Ghali; Danielle Southern; Hude Quan. University of Calgary, Calgary, AB, Canada. (Tracking ID #1896442)

BACKGROUND: Health data coded using the International Classification of Diseases have been widely used to identify the occurrence of adverse events that may be linked to suboptimal safety and quality of care. Some countries record ‘diagnosis timing indicators’, which allow data users to distinguish whether a coded diagnosis or event was present at time of admission as opposed to one that occurred after admission. Using this diagnosis-timing data element, we undertook a multistep process to produce novel patient safety indicators (PSIs).

METHODS: We queried 2,416,413 records in the Canadian national Discharge Abstract Database (DAD) for April 1, 2009 through March 31, 2010. All listed diagnosis codes were compiled, and 2,613 unique diagnoses were found to have a type ‘2’ diagnosis designation, indicating that the diagnosis was not present at time of admission and thus arose after admission. Next, we asked an international panel of experts in patient safety and quality of care to participate in a Modified Delphi review process in which they undertook a two-phase review of all of the identified codes, to determine (using a 9-point scale) the subset of post-admission diagnoses that are highly likely to have a strong link quality of care and/or patient safety. The subset of identified codes were then grouped into clinically-related subgroups, and the frequency of events was determined in a database of all hospitalizations records from the province of Alberta, Canada.

RESULTS: Of the 2,613 diagnosis codes recorded at least once in Canadian data as a diagnosis arising after admission, 640 codes were judged by the expert panel to have a strong link to quality of care and patient safety. These codes were then reviewed by a second group of experts in patient safety to create summary PSI code groupings. These included: a single global PSI for ‘any adverse event’; a smaller grouping of codes for ‘serious adverse event threatening to life or vital organs’; and then a number (n = 17) of other specific code groupings for specific types of events such as ‘cardiac complications, ‘respiratory complications, ‘hospital-acquired infections’, and ‘venous thromboembolic events’. The annual event rates in Alberta provincial data for the global PSI of ‘any adverse event’ ranged from 4.5% in 2005 to 5.6% in 2011. The event rate for the ‘serious adverse event’ indicator was 0.8%. The event rates for the other specific indicators ranged from a high of 2.0% for ‘Hemorrhagic Events’ to a low of 0.03% for the ‘obstetrical complications affecting fetus’ PSI.

CONCLUSIONS: The methodological work presented here utilizes the unique potential of diagnosis-timing indicators to produce a clinically-relevant listing of diagnosis codes that can be used to create novel PSIs that overcome some of the shortcomings of existing PSI systems. The resulting work has great potential to inform future approaches to health system monitoring, adverse event surveillance, and quality/safety improvement.

DESCRIBING THE STRUCTURE AND PATTERNS IN THE PHYSICIAN-PATIENT INTERACTIONHarry B. Burke¹; Dorothy Becher¹; Patrick G. O’Malley¹; Jeffrey L. Jackson²; Ronald W. Gimbel¹. ¹Uniformed Services University of the Health Sciences, Bethesda, MD; ²Zablocki VAMC, Milwaukee, WI. (Tracking ID #1937588)

BACKGROUND: Although there is an extensive literature describing the semantics of physician-patient interactions, i.e., judgments regarding the meaning of participants’ words and individual utterances, little is known about the interactional structures that constitute the physician-patient encounter. Structural units are the categories of verbal interactions that the participants use to achieve their goals (categories are described below). We hypothesized that these structural units would occur in a systematic manner across interactions.

METHODS: The unit of analysis was not the word or the utterance; it was the structural unit. Twenty audio-recorded established (rather than new) patient primary care outpatient physician-patient interactions were coded into structural units in terms of their interactional categories. A structural unit is composed of adjacent sequential utterances of the physician and patient that, taken together, meet one of the eight defined structural categories. The same type of structural unit could occur at various times throughout the interaction and at each occurrence the unit was coded and its location noted. Eight categories of interactional units were sufficient to describe the interactions: NEW, discussing a new medical problem; EXT, discussing an existing medical problem; WEL, discussing wellness and screening; REV, reviewing what had happened to the patient since the last visit including labs and medications; ADM performing administrative tasks such as ordering tests and prescriptions and filling our forms; SUM, summarizing the interaction; AGR, participant agreement; and SOC, discussing social topics.

RESULTS: We coded all pairs of adjacent units. There were 462 pairs across the 20 interactions. The inter-rater category agreement was 94%. For most pairs we set a threshold of occurrence at 10%. This analysis focused on the NEW and EXT units. The table below shows the relative frequency of the adjacent unit pairs. Relative frequency is the pair frequency divided by the overall frequency of the target unit. In terms of all the interactional pairs, NEW occurred in 10% and EXT occurred in 27% of the pairs. The arrow designates the order of occurrence of the adjacent units. PRIOR TO NEW.....Relative freq.....SUBSEQUENT TO NEW......Relative freq. NEW ⇒ NEW...........0.06..................NEW ⇒ NEW........................0.06 EXT ⇒ NEW............0.15..................NEW ⇒ EXT..........................0.29 REV ⇒ NEW............0.33..................NEW ⇒ EXT..........................0.29 ADM ⇒ NEW...........0.10 .................NEW ⇒ ADM........................ 0.08 SOC ⇒ NEW...........0.21..................NEW ⇒ SOC.........................0.20 PRIOR TO EXT.....Relative freq.....SUBSEQUENT TO EXT.....Relative freq NEW ⇒ EXT.......... 0.11.................EXT ⇒ NEW.......................0.06 EXT ⇒ EXT............0.34.................EXT ⇒ EXT.........................0.34 REV ⇒ EXT............0.21.................EXT ⇒ REV........................0.21 ADM ⇒ EXT...........0.16.................EXT ⇒ ADM.......................0.23 SOC ⇒ EXT...........0.14.................EXT ⇒ SPC........................0.10 It was rare for a new medical problem to be adjacent to a new medical problem (NEW ⇒ NEW = 0.06), but common for an existing medical problem to be adjacent to an existing medical problem (EXT ⇒ EXT = 0.34). Reviews (REV) occurred with the same frequency before and after new problems (NEW), 33%, 29% and existing problems (EXT), 21%, 21%. Likewise, social topics (SOC) occurred with the same frequency before and after new problems (NEW), 21%, 20% and existing problems (EXT), 14%, 10%. There was an inverse relationship between new and existing problems in terms of the occurrence of adjacent reviews and social topics.

CONCLUSIONS: Structural units can describe physician-patient interactions. They appear to occur in a systematic manner and they can assist us in understanding of how physicians and patients interact to achieve their goals for the interaction.

DETERMINANTS OF HOSPITAL READMISSIONS AMONG LOW INCOME AND MINORITY OLDER ADULTSUgochi K. Ohuabunwa^1,2; Jonathan Flacker^1,2; Queenie E. Jordan². ¹Emory University, Atlanta, GA; ²Grady Hospital, Atlanta, GA. (Tracking ID #1942171)

BACKGROUND: Low income and minority seniors constitute a vulnerable group during care transitions. Determinants of failure in transitions of care and rehospitalization in minority and low income populations have not been well defined. Understanding these determinants and consideration of the specific needs of this group of patients is essential in designing a transitional care model that will be effective in this patient population.

METHODS: We evaluated Medicare beneficiaries who were frequently hospitalized at our 953 bed safety net hospital from January to December 2010 by secondary data analysis of hospital administrative data to determine risk factors associated with repeated hospitalizations.

RESULTS: A total of 3790 patients were admitted during this period accounting for 4933 admissions. We classified all patients (n = 189) admitted 2 or more times in a 6month period as high risk. Our unadjusted all cause readmission rate was 25%. Factors found to be significantly associated with readmission were female gender (60%), African American race (94%), age >75year, patients residing in the Zip Code areas 30318, 30310 and 30311. The most common presenting diagnoses were congestive heart failure, psychoses, chronic obstructive pulmonary disease, septicemia and renal failure.

CONCLUSIONS: Definition of these determinants will form a basis for further needs assessment and the design of a transitional care model that will be effective in this patient population.

DETERMINING THE ACADEMIC HOSPITALIST SAFE CENSUS: RESULTS FROM A PHYSICIAN SURVEY AND HOSPITALIST SERVICE ANALYSISAmy Baughman^2,1; Warren Chuang¹; Denisa Gace¹; Daniel P. Hunt¹. ¹Massachusetts General Hospital, Boston, MA; ²Brigham and Women’s Hospital, Boston, MA. (Tracking ID #1940088)

BACKGROUND: Finding the appropriate patient-to-physician ratio is critical for optimizing patient care and physician satisfaction and productivity. Current literature in this area to guide hospitalist models of care is limited. However, it is well established that excessive patient responsibilities and work load can result in worse patient outcomes as well as increased physician burnout and dissatisfaction. Our hospitalist group has grown considerably which enabled our group to study the impact of higher census. In 2009, our service cared for 22 to 68 patients each day; by 2012, we cared for 54 to 127 patients. Despite hiring more staff, the individual physician census also increased. This challenged even our most experienced hospitalists to round on patients and complete discharges and admissions within their twelve-hour shift. When establishing a safe census, programs must consider regionalization of patients, ancillary staff or resident availability, case-mix index and physician characteristics. On our non-regionalized service, an individual hospitalist cares for patients scattered across several floors and buildings. On our regionalized service, however, the complexity of patients is higher. All of our physicians work without ancillary staff such as nurse practitioners or residents and significant physician time is required for administrative tasks, consultant and patient communication, and EHR documentation. Our goal was to determine how many patients a hospitalist physician at a tertiary academic medical center can safely care for during a typical 12h shift.

METHODS: We describe a survey based method for identifying an optimal and safe census among hospitalists at our tertiary academic medical center. This assessment occurred in 2012 as a quality improvement initiative to gather feedback in several areas including optimal work flow and physician satisfaction. We also separately analyzed the impact of physician census on in-patient quality metrics including hospital readmission rates, length of stay and pre-noon discharges from 2010 to 2013. We then evaluated whether results from the physician survey were consistent with our quality metric outcomes.

RESULTS: Survey results showed that patient census was the most important factor contributing to physician satisfaction (n = 35, Response Rate = 85%). In that year, daily physician census varied from a low of six to a high of 14. We found that 92% of physicians felt that the ideal patient-to-physician ratio was 10 or less for providing the highest quality of care. For half of physicians, the maximum patient number for safe care was 12; however the other half reported lower values of 10 or 11 patients. We reviewed monthly readmission rates, length of stay and pre-noon discharge rates from 2010 to 2013 for our hospitalist group. During this time, the monthly physician census ranged from 8.6 to 11.8 patients. We found that the readmission rate was slightly higher when the physician census was more than 11 (P value < 0.05). Average length of stay and the percentage of pre-noon discharges did not change significantly with higher physician census.

CONCLUSIONS: Surveying physicians is a simple and effective method for identifying safe patient-to-physician ratios. For hospitalists at large tertiary medical centers with similar patient complexity, a safe census of ten is ideal for providing the highest quality of in-patient care. Evaluation of readmission rates, a crude measure of quality, supported this finding as readmission rates increased when census increased to more than 11.

DEVELOPING PATIENT-CENTERED DEFINITIONS AND RATINGS OF ENABLING SERVICESRobin Clarke¹; Anne Escaron¹; Rosy Chang Weir²; Petra Stanton³. ¹University of California, Los Angeles, Los Angeles, CA; ²Association of Asian Pacific Community Health Organizations, Oakland, CA; ³California Primary Care Association, Sacramento, CA. (Tracking ID #1938362)

BACKGROUND: Community Health Centers (CHCs) are safety net practices that have a long history of delivering non-medical or “enabling” services intended to reduce the social barriers to care faced by their low-income patients. Enabling services include providing transportation, performing community outreach, and delivering case management. The implementation of the Affordable Care Act offers new opportunities to reimburse these patient-centered services, but limited evidence exists to guide CHCs and other providers in deploying effective services. We sought to begin to address this evidence gap by developing a research framework for analyzing this field of health care services and by starting the initial evaluation of these services.

METHODS: We adapted the RAND/UCLA modified Delphi method typically used to assess appropriateness of medical interventions. Using a community-partnered research team, we had three phases to our project. In the first phase, we used a literature review and an in-person meeting of 17 CHC patients and providers to define broad categories of services, discrete enabling services, and the variable levels of intensity by which a service can be delivered. An example of an intensity variable is: Health education delivered by a licensed personnel vs. by a non-licensed community member. In the second phase, we convened an expert panel composed of 13 CHC patients, providers, community patient advocates, and policy-makers. This panel used the validated RAND/UCLA 1–9 scale and iterative rounds of ratings to rate each enabling service as ineffective, equivocally effective, or effective. Effectiveness was defined as the degree by which a service increases a patient’s access to, use of, and understanding of his or her medical care. In the last phase, we analyzed the ratings to identify the panel’s recommendations by calculating each service’s median effectiveness rating and the level of agreement or disagreement amongst the panelists. We used a Friedman test and a subsequent Wilcoxson Signed Rank test to determine whether effectiveness varied based on level of intensity.

RESULTS: The literature search identified 40 peer-reviewed articles that tested the effectiveness of enabling services. Based on these articles and their own knowledge, the patients and providers in the first group defined six categories: health education & supportive counseling; case management; outreach; interpretation; transportation; and financial counseling & eligibility assistance. Within these categories, 112 unique services were described, of which the expert panel elected to rate 77. Each of these services had between 1 and 4 levels of intensity at which it could be delivered resulting in a total of 181 unique service-variable combinations being rated by the expert panel. Health education & supportive counseling had the lowest median effectiveness rating as a category at 6.6 (out of 9) and outreach the highest at 7.5. Within each category, we identified the services that had a median effectiveness of 8–9 (or 9 for outreach) as the most highly recommended services. Twenty-four services across all categories met this threshold. Lastly, we found that the variable level of intensity had a significant effect on a service’s rated level of effectiveness for some but not all services. For example, licensed professionals were rated significantly higher for developing care plans, whereas community health workers were equally effective in executing that care plan.

CONCLUSIONS: Because of their heterogeneity and the difficulty testing them against clear outcomes, enabling services make up a relatively understudied field of healthcare. In this study, our community-academic research team brought together two patient and stakeholder panels to define three constructs - six broad categories, 112 granular-level services, and various service intensity variables—as a framework for evaluating the effectiveness of these services. While not exhaustive, this list can serve as an important, initial classification scheme to guide further research. For providers and policy-makers, the list of most highly recommended services is the first report of its kind to provide guidance about priority services based on the perspectives of underserved patients and their representatives. In addition, by rating at the level of the service intensity, we identified opportunities to target resources to services where they are most needed to improve effectiveness. In many cases, the highest level of intensity did not necessarily increase perceived effectiveness. This study was limited by the dearth of literature suitable for the expert panel and the small number of experts used to create the recommendations. Nevertheless, given the potential that enabling services have for reducing social barriers to care, the categorization and ratings developed by this study provide a useful framework for further studying these services.

DEVELOPMENT AND APPLICATION OF A CLASSIFICATION SCHEME FOR CARE COORDINATION ACTIVITIES IN AN ACADEMIC PRIMARY CARE SYSTEMNazleen Bharmal¹; Robin Clarke¹; Paul Di Capua¹; Indu Gupta¹; Brian Doyle¹; Aliza Ali²; Asad Malim³; Brian S. Mittman^4,5. ¹UCLA, Los Angeles, CA; ²Lake Erie College of Osteopathic Medicine, Bradenton, FL; ³College of Science and Health, Charles R. Drew University, Los Angeles, CA; ⁴VA Greater Los Angeles Healthcare System, Los Angeles, CA; ⁵Kaiser Permanente Southern California, Los Angeles, CA. (Tracking ID #1939165)

BACKGROUND: Primary care redesign aims to improve patient experience, improve health, and reduce costs. One opportunity to transform practice care delivery is to introduce supports to enhance primary care team effectiveness. In the field of care coordination, the RN-based case manager model is defined and well-understood; care managers may have a specific group of patients with chronic conditions for whom they develop individualized care plans and/or train in self-management skills. However, one of the limitations of the limited-panel, high-utilizer care manager model is that it may only touch on a small percentage of patients. In our large health system comprised of 28 primary care clinics throughout Los Angeles County, we embedded non-licensed care coordinators (CC) into clinics to integrate with primary care teams. CCs would act as PCP extenders, assess patients’ care needs, coordinate patients’ care with other providers and settings, and connect patients to community resources and social services. To better describe and ultimately analyze the role of CCs in our health system, we developed and applied a classification scheme to categorize CC activities.

METHODS: The first five general internal medicine practices to incorporate a CC were included in the study. CC encounters were defined as any activity performed by a CC that (a) contributed to the development and/or implementation of a plan of care for a patient or family and (b) was documented by the CC in an online Patient Care Coordination System (PCCS) database established to track CC activities. A classification scheme to broadly categorize CC activities was developed iteratively by adapting models cited in a systematic review of care coordination published by the Agency for Healthcare Research & Quality (AHRQ) plus literature on case management, discussions with CCs, and our exploratory review of the PCCS database. Five coders, in total, coded all activities with high inter-rater reliability (kappa 0.89).

RESULTS: The five selected practices represented a diverse group of community and academic practices focused on family medicine, internal medicine, or geriatrics. The PCCS database contained 8,036 CC encounter records entered from June 2012 to July 2013 among the five primary care practices. Our 5-category classification scheme is presented in the table. The proportion of CC activities coded in each category was 38% for execution of care, 32% for coordination of transitions, 15% for link to community resources or self-management support, 10% for monitor and follow-up, 1% for patient assessment, and 5% for no classifiable activity.

CONCLUSIONS: CC documented activities provided evidence to the broad range of potential impact this model can have on patients, including helping patients navigate the health system, improving access and communicating across the care spectrum. CC activities are a vital part of helping primary care practices deliver coordinated, accessible, comprehensive, and patient-centered care; a valid and reliable classification scheme for these activities is an important tool for research and practice. The scheme we developed will be incorporated into CC training and a revised PCCS database and documentation system. The scheme will also facilitate subsequent phases of our research, including an assessment of the impact of CC activities and the degree of CC integration into primary care practices on key outcomes, including patient and provider satisfaction, chronic disease prevention and management, and preventable emergency room utilization.

Care Coordinator Activities by Mutually Exclusive Process Categories

DEVELOPMENT AND EVALUATION OF A PATIENT EMPOWERMENT LETTER IN A PRIMARY CARE SETTINGDan Matlock; Eva M. Aagaard; Adam Abraham; Wagner J. Schorr-Ratzlaff; Nivedita D. Mahidhara; Scott De La Cruz; Lisa Schilling; Brandon Combs; Carmen L. Lewis. University of Colorado, Aurora, CO. (Tracking ID #1939663)

BACKGROUND: Patients often feel disempowered when talking with their physicians. One study noted that patients feared questioning their doctors because they might be dismissed or labeled “difficult.” As part of the Patient Centered Medical Home (PCMH) initiative at the University of Colorado, we developed a letter to be given to patients prior to their clinic appointment to both empower patients and to help them set an agenda. Herein, we describe the development and evaluation of this letter.

METHODS: Development phase: Initially, the idea was to give patients something prior to their visit where they could write down their agenda. The letter was developed iteratively and piloted with patients. Ultimately, it included language designed to empower patients with phrases like “Please don’t feel that we will be upset if you have more questions.” The letter also included several lines for patients to write their “main goals, fears, and worries for the visit that day.” Evaluation phase: All adult patients aged 18–99 seen on one team at an internal medicine clinic over a two-week time period were included. The letter was evaluated using a pre-post design with the first week being the control and the second week including the letter. The primary outcomes were patient empowerment using the low-literacy version of the validated patient activation measure (13 items, 4 point Likert scale, scored 0–100) and patient satisfaction using the hospital mandated patient satisfaction measure (10 items, 5 point Likert scale, scored 0–5). Patients were also given the opportunity to provide open-ended feedback on the letter. Differences between mean scores were tested using t-tests.

RESULTS: We received a total of 171 responses (85 in week one and 86 in week two) from 248 distributed surveys (response rate = 69%). The sample was 63% female, 39% under the age of 55, 37% had more than 4-year college degree, 76% Caucasian and 56% married. There was no difference in the Patient Activation scores between the control and intervention periods (71.4 vs. 68.9/100, p = 0.31). There was a small but statistically significant decrease in satisfaction scores between the control and intervention (4.86 vs. 4.56/5, p < 0.01) and overall satisfaction was very high. Patients had positive things to say about the survey: “Great letter and very impressive that your office is distributing this and indicating the need that every patient should be proactive with their own health and healthcare.” Suggestions on improving the letter including: arranging the key points into a bulleted list, shortening the overall length of the letter (simplifying) and including goals of care in patient print-outs.

CONCLUSIONS: Overall, the letter had no effect on patient activation and there was a small but significant decrease in patient satisfaction scores although overall satisfaction scores were very high. It is possible that satisfaction scores lowered slightly because patients’ agenda written on the letter were not addressed. Overall, the qualitative feedback suggests that the letter was well-received. We are currently working on a shorter version of the letter based on the results of this study.

DEVELOPMENT AND VALIDATION OF THE MEDICAL PROFESSIONALISM BEHAVIOR ASSESSMENT TOOLPreston Reynolds¹; James Martindale². ¹University of Virginia, Charlottesville, VA; ²University of Virginia, Charlottesville, VA. (Tracking ID #1940065)

BACKGROUND: Assessment of professionalism behaviors has become a priority with launching of the Milestones Initiative by the Accreditation Council on Graduate Medical Education. The Medical Professionalism Behavior Tool (MPB) was developed by P Reynolds with input from clinician-educators, education researchers, and internal medicine residents beginning in 2008. It is modified from the NBME’s Assessing Professional Behaviors tool with inclusion of behaviors relevant to cultural competency.

METHODS: As part of an annual internal medicine resident seminar series, P Reynolds began introducing the concept of evaluating professional behaviors in 2008. In subsequent years, she reviewed with the residents various behaviors and then a 25-item evaluation tool she developed modified from the NBME’s APB with inclusion of behaviors specific to cultural competency. Based on their feedback, the tool was reduced to 15 items with retention of the cultural competency skills. The MPB was then used by all of the medicine residents in three subsequent annual seminars as part of a self-reflective exercise where they were asked to rate themselves and then discuss with another resident one or two areas they specifically identified as needing improvement. These data were entered into a spreadsheet making the responses anonymous, and then subjected to statistical analysis to determine the degree to which evidence of reliability and validity was present in the MPB as an assessment tool. The MPB was also reviewed by 11 of 13 residency program directors at the University of Virginia for content validity

RESULTS: All of the residency program directors thought the MPB captured behaviors relevant to evaluation of professionalism in clinical medicine. Statistical analysis of the internal consistency (reliability) and construct validity associated with the MPB yielded strong results with Cronbach’s alpha = 0.84 and thus providing good statistical evidence of reliability and validity.

CONCLUSIONS: The strength of the MPB is that it is only 15-items, reflects behaviors central to medical professionalism as described in The Physician Charter, and also incorporates behaviors specific to culturally competent clinical care. Fourteen states now require demonstration of cultural competency as part of medical licensure making the MPB an important assessment tool to shape residency training to include skills training in medical professionalism, and in cultural competency. Additionally, the MPB is a tool that is reliable and valid and thus, can be used in formative and summative assessment.

DEVELOPMENT OF A SCREENING TOOL FOR IDENTIFYING HOUSING-RELATED BARRIERS TO CARE AMONG LOW-INCOME WOMENAmbili Ramachandran¹; Samantha Morton²; Naomi Y. Ko³; Kerrie Nelson⁴; Maria Castano³; Emily Bergling³; Sharon Bak³; Jennifer Cedor³; Wanda Turner³; Tracy A. Battaglia³. ¹Boston University School of Medicine, Boston, MA; ²Medical-Legal Partnership | Boston, Boston, MA; ³Boston University School of Medicine, Boston, MA; ⁴Boston University School of Public Health, Boston, MA. (Tracking ID #1934188)

BACKGROUND: Social determinants of health include access to safe, healthy, and affordable housing. Support services such as patient navigation frequently target and resolve this barrier to care, but vulnerable patients sometimes need legal advice or advocacy to address these housing issues. It is often difficult in a health care setting to appropriately and efficiently determine which patients have housing problems that require legal advice or assistance. We developed a screening tool for use by patient navigators to identify individuals who would benefit from legal advice or assistance regarding housing-related barriers to care.

METHODS: The screening tool was designed in collaboration with Medical-Legal Partnership | Boston (MLP | Boston) and featured questions regarding utility expenses, mortgage and rent payments, foreclosures and evictions, and housing conditions. Dichotomous responses were designated “positive” or “negative” based on current Massachusetts laws protecting tenants, homeowners, utility consumers, and shelter residents, and at least one positive response qualified as a positive screen. After piloting, the screening tool was administered to female patients with public or no health insurance attending a breast health clinic at a safety-net hospital. Performance of the screening tool was validated against a comprehensive, in-person, one-hour legal interview by an attorney followed by a legal evaluation by MLP | Boston, which represented the “gold standard” for detecting housing-related barriers to care. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the screening tool were computed. Descriptive statistics on demographic characteristics were also calculated.

RESULTS: Of 74 eligible women approached, 35 (47%) of a targeted sample of 50 consented to participate: mean age 42years, 54% African-American, 30% White, 8% Latina, and 8% Other Race. Most subjects had state subsidized health insurance (43%), Medicaid (26%), or Medicare (11%), and 83% of women had incomes under 200% of the federal poverty level. While 20% of subjects owned their home, 57% rented, 17% stayed with friends or family, and 6% had other housing arrangements. Almost 40% of subjects rated their health as fair or poor. Among the 31 out of 35 women who completed full legal evaluation at the time of analysis, the screening tool had a sensitivity of 89% (17/19), specificity of 42% (5/12), PPV of 71% (17/24), and NPV of 71% (5/7). Based on the legal evaluation, over 60% of women had housing-related barriers to care. Among owners and renters, the screening tool was most often positive for threatened disconnection of utility services (10/24 or 42%), of which 3 were false positives among renters only.

CONCLUSIONS: In this population of low-income, predominantly publicly-insured women, housing-related barriers to care were widespread. When compared to the comprehensive legal assessment by the attorneys, our screening tool had acceptable sensitivity, but modest specificity, for identifying housing-related barriers to care that require legal advice or assistance to resolve. After completion of targeted enrollment, future work will refine which screening questions best identify subjects whose housing-related barriers to care are amenable to legal advocacy and which responses trigger the need for access to legal advice or assistance.

DEVELOPMENT OF QUALITY INDICATORS FOR THE CARE OF WOMEN WITH ABNORMAL UTERINE BLEEDINGDonna L. Washington^1,2; Marjorie Danz^1,3; Kristina M. Cordasco^1,2. ¹VA Greater Los Angeles Healthcare, Los Angeles, CA; ²University of California, Los Angeles, Los Angeles, CA; ³RAND Corporation, Santa Monica, CA. (Tracking ID #1936245)

BACKGROUND: Abnormal uterine bleeding (AUB) is among the most common gynecologic complaint of reproductive-age women in ambulatory care settings, with up to 30% of women seeking medical care for AUB during their reproductive years. International consensus recommendations for standardizing the nomenclature and methods for investigating and categorizing potential AUB etiologies (the FIGO classification system for causes of AUB) were recently developed as an approach to improve our understanding and management of AUB. We sought to translate consensus recommendations and evidence-based guidelines for AUB care into quality indicators (QIs) that are measurable from electronic health records (EHRs), with the ultimate goal of applying these QIs to assess variations in clinical management and aid in women’s health education for healthcare providers.

METHODS: To develop QIs, we first performed an extensive literature review to develop a set of potential QIs for the evaluation and management of women with AUB, targeting aspects of care that are within the purview of primary care providers (PCPs). For each potential QI, we specified a numerator, denominator, exclusions, relevant definitions, period of assessment, underlying evidence and data source. We then used a modified Delphi consensus technique based on the RAND/University of California-Los Angeles Appropriateness Method, which consisted of three steps. In step one, nine experts (women’s health PCPs, obstetrician/gynecologists, quality measurement experts, policy stakeholders) independently ranked the indicators using nine-point scales on three metrics: consistency with established guidelines; importance to women’s health; and reliability of measurement from Veterans Health Administration (VHA) EHRs. In step two, we conducted an expert panel tele-conference with internet-based sharing of presentation materials, in which indicators were discussed by the panelists. In the final step, all indicators were independently rated by panelists a second time, using the same nine-point scales. QIs were selected for final inclusion if they had a median score of 7 or higher on all three metrics, and high consistency in their ratings, based on the interpercentile range adjusted for symmetry (IPRAS), using an interpercentile range of 0.3–0.7.

RESULTS: The consensus process resulted in selection of 19 QIs covering multi-dimensional aspects of AUB care, spanning reproductive and post-menopausal life phases, and including profuse vaginal bleeding, and use of hormonal contraception/IUDs. Five QIs relate to documentation of critical aspects of the history; 6 to diagnostic evaluation; and 8 to management. Evaluation QIs highlight the need to recognize pregnancy, need for emergency care, and elevated endometrial cancer risk. Management QIs focus on current profuse bleeding, follow-up of diagnostic studies, and indications for gynecologist referral. Areas of disagreement in panelists’ ratings and discussion related to the acceptable timeframe for completion of different actions, with generalist panelists being more likely to consider the constraints of the ambulatory care setting in their decision-making.

CONCLUSIONS: This project resulted in development of a multi-dimensional set of QIs for primary care AUB evaluation and management, based on published data, consensus recommendations and guidelines, and a modified Delphi process. These QIs are being formulated into an EHR quality of care assessment tool. Once this tool is pilot-tested for feasibility, it will be applied on a larger scale to measure the care quality provided to women with AUB in VHA primary care settings. This rigorously designed set of QIs should facilitate measuring and improving the quality of care for women with suspected AUB. Future work should focus on developing educational programs that are concordant with these QIs and assess potential changes in quality outcomes associated with their implementation.

DEVELOPMENT OF THE IORA WORRY SCORE AND COMPARISON TO SUBJECTIVE PROVIDER WORRY SCORES IN MULTIPLE PRIMARY CARE PRACTICESDaniel P. Croft¹; Benjamin Berk²; Whitney Kramer²; Joel Lazar^2,3; Rushika Fernandopulle²; Eugene Nelson³. ¹Dartmouth Hitchcock Medical Center, Lebanon, NH; ²Iora Health, Cambridge, MA; ³The Dartmouth Institute for Health Policy & Clinical Practice, Lebanon, NH. (Tracking ID #1927628)

BACKGROUND: The degree of provider concern for individual patients significantly impacts panel management decisions and subsequent resource allocation within primary care practices. As current decision support tools do not effectively capture provider concern levels, the Iora Worry Score (IWS) was developed to reflect providers’ level of concern for individual Iora Health patients. The primary outcome of interest in this study was the strength of agreement between the IWS and the subjective provider worry score (SPS). This outcome was validated against an established stratification tool, the Milliman Advanced Risk Adjusters (MARA) score. A secondary outcome of interest was the relationship between both worry scores (IWS and SPS) and the average number of primary care visits.

METHODS: Iora Health is a venture-funded health care company based in Cambridge, MA, building a new model of primary care that serves a broad spectrum of patient populations, with a special focus on patients with complex chronic diseases. A patient’s IWS was defined as a provider’s general level of concern for a patient in terms of the anticipated need for team-based review and potential follow-up. The IWS was reported on a scale of 1 (least concern) to 10 (most concern) and was created using 77 evidence-based utilization, clinical, administrative and patient-reported variables. The IWS was applied to 5,773 patients across four Iora Health practices in a retrospective review. Iora Health providers assigned a SPS to each patient within a randomly selected subset of 352 patients for inclusion in the analysis of the primary outcome. The Kendall tau rank correlation coefficient was used to determine the strength of agreement in the analysis of IWS, SPS and MARA score.

RESULTS: The randomly selected cohort of 352 patients accurately represented the demographics of the entire Iora Health system population. Though few significant demographic differences were found across the four practices, there was significant variation in the IWS across practices (p < 0.0001). As the primary outcome, the statistically significant agreement noted between the IWS and the SPS (p < 0.0001; tau = 0.47) was stronger than the agreement between the comparator MARA score and the SPS (p < 0.0001; tau = 0.39). The SPS and the average IWS across all practices both displayed a statistically significant positive relationship with the average number of primary care visits (p < 0.0001, tau = 0.43 and p < 0.0001, tau = 0.44 respectively).

CONCLUSIONS: When compared to an accepted stratification tool, the IWS had a stronger degree of agreement with subjective provider concern levels. Implementing the IWS electronically through Iora Health’s care collaboration platform can save provider time by automating the triage of patients for team review and increase the responsiveness of Iora Health to concerning changes in individual patient worry scores. Ongoing refinement of the IWS to improve proactive panel management will be an important effort within the overall movement toward improving primary care delivery and patient outcomes.

DIABETES CONTROL IN OLDER PATIENTS IN PRIMARY CARE CLINIC VS. ENDOCRINOLOGY CLINICZoobia Chaudhry; Phil Levin; Lee Bromberger; Hsin-Chieh Yeh. Johns Hopkins School of Medicine, Baltimore, MD. (Tracking ID #1935937)

BACKGROUND: With increasing prevalence of diabetes, the demands for appropriate management by primary care providers as well as endocrinologists have amplified significantly. Although primary care providers see the vast majority of patients with type 2 diabetes, very few studies compared diabetes control in patients seen in primary care setting vs. in endocrinology clinic. A study in early 2000s reported a better quality of care in diabetes clinic than in primary care clinic, but no recent data indicated whether the disparity remained. We therefore conducted a retrospective study to compare HbA1c, blood pressure, and lipids controls in diabetic patients age 65 and older who were all eligible for Medicare coverage.

METHODS: A cross-sectional medical record review for older diabetic patients receiving care in 2010 and 2011 was conducted in a primary care clinic (PCC) and in an endocrinology clinic (EC) located in a close geographic proximity in Baltimore City. Data from the PCC included all older patients seen during the study period (n = 713); data from the EC was a random sample of older patients (n = 100) from medical chart reviews.

RESULTS: Patients seen in the PCC were older (mean age: 75.5 ± 6.7 vs. 73.3 ± 7.0, p = 0.002), were more likely to be black (75% vs. 14%, p < 0.0001), had higher body mass index (BMI, 30.9 ± 6.4 vs. 28.2 ± 7.7; p < 0.0001), and were less likely to be on insulin (24% vs. 38%, p = 0.004). There was no difference in gender distribution (women 56% vs. 54%, p = 0.69). In the univariate analysis, patients seen in the PCC and EC had similar mean HbA1c (PCC 6.93 ± 1.5%, EC 6.99 ± 1.5%; p = 0.69), systolic blood pressure (PCC 133 ± 18mmHg, EC 130 ± 15mmHg; p = 0.132), and total cholesterol (PCC 166 ± 40mg/dl, EC 162 ± 36mg/dl; p = 0.44), but had lower diastolic blood pressure (67 ± 12mmHg vs.72 ± 10mmHg; p = 0.0001) and higher HDL-cholesterol (56 ± 19mg/dl vs. 51 ± 18mg/dl; p = 0.006). In the multivariable linear regressions adjusting for age, race, sex, BMI, and insulin use, compared to patients seen in PCC, patients seen in EC had a significantly lower mean HbA1c (beta-coefficient: −0.57%; 95% confidence interval (CI): −0.94% to −0.19%; p = 0.003), but a higher DBP (beta-coefficient: 4.0mmHg, 95% CI: 0.88 to 7.14mmHg; p = 0.012).

CONCLUSIONS: Glycemic control in older patients with type 2 diabetes was better in patients seen in EC as compared to those seen in PCC, partly due to the use of insulin. On the other hand, patients seen in PCC had a better control on diastolic blood pressure. Future research should explore the differences between the primary care clinic and endocrinology clinic on treatment strategies and other aspects of clinical practices such as counseling on lifestyle changes.

DIABETES EMPOWERMENT PROGRAM: CHANGES IN PATIENT SELF-CONFIDENCE, DIABETES SELF-MANAGEMENT, AND HEALTH OUTCOMESMonica E. Peek; Nora Geary; Yue Gao; Daniel J. Rowell; Yolanda O’Neal; Tonya Roberson; Julie L. Whyte; Nyahne Bergeron; Marshall Chin. University of Chicago, Chicago, IL. (Tracking ID #1939499)

BACKGROUND: African-Americans disproportionately suffer from diabetes and its complications. Although culturally tailored education can improve diabetes self-management and health outcomes among this population, little prior work has combined diabetes education with patient/provider communication training to empower African-American patients to better manage their disease.

METHODS: As part of a larger intervention, we developed the Diabetes Empowerment Program, a 10-week program that combined culturally-tailored diabetes education with skills training in patient/provider communication and shared decision-making. The education classes were modeled after the BASICS diabetes curriculum, and were adapted based on principles of health literacy and adult learning theory. Session length was adjusted to reduce the volume of information shared per session and audio visual aids supplemented written materials. The shared-decision making classes focused on building patients’ skills and confidence in asking more questions, giving more information, clarifying/restating what the doctor says, and communicating health care preferences. The classes utilized role-play, narrative, group interaction and problem-solving. Following the 10-week session, patients were invited to join a monthly support group for ongoing learning, skills building and social support. Patients were recruited to participate in the classes from two university clinics and four federally qualified health centers. We utilized in-person surveys to measure the following: 1) patients’ self-confidence about shared decision-making and diabetes self-management (i.e. decision-making empowerment and diabetes self-efficacy), 2) patient’s behaviors regarding diabetes self-management behaviors and shared decision-making (i.e. patients’ perceived involvement in care, blood sugar testing, exercise, healthy eating, and foot care), and 3) intermediate health outcomes (i.e. HbA1c, systolic blood pressure, weight and LDL cholesterol). Data were collected at baseline, immediately following the intervention (post), 3-months, and 6-months post intervention. We used linear mixed models to assess changes between baseline data and three follow-up points, controlling for age, gender, education, health insurance, number of years since diabetes diagnosis, self-reported health status, and co-morbidities (e.g. stroke, asthma, hypertension, hyperlipidemia).

RESULTS: Among the 118 patients (with at least one follow-up data point), the mean age was 57years, 78% were female, 97% were African American and 40% were uninsured or Medicaid-insured. The average baseline HbA1c was 8.8%. Patients’ self-confidence about diabetes care significantly improved after program participation in decision-making empowerment and diabetes self-efficacy (Table 1). Significant improvements in diabetes self-management behaviors were also observed after program participation. Patients perceived themselves to be more involved in their care, and reported higher frequencies of exercising, testing blood sugar, and doing foot examinations. No significant findings were observed in healthy eating. After program participation, significant changes in intermediate health outcomes were also observed. There was a significant decrease in HbA1c from baseline, which was sustained at the 6-month follow-up. Additionally, a significant decrease in weight was observed at the 6-month follow-up. No significant changes in LDL or systolic blood pressure values were observed.

CONCLUSIONS: Our findings suggest that combining culturally-tailored diabetes education with skills training in patient/provider communication can be an effective strategy to improve self-confidence and behaviors in self-management and shared decision-making, and also improve diabetes-related health outcomes. Such strategies may serve to reduce diabetes disparities among African-Americans.

Changes in self-confidence, diabetes self-management behaviors, and intermediate health outcomes in the Diabetes Empowerment Program

(a) Scale from 0 to 100, low to high (b) Number of days per week No significant changes observed in systolic blood pressure, LDL, or healthy eating.

DIABETES PREVENTION: ARE WE TRANSLATING EVIDENCE INTO PRACTICE?Tannaz Moin^1,2; Jinnan Li²; O. Kenrik Duru²; Susan Ettner²; Norman Turk²; Robert H. Luchs³; Abigail M. Keckhafer³; Anya Kirvan³; Carol Mangione². ¹VA Greater Los Angeles Healthcare System and HSR&D Center for Healthcare Innovation, Los Angeles, CA; ²David Geffen School of Medicine at UCLA, Los Angeles, CA; ³UnitedHealth Group, Minneapolis, MN. (Tracking ID #1939018)

BACKGROUND: Pre-diabetes is a major health care burden that is associated with an increased risk of type 2 diabetes and related complications. More than 10years ago, several landmark studies demonstrated that both lifestyle changes and metformin use can significantly reduce the risk of progression to diabetes compared to placebo. Soon after, the safety, tolerability and cost-effectiveness of metformin use in patients with pre-diabetes was also demonstrated. Beginning in 2008, the use of metformin was included in national diabetes prevention guidelines by the American Diabetes Association (ADA). To date, very little is known about the uptake and translation of the evidence supporting metformin use in patients with pre-diabetes. The goal of this analysis was to characterize national patterns of metformin use among patients with pre-diabetes and examine whether a health plan designed specifically for patients with pre-diabetes and diabetes (the “Diabetes Health Plan,” or DHP), which eliminates copayments for metformin and enhances access to care management, could impact the translation of evidence into practice.

METHODS: We conducted a retrospective analysis of claims data from a national commercial health plan between 2009 and 2012. We included patients with pre-diabetes between 19 and 63years of age with consecutive enrollment over 3years and compared those who were enrolled in the DHP to those enrolled in a standard medical plan. A diagnosis of pre-diabetes was defined by two or more ICD-9 diagnoses of 790.2× (during an inpatient, outpatient, or ED claim), or last HbA1c value of 5.7–6.4%, or last Fasting Plasma Glucose (FPG) value of 100–125ng/dl or last oral glucose tolerance test (OGTT) value of 140–199ng/dL. We excluded patients with a history diabetes, gestational diabetes, or use of anti-glycemic medications at baseline.

RESULTS: We analyzed data from 23,843 patients with pre-diabetes and compared results among those enrolled in the DHP (n = 935) to those enrolled in standard medical plans (n = 22,908). The prevalence of metformin use among patients with pre-diabetes was similarly low in both groups at baseline (1.9% in the DHP [n = 18] and 1.7% in the standard plans [n = 382]). Over the following 2years we observed minimal change in the proportion of patients with pre-diabetes who were newly prescribed metformin whether or not they were enrolled in the DHP (2.1% new metformin prescriptions in the DHP plan [n = 19] and 1.2% in the standard plans [n = 262]).

CONCLUSIONS: Our findings indicate that metformin is rarely used in the management of pre-diabetes, despite a strong evidence base in the literature for over 10years. This is a concerning gap in the clinical approach to pre-diabetes and potentially impacts up to one third of the U.S. population currently estimated to have pre-diabetes. Enrollment in a pre-diabetes health plan, which eliminates medication and physician visit copays and increases pre-diabetes disease awareness, minimally increased the rates of metformin use among patients with pre-diabetes. Further studies are needed to understand the root causes of this existing gap and possible interventions to help promote the translation of evidence into future practice.

DIAGNOSING AND COUNSELING OBESE PATIENTS IN THE PRIMARY CARE SETTINGAdam J. Balzer; Cynthia Phan; Priya Radhakrishnan. St. Joseph’s Hospital and Medical Center, Phoenix, AZ. (Tracking ID #1938375)

BACKGROUND: The incidence of obesity in the U.S. has expanded to greater than 30% of the population and is now widely regarded as a national epidemic. Obesity is defined as a body mass index of greater 30kg/M2. For years, it has been clearly linked to several chronic diseases including coronary artery disease, hypertension, and diabetes mellitus. Studies are also demonstrating that obesity is an independent risk factor for prolonged hospitalizations, increased ambulatory care utilization, and increased healthcare costs. While healthcare providers are very aware of detrimental impact of obesity, a systematic approach to obesity as a disease entity and the implementation of a treatment plan does not occur routinely. With the expansion of electronic health record systems and underlying requirement of “meaning use” rules, BMI must be recorded at each clinic visit. If the BMI is outside parameters, a follow-up visit must be documented.

METHODS: With the transformation into a electronic medical record system in the Internal Medicine, we undertook a quality improvement project. Chart review was performed on patients from May 2012 to June 2012 of patients that had appointments at the Internal Medicine Clinic Health Center. Data was collected and analyzed. We looked at the percentage of patients with a documented body mass index of 30 or greater. A chart review was performed for each patient with a BMI of greater than 30 to see if obesity was diagnosed and if counseling was performed.

RESULTS: From 5/1/2012 to 6/18/2012, 744 encounters were analyzed from both residency clinic and faculty practice. Forty-three percent of 744 encounters had a clinical diagnosis of obesity listed in the problem list. Of these, 60% of those diagnosed received counseling. Of the 744 encounters, there were 557 patients made up the 744 encounters; 142 patients were seen at least two times and 61% of those patients were diagnosed with obesity.

CONCLUSIONS: The results of this study demonstrate that BMI is consistently measured and recorded in the EMR, however, obesity is not consistently recorded as a diagnosis in the problem list and treatment options are not routinely documented in the EHR. It is interesting to note that despite having multiple encounters, obesity was consistently not clinically diagnosed in 39% of the patients. This quality improvement project demonstrates the importance of linking education with meaningful use metrics to make a clinical impact. We hope these findings initiate further research on provider understanding of EHR utilization and potential barriers to recognizing and treating obesity in the primary care setting.

DID THE UNITED STATES PREVENTIVE SERVICES TASK FORCE 2005 GUIDELINES FOR ABDOMINAL AORTIC ANEURYSM SCREENING AFFECT MORTALITY IN THE UNITED STATES POPULATION?Devin B. Malik; Sourabh Aggarwal. Western Michigan University School of Medicine, Kalamazoo, MI. (Tracking ID #1938931)

BACKGROUND: An abdominal aortic aneurysm (AAA) is defined when infra-renal aortic diameter is at least 3.0cm. The United States Preventive Services Task Force (USPSTF) made recommendation in 2005 that all men between the age of 65 to 75years and who have ever smoked should be screened one time for AAA by abdominal ultrasonography. However, the clinical impact of these recommendations are unknown in the American population.

METHODS: We queried Healthcare Cost and Utilization Project’s Nationwide Inpatient Sample (NIS) for AAA and AAA rupture using ICD9 codes 441.4 and 441.3 respectively. The NIS represents 20% of all hospitals data in US. All the data was extracted for years 2000–2010. The prevalence and in-hospital mortality for pre-screening years (2000–2004) was compared with post-screening years (2006–2010). Chi square was used to find statistical significance.

RESULTS: A total of 527,801 hospitalizations secondary to AAA and AAA rupture were analyzed for the study period. AAA prevalence decreased from 61.72 to 58.77 per 10,000 total hospitalizations with in-hospital mortality decreasing from 3.5 to 2.12% (p value <0.001). On sub-analysis, the decrease in prevalence was chiefly in 65–84 age group (77.19 to 74.54 per 100 AAA admissions, p < 0.001), with increase in 84+ age group (6.37 to 8.6 per 100 AAA admissions, p < 0.001) and no significant change in 45–64 age group (16.16 to 16.51 per 100 AAA admissions, p value >0.05). Prevalence increased in males (77.84 to 78.13 per 100 AAA admissions, p < 0.001) and decreased in females (22.13 to 21.81 per 100 AAA admissions, p < 0.001). Decrease in mortality was uniform in all age and gender sub-groups. The prevalence of AAA rupture decreased from 9.51 to 7.03 per 10,000 total hospitalizations (p < 0.001). On sub-analysis decrease in prevalence was reciprocated in 65–84 age group (70.08 to 64.94 per 100 AAA rupture admissions, p value <0.001) and males (73.23 to 71.7 per 100 AAA rupture admissions, p value <0.001). However, prevalence of AAA rupture increased in age group 45–64 (14.41 to 15.43 per 100 AAA rupture admissions, p value <0.001), age 84+ (15.21 to 19.2 per 100 AAA rupture admissions, p value <0.001) and females (26.28 to 28.28 per 100 AAA rupture admissions, p value <0.001).

CONCLUSIONS: Our study reveals that post screening recommendations, hospitalizations for AAA decreased significantly with decrease only reciprocated in the age group 65–84years. Also hospitalizations and in-hospital mortality from AAA rupture decreased in males and age group 65–84, with increase in hospitalizations for the 45–64 and 84+ age group and females. The possible explanation is better screening as outpatient resulted in decreased morbidity with decreased hospitalizations from AAA and AAA rupture in susceptible population. Our study also makes the case to consider extension of recommendations to include other susceptible groups.

DIFFERENCES IN ACCESS TO HEALTH CARE AMONG LESBIAN, GAY, AND BISEXUAL POPULATIONS IN NEW YORK CITYKristin A. Swedish¹; Marcus A. Bachhuber²; Viraj V. Patel¹. ¹Montefiore Medical Center, Bronx, NY; ²University of Pennsylvania, Philadelphia, PA. (Tracking ID #1938935)

BACKGROUND: Lesbian, gay, and bisexual (LGB) groups in the United States experience significant health care disparities across a wide range of conditions. Despite being considered a priority area to target in Healthy People 2020, there is limited nation-wide data available on these populations. Previous research suggests that LGB members of racial or ethnic minorities often face the highest level of health disparities. We examined differences in access to care by self-identified sexual orientation and race/ethnicity in New York City (NYC).

METHODS: We used multi-year pooled data from the 2009–2011 NYC Community Health Surveys, a probability-based survey using a stratified random sampling. The Community Health Survey is based upon the National Behavioral Risk Factor Surveillance System and samples approximately 10,000 adults from all five boroughs of NYC (response rates 37.7–40% for 2009–2011). We examined two primary outcomes: 1) identifying someone as a primary care provider (PCP) and 2) receiving medical care when needed in the prior 12months. Self-identified sexual orientation was the primary independent variable, accounting for gender. We conducted descriptive analyses and multivariate logistic regression analyses adjusting for age, race/ethnicity, nativity (US vs. foreign born), education, employment status, insurance status, and primary language spoken at home. All analyses incorporated survey weights.

RESULTS: Survey respondents included 603 gay men, 204 lesbian women, 115 bisexual men, and 133 bisexual women, with mean age 48.9years (SD 14.9). The majority of LGB individuals were White (59.5%), college-educated (62.9%), employed or self-employed (64.9%), privately insured (57.5%), and US-born (82%). Lesbian and bisexual women were significantly less likely than heterosexual women to identify a PCP (adjusted odds ratio [AOR] 0.46, 95% confidence interval [CI] 0.45–0.47 and AOR 0.66, 95% CI 0.65–0.68, respectively). Lesbians and bisexual women were also less likely than heterosexual women to have received medical care when needed in the prior 12months (AOR 0.73, 95% CI 0.71–0.75 and OR 0.41, 95% CI 0.40–0.42, respectively). In contrast, when compared to heterosexual men, gay and bisexual men were significantly more likely to have a PCP (AOR 1.09, 95% CI 1.07–1.11 and AOR 1.22, 95% 1.18–1.27, respectively). Both gay and bisexual men were also more likely than heterosexual men to have received medical care when needed in the prior 12months (AOR 1.36, 95% CI 1.33–1.39 and OR 1.54, 95% CI 1.47–1.62, respectively). When stratified by race/ethnicity, lesbian and bisexual Black and Hispanic women were significantly less likely than heterosexual white women to have received medical care when needed in the prior 12months (AOR 0.50, 95% CI 0.49–0.52 and AOR 0.34, 95% CI 0.33–0.35, respectively). Gay and bisexual Hispanic men were significantly less likely than heterosexual white men to receive medical care when needed in the prior 12months (AOR 0.58, 95% CI 0.56–0.60).

CONCLUSIONS: Lesbian and bisexual women in NYC have worse access to health care when compared with heterosexual women; these disparities are more pronounced in Black and Hispanic women. It is notable that similar disparity is not seen among gay and bisexual men in NYC, apart from the Hispanic subgroup. It is possible that the HIV epidemic may have altered the way gay and bisexual men utilize health care in NYC. It is also possible that lesbian and bisexual women fear greater stigma in the health care setting. Further research is needed to elucidate these disparities. The findings of this study should be used to help public health officials design interventions and policies to increase access to care in LGB populations.

DISCHARGE ROUNDS: IMPLEMENTATION & EFFECTS OF A SYSTEMATIC APPROACH TO EARLY DISCHARGES ON PATIENT THROUGHPUTKrisda Chaiyachati^1,2; Andre N. Sofair^1,2; Jeremy I. Schwartz^1,2; David Chia^1,2. ¹Yale School of Medicine, New Haven, CT; ²Waterbury Hospital, Waterbury, CT. (Tracking ID #1922266)

BACKGROUND: Patient throughput is an important aspect of hospital medicine. It has effects on patient satisfaction and safe transitions of care from home-to-hospital by decreasing emergency room wait times, time to evaluation by a physician, and time to hospital admission. Early discharges benefit these upstream events by creating bed space and allowing support staff to plan accordingly, setting up a cycle of events necessary for new hospital admissions to begin.

METHODS: In order to improve throughput, our internal medicine residency program systematically instituted daily early morning discharge rounds on the inpatient teaching service on August 1, 2013. Discharge rounds require that each medical team identifies potential discharges and completes the necessary paperwork the day prior to anticipated discharge. The following day, these patients are evaluated early and discharge orders are initiated in real-time. We conducted a retrospective pre-post study to determine whether discharge rounds were effective in changing when discharge orders were initiated and when patients actually left the hospital. To account for changes in practice patterns, we compared the mean time outcomes one month pre- and post-intervention. Moreover, we followed the intervention for two additional months to assess for sustainability.

RESULTS: A total 159 patients in July and 152 in August were discharged by the teaching service. The intervention was associated with earlier discharge order initiation by 59min (1:07PM vs. 12:08PM, p = 0.001). This effect was consistent for patients being discharged to home (12:54PM vs. 12:03PM, p = 0.01) and to facility (1:32PM vs. 12:22PM, p = 0.05). The percentage of early discharges, namely discharges initiated before 11AM, increased from 21.4 to 38.8% (p =< 0.001). The intervention was associated with patients leaving the hospital earlier 50min (3:21PM vs. 2:31PM, p = 0.005). This was significant in patients being discharged to home (2:46PM vs. 2:05PM, p = 0.05) but not in those discharged to facility (4:22PM vs 3:46PM, p = 0.28). These beneficial effects, however, waned in the subsequent 2months compared with pre-intervention times. In September and October, the mean discharge order initiation times were 1:00PM (p = 0.69) and 1:11PM (p = 0.82), respectively, while the mean discharge times were 3:22PM (p = 0.94) and 3:42PM (p = 0.22), respectively.

CONCLUSIONS: The systematic implementation of early discharge rounds significantly decreased the time to initiation of discharge orders and subsequently the time of patient discharge in the immediate period after the intervention. Academic internal medicine services may help improve patient throughput by implementing this intervention. However, as in other quality improvement efforts, consistent reinforcement and reevaluation are likely required for sustainability.

DISCORDANT WEIGHT PERCEPTION AND PHYSICAL ACTIVITY IN OVERWEIGHT AND OBESE WOMEN: IS WEIGHT MISPERCEPTION PROTECTIVE?Nathalie Moise¹; Robert Sciacca²; Juviza Rodriguez²; Ian M. Kronish¹; Elsa-Grace V. Giardina². ¹Columbia University Medical Center, New York, NY; ²Columbia University Medical Center, New York, NY. (Tracking ID #1926824)

BACKGROUND: One third of U.S. adult women (35.8%) are obese, placing them at increased risk for cardiovascular disease and mortality. Women are approximately twice as likely to be inactive as men while overweight and obese individuals are less likely to exercise than their normal weight counterparts. Understanding drivers of behavioral change in this population is crucial. Prior studies have suggested that within overweight and obese men, weight underestimation is associated with physical inactivity. However, these studies have relied on self-report, both of weight perception and body mass index (BMI), and have been inconclusive regarding the relationship between weight perception and physical activity in women, whose responses may be particularly influenced by social desirability bias. To address this gap, we assessed whether weight perception measured with a figural rating scale is associated with physical inactivity in overweight and obese women.

METHODS: A cross sectional analysis was conducted on a primary prevention cohort of 580 overweight and obese (BMI ≥ 25kg/m2) women recruited from the primary care clinics of Columbia University Medical Center. Exclusion criteria were known coronary artery disease, cardiovascular disease, stroke, pregnancy and age ≤18years. BMI was calculated from measured heights and weights. The outcome variable was based on American Heart Association recommendations of at least 30min of physical activity on most days, which was subdivided into (a) walking exercise and (b) non-walking exercise. Participants indicated their “self” figure on the Stunkard 9-figure rating scale (FRS), a widely used and validated measure for assessing weight status. Accurate weight perception was determined by comparing participants’ actual BMI with the BMI matched to the selected figure: Underestimators selected underweight or normal weight figures on the FRS and accurate estimators selected overweight or obese figures that matched their BMI. Multivariable logistic regression was used to analyze the relationship between accuracy of weight perception and physical activity after adjusting for age, race, education and BMI.

RESULTS: Mean age was 48 + 14years; 45.3% were obese, 65.3% were Hispanic, 41.6% had hypertension, 30.7% had hyperlipidemia, 13.4% had diabetes mellitus and 28.1% had 2 or more cardiovascular risk factors. Among participants, 49.2% correctly perceived their weight and 82.0% reported they were physically active; 64.6% reported walking exercise and 41.8% non-walking exercise. Those who underestimated their weight, compared to accurate estimators, were more likely to be non-Hispanic black (12.9% vs. 6.7%, p < 0.0001), Hispanic (70.8% vs. 60.0%, p = 0.0001), have 9–12years of education as opposed to >12years of education (40.6% vs. 30.6%, p = 0.02) and be obese (57.8% vs. 32.3%, p =< 0.0001). Those who underestimated their weight reported more physical activity (adjusted odds ratio [AOR] = 1.89, 95% CI, 1.19–2.99, p = 0.0072) and non-walking exercise (AOR = 1.52, 95% CI, 1.05–2.21, p = 0.026), but not walking exercise (Table 1).

CONCLUSIONS: Among overweight and obese women, contrary to our hypothesis, those who underestimated their weights were more likely to report increased physical activity. Accurately perceiving oneself as overweight or obese may be a cognitive barrier to physical activity participation. Additionally, we found that obesity, lower educational attainment and minority race were associated with physical inactivity; physicians should continue to focus on these at risk groups. Because we relied on self-report for physical activity, it may be that individuals who underestimate their weight report greater physical activity. Future studies using objective measures of physical activity are needed to better understand the directionality of this association. Notwithstanding this limitation, clinicians should be aware of weight misperception when counseling overweight and obese patients and should be cautious about reinforcing a self-image that is counter to physical activity.

Adjusted Odds Ratios for physical activity, non-walking exercise and walking exercise.

*p < 0.05 AOR = Adjusted Odds Ratio

DISCRIMINATION PERCEIVED BY FREQUENT EMERGENCY DEPARTMENT USERS. IS DISCRIMINATION A RISK FACTOR FOR VULNERABILITY?Stéphanie Baggio¹; Jean-Bernard Daeppen²; Bernard Burnand²; Olivier Hugli²; Ornella Ruggeri²; Jean-Blaise Wasserfallen²; Karine Moschetti²; Philippe Staeger²; Séverine Alary²; Katia Iglesias²; Marina Canepa Allen²; Patrick Bodenmann². ¹Lausanne University, Lausanne, Switzerland; ²Lausanne University Hospital, Lausanne, Switzerland. (Tracking ID #1930275)

BACKGROUND: Frequent Emergency Department (ED) users are often vulnerable people with many risk factors and health needs. They are also perceived to be time-consuming, “illegitimate” patients, and thus may feel discriminated against during their hospital visit. Because perception of discrimination, non-adherence to medical treatment, and poorer health status are associated, frequent ED users’ perception of discrimination may constitute an additional risk factor for vulnerability. The aim of this study was to investigate the relationship between perceived discrimination and vulnerability among frequent ED users.

METHODS: Data were part of a baseline assessment of a randomized controlled trial designed to improve frequent users’ quality of care in Lausanne University Hospital, one of the two French-speaking Swiss tertiary care hospitals, with over 50,000 annual ED visits. In total, 250 frequent ED users (≥5 attendances during the previous 12months; ≥18years) were interviewed between May 2012 and July 2013. From a previously published questionnaire, 15 dichotomous sources of discrimination, grouped into five causes of discrimination (any cause, immigration-related, age/gender, physical characteristics/sexual orientation, and social/economic status) were assessed. Vulnerability was assessed using health status: objective health status evaluated by health-care practitioners and including somatic-, mental-health-, behavioural-, and social issues (dichotomous variables), and subjective health status self-rated by the frequent ED users, including health-related quality of life (WHOQOL), quality of life (EUROQOL), and satisfaction with care (mean-scores). Associations between perceived discrimination and health status were tested with exact Fischer tests and Mann-Whitney U tests.

RESULTS: Participants were 46.2 ± 18.9years old on average, 57.2% were men, 47.8% Swiss, 17.7% European, and 34.5% non-European. A total of 35.2% reported at least one source of discrimination. Objective health status was not significantly related to perceived discrimination (e.g. 35.8% of users discriminated against had somatic issues versus 33.8% of users not discriminated against, p > 0.05, see Table 1). Contrariwise, experiencing discrimination was associated with worse subjective health status (e.g. users discriminated against reported a mean-score of 28.16 for WHOQOL health status, versus 39.44 for users not discriminated against, p < 0.001, see Table 1).

CONCLUSIONS: Perceived discrimination was partially associated with increased vulnerability. Indeed, frequent ED users who experienced discrimination were more likely to self-rate their own health as lower, while health-care practitioners’ rating was not related to perceived discrimination. Therefore, experiencing discrimination was not related to “real” health status, but to subjective health status’s evaluation. Perceived discrimination seemed to be associated with a negative overall picture of their own health and health care services among frequent ED users, even if their health was not really worse. Health-care practitioners should be aware of this phenomenon, and perceived discrimination may also be of concern in improving interventions and providing optimal care to frequent ED users.

Associations between perceived discrimination and health status

(a) Objective health status — exact Fischer test; subjective health status — Mann-Whitney U test.

DISPARITIES IN GLYCEMIC AND LIPID CONTROL AMONG DIABETIC PATIENTS IN THE VETERANS ADMINISTRATION: IMPLICATIONS IN THE ERA OF THE AFFORDABLE CARE ACTLeChauncy D. Woodard¹; Salim S. Virani¹; Laura A. Petersen¹; Amber Amspoker¹; Tracy Urech¹; David Ramsey¹; Omolola E. Adepoju¹; Praveen Mehta²; Jeffrey A. Murawsky²; Aanand D. Naik¹. ¹Michael E. DeBakey Medical Center, HSR&D, Houston, TX; ²VISN 12 (Great Lakes), Westchester, IL. (Tracking ID #1931717)

BACKGROUND: Racial and gender disparities in diabetes care are well documented. The Patient Protection and Affordable Care Act offers promise for reducing such disparities by increasing access to care and aligning financial incentives to promote high-quality, comprehensive primary care through delivery models such as the patient-centered medical home (PCMH). Given its widespread transformation to the PCMH model of care, the Veterans Health Administration (VA) serves as an ideal setting in which to examine the impact of universal healthcare access and patient-centered, team-based care on disparities in diabetes outcomes. Thus, we assessed racial and gender disparities in diabetes care among patients receiving care in the VA subsequent to PCMH implementation.

METHODS: We used VA structured clinical data from electronic fields to identify patients with diabetes who received primary care between June 1, 2011 and May 31, 2012, in one Midwestern VA Network encompassing 7 facilities. We assessed differences in glycemic and lipid control between male and female Veterans with diabetes using two VA quality indicators [Hemoglobin (Hb) A1c < 9%, low-density lipoprotein cholesterol (LDL-C) < 100mg/dL]. We then used multilevel logistic regression, accounting for facility-level clustering, to determine whether female gender or race was associated with glycemic and LDL-C control.

RESULTS: Of the 203,897 patients who received care during our study interval, we identified 37,644 patients (18.5%) with diabetes. The majority of patients were male (97.1%) and white (73.1%). Male patients were older than female patients (mean age 62.9 vs. 54.4years, p < 0.0001), and were more likely to have co-existing cardiovascular disease (31.6% vs. 12.8%, p < 0.0001) and hypertension (84.5% vs. 73.5%, p < 0.0001), but were less likely than female patients to have co-existing depression (21.2% vs. 38.3%, p < 0.0001). Compared to men, female patients were less likely to achieve glycemic (84.2% vs. 81.0%, p = 0.004) and lipid (72.5% vs. 57.9%, p < 0.0001) control. Adjusting for age, illness burden, comorbidities, receipt of care at a teaching facility, number of primary care visits, and provider type, females were as likely as males to achieve glycemic control (OR 1.07; 95% CI, 0.87–1.33), but were less likely to have controlled LDL-C levels (OR 0.77; 95% CI 0.65–0.92). Further, white patients had significantly higher odds than non-white patients of achieving both glycemic (OR 1.21; 95% CI 1.10–1.33) and lipid (OR 1.29; 95% CI 1.20–1.40) control.

CONCLUSIONS: Despite universal access to care and widespread PCMH implementation within the VA, both gender and racial disparities in diabetes care persist. Of course, these findings could be due to patient preferences or lower cardiovascular risk among females and thus, clinically appropriate. Our findings suggest that additional efforts to adapt the PCMH model of care to diverse populations by accounting for patient characteristics and patient preferences for care may lead to additional gains in eliminating health disparities among chronically ill patients.

DISPARITIES IN THE TREATMENT OF DEPRESSION AMONG LOW-INCOME RACIALLY AND ETHNICALLY DIVERSE PATIENTS WITH TYPE 2 DIABETESCaroline Presley; Ken Wallston; David Schlundt; Chandra Y. Osborn; Sunil Kripalani; Shari D. Barto; Russell L. Rothman. Vanderbilt University Medical Center, Nashville, TN. (Tracking ID #1928731)

BACKGROUND: Comorbid depression among patients with Type 2 Diabetes (T2DM) is common and has been associated with decreased adherence to self-care and worse glycemic control (A1C). There have been few studies examining disparities in the prevalence of depressive symptoms, antidepressant use, and adequacy of treatment among low-income patients with T2DM. Therefore, we examined the prevalence of depression and adequacy of treatment among a low-income, racially and ethnically diverse sample of patients with T2DM.

METHODS: 406 patients with T2DM were enrolled in a randomized controlled trial evaluating a health literacy intervention for diabetes care at 10 Tennessee Department of Health safety net clinics. At enrollment, data on patient characteristics and clinical information were collected via self-report and from the medical record. Depressive symptoms were assessed with the Center for Epidemiological Studies Depression Scale (CES-D 20, range 0–60), with a score of ≥16 indicating possible depression. Patients were considered to be prescribed an antidepressant if their medication list included an SSRI, an SNRI, a TCA, or an atypical antidepressant. Descriptive statistics characterized the sample and identified prevalence rates. Chi-squared analyses and t-tests were used to examine the bivariate relationships between patient characteristics and the presence of depressive symptoms and, separately, depression treatment status. We performed logistic regression analysis of antidepressant use adjusted for age, gender, race, ethnicity, duration of diabetes, education level, level of depressive symptoms (CES-D score), and study intervention status to evaluate the factors impacting antidepressant use in this patient population.

RESULTS: The sample was on average 51years old (SD 9.6), 61% were female, 18% were African American, and 24% reported Latino ethnicity. Mean education was 11.1years (SD 3.4). The majority (88%) reported no health insurance and annual household incomes <$20,000 (83%). Mean A1C was 9.2 (SD 2.1). The majority (53%) had CES-D scores ≥16. There were similar rates of depressive symptoms by race and ethnicity. Despite the high prevalence of depressive symptoms in the overall sample, only 18% of all patients and 24% of patients with CES-D scores ≥16 were prescribed an antidepressant. Depressive symptoms were higher among patients prescribed an antidepressant compared to those not prescribed an antidepressant (CES-D score 25 (SD 14.5) vs. 17 (SD 11.9), p < 0.0001). In adjusted analysis, African American/Blacks (AOR 0.25, p = 0.006) and Latinos (AOR 0.16, p = 0.005) were significantly less likely to be on an antidepressant compared to non-Hispanic Whites.

CONCLUSIONS: We found a high prevalence of depressive symptoms among a vulnerable population with T2DM. Only a fraction of patients with possible depression were prescribed an antidepressant. Among patients who were prescribed an antidepressant, there were high rates of depressive symptoms suggesting ineffective treatment. African American and Latino patients were significantly less likely to be treated with an antidepressant even with adjustment for presence of depressive symptoms. These data suggest that depression is often unrecognized and inadequately treated in low-income patients with T2DM, especially racial and ethnic minorities, which may have a negative impact on their diabetes control and outcomes.

DO CANCER PATIENTS FARE WELL IN THE ICU?Vijairam Selvaraj; Jane Garb; James Stewart; Jay Steingrub; Thomas Higgins. Baystate medical center, Springfield, MA. (Tracking ID #1895999)

BACKGROUND: Physicians and patients may have unrealistic expectations about prognosis with cancer, with both over- and under-estimates of mortality. While severity of illness at ICU admission is frequently evaluated using the APACHE (Acute Physiology and Chronic Health Evaluation) system, these scores are designed to assess large cohorts rather than individuals and may help with prognostication, but cannot be used to triage ICU admission in isolation. There is long-standing controversy over “potentially ineffective care” and evidence for a high degree of regional variation in caring for patients with poor prognosis cancer. However, newer studies suggest reduced mortality rates among critically ill cancer patients with the development of newer procedures such as non-invasive ventilation. Our hypothesis is that subgroups of patients with cancer may have APACHE-adjusted outcomes similar to non-cancer patients.

METHODS: A retrospective cross-sectional study of all patients above 18years admitted to the ICU from May 1, 2009 to April 30, 2013 with chronic health diagnoses of leukemia, lymphoma/myeloma, solid tumor with metastasis and immunosuppression at Baystate Medical Center, Springfield, MA. Exclusion criteria was consistent with APACHE-IV methodology. A control group was selected of patients without an APACHE chronic health diagnosis. Primary outcomes were APACHE-IV score and Standardized Mortality Rates (SMR) during ICU and hospital stay. Secondary outcomes included length of stay in ICU and hospital, and mechanical ventilation on day 1 of admission. Baseline factors in predicting APACHE score were analyzed using multiple regression. Group differences in secondary outcomes were reported descriptively.

RESULTS: 5486 subjects were available for the study: 334 immunosuppressed patients (6%); 91 leukemia patients (2%); 50 lymphoma patients (1%); 162 metastatic cancer patients (3%); and 4849 non-cancer controls (88%). All groups were significantly different from controls in Acute Physiology Scores (APS) (p < 0.001), after adjusting for age. Mean APS scores for immunosuppressed were 74.3, leukemia 83.6, lymphoma 83.7 and metastatic cancers 68.6; controls were 60.3 (p < 0.0001). The age-adjusted regression model accounted for 15% of the variability in APS scores. While APACHE score significantly discriminated diagnosis groups from controls, there were no significant differences between any groups or controls on APACHE-based SMR’s.

CONCLUSIONS: Our study showed patients with cancer or immunosuppression had survival similar to that of non-cancer patients in the ICU; with no difference noted among groups. Because this is a single-site retrospective study, we cannot exclude that triage decisions may be selecting ICU admissions from patients with better prognoses. Further investigation could utilize the national APACHE database to create more generalizeable results. Based on this preliminary work, cancer and immunosuppresed patients do not appear to have a mortality penalty in the context of APACHE severity adjustment. This knowledge can help with ICU resource planning as well as timing for palliative care.

SMR for ICU and Hospital Mortality

DO PHYSICIANS SPEND LESS TIME WITH PATIENTS IN CONTACT ISOLATION? A TIME-MOTION STUDY OF INTERNAL MEDICINE INTERNSCody Dashiell-Earp¹; Douglas Bell¹; Alexis Ang¹; Daniel Uslan^2,1. ¹University of California, Los Angeles, Los Angeles, CA; ²University of California, Los Angeles, Los Angeles, CA. (Tracking ID #1930419)

BACKGROUND: Studies have shown that isolation precautions used by hospitals to control the spread of resistant organisms may have unintended consequences, including a reduction in time spent with health care providers, lower patient satisfaction and more preventable adverse events.

METHODS: The purpose of this study was to measure the time spent by internal medicine interns in contact with patients in isolation compared to non-isolation hospital rooms. Using RFID tags attached to hospital identification badges, we collected real-time data on the location of 15 internal medicine interns working in our hospital between October 1, 2012 and December 31, 2012. These devices enabled us to record the amount of time spent by interns inside isolated and non-isolated patient rooms. In total, there were 1,156 encounters with isolated patients and 2,467 encounters with non-isolated patients. A mixed model was used to compare the average time spent by each intern with these two patient groups.

RESULTS: Interns visited isolated patients less often (2.3 compared to 2.5 visits per day, P < 0.0001) and spent less time per visit with isolated patients (2.2 compared to 2.8min per visit, P < 0001). Thus, on average, interns spent 5.2min per day with each of their isolated patients compared to 6.9min per day with each of their non-isolated patients (P < 0.0001).

CONCLUSIONS: Internal medicine interns spent significantly less time in the rooms of patients in isolation. This difference was more pronounced for some interns compared to others. Reduced physician contact time may be one of the driving forces behind differences in the quality of care experienced by patients in isolation. Further research is needed, both to better define the patient population for whom the benefits of contact isolation outweigh the risks and to develop strategies to ameliorate those risks for those who must be placed into isolation.

DO PRIMARY CARE PHYSICIANS WHO DELIVER BETTER QUALITY OF CARE ALSO USE LESS HEALTH CARE RESOURCES?Carine Yelibi; Charlotte E. Ward; Yuchiao Chang; Jeffrey M. Ashburner; Clemens S. Hong; Steven J. Atlas. Massachusetts General Hospital, Boston, MA. (Tracking ID #1940237)

BACKGROUND: Access to and a long-term relationship with a primary care provider (PCP) are associated with better health outcomes and decreased rates of acute care utilization and costs. However, it is less clear whether PCPs that deliver higher quality care also reduce utilization and costs. We hypothesized that PCPs delivering higher quality care have lower rates of emergency department (ED) utilization, admissions, and readmissions.

METHODS: We performed a physician-level, retrospective cohort analysis, using data from calendar year 2012 from electronic data repositories, on 173 PCPs and 105,766 patients receiving care in the MGH primary care practice-based research network. A quality composite measure was created using nine individual HEDIS measures at the patient level, where the denominator was the number of measures for which the patient was eligible, and the numerator was the number of measures the patient achieved (e.g. test completion or goal attainment). The composite measure was dichotomized based on whether the patient received 100% of services that they were eligible for. The nine HEDIS measures included 3 prevention measures: breast, cervical, and colorectal cancer screening; and 6 chronic disease measures: hemoglobin A1c testing in the prior year and proportion with levels >9.0% in diabetic patients, low-density lipoprotein cholesterol testing in the previous year and proportion with levels <100mg/dL for patients with diabetes or coronary artery disease. The outcomes of interest were the number of visits to the ED, inpatient services and the number of 30-day readmissions during the calendar year 2012. These were assessed at the PCP level as a rate of visits/100 patients. Ranking for each physician was created based on the log-odds of achieving 100% of the HEDIS composite measure using logistic regression models controlling for patient age, gender, ethnicity, language, insurance status, Charlson score, and visit frequency. The relationship between physician quality ranking and utilization outcomes was assessed using linear regression. Rankings were also grouped into quartiles and differences between top and bottom quartiles were assessed using a t-test or Wilcoxon Mann-Whitney test.

RESULTS: Of the 173 PCPs eligible for analysis, 55% were female, and their mean age was 48. We excluded physicians who were in practice for less than 1year and those who left during the year of analysis. The percent of eligible patients in a PCP’s panel who achieved 100% of eligible HEDIS measures was 68% (IQR: 63.6–74.2%). High physician ranking was predictive of lower ED utilization (p < 0.0001), admissions (p = 0.0001) and 30-day readmissions (p = 0.009). When comparing physicians in the top and bottom quartiles of quality ranking, physicians in the top quartile had significantly lower ED utilization rates (13.1 vs 21.1, p < 0.0001). Physicians in the top quartile also had significantly lower admission rates than physicians in the bottom quartile (7.3 vs 9.8, p = 0.007). There was no significant difference in 30-day readmission rates between top and bottom quartile physicians.

CONCLUSIONS: Among primary care physicians within a large, academic network, we found that physicians with higher performance based on a composite of nine quality of care measures was associated with lower ED utilization, admission rates, and 30-day readmission rates. In addition to access to primary care and having a usual source of care, our study suggests that the patients cared for by PCPs with better performance on quality metrics may have lower rates of ED and hospital utilization. Whether patients of such PCPs also have lower health care costs requires further study.

DO RESIDENT-RN ROUNDS REDUCE CALLS TO THE ONCALL PERSON? (D. R. R. R. O. P.): A QUALITY IMPROVEMENT PROJECT IN THE ICUMaria Yballe^1,2; Sehrish Memon^1,2; Sandra Kurup^1,2; Gen Bahrt¹; Jeana Reyes¹; Michael P. Carson^1,2. ¹Jersey Shore University Medical Center, Neptune, NJ; ²Rutgers - Robert Wood Johnson Medical School, Piscataway, NJ. (Tracking ID #1913456)

BACKGROUND: On-call ICU/CCU residents receive many calls from nurses during their overnight shift regarding orders that, in theory, could have been placed by day team residents. These interrupt nursing and physician care, and could distract caregivers from more urgent matters, or contribute to sleep deprivation. The first objective was to determine the number of, and reasons for, calls from nursing to the on-call night resident. The second objective was to identify the most common reasons, modify the process so that the day team would address these orders prior to leaving, and then resample to determine if this educational intervention reduced the number of calls made to on-call residents.

METHODS: Baseline Call Rate: The ICU night resident logged and categorized calls for 20 nights to determine the baseline. Based on the common reasons for calls, we developed a laminated card listing the common issues, distributed it to the nursing and resident staff to use as a reference during end of Day-shift sign-out rounds, and oriented them to the process. Re-Evaluation Call Rate: calls were reassessed 2weeks later for 10 nights. We coordinated with the ICU nurse manager, nurse educator, and staff nurses. T-test was used to compare the Baseline and Re-Evaluation call rates.

RESULTS: Baseline: 592 calls over 20 nights, Mean 29.6+/−8 calls/night, Range 14–46. The top seven reasons were associated with 81% of the calls: electrolytes 35%, medication change 15%, ventilator change 11%, order for routine morning labs 11%, insulin 5%, diet/NPO 3%, NG-tube 3%. Re-evaluation: There were 133 calls over 10 nights; Mean 13+/−4 calls/night (range 5–19) for an average decrease of 16 calls/night. A matched analysis of the 10 lowest nights during the Baseline period vs. the 10 during the Re-evaluation yielded a significant average decline of 10 calls/night [95% C.I. (4,16)]. Comparing the 10 highest Baseline nights to the 10 Re-evaluation yielded an average decrease of 26 calls/night [95% C.I. (14,37)]. Calls/night decreased as follows: Electrolytes drop of 3, medication decreased from 4 to 1, routine labs/imaging from 3 to 1.

CONCLUSIONS: A simple program educating the residents to address common orders at the end of the day shift, and nurses to remind the residents to do so, was effective and decreased calls to the night residents by 55%. This decrease in calls regarding issues that should be addressed by the primary day team was noticed and appreciated by the night resident and nursing staff. The day shift process has been modified, and a repeat survey will be done to assess resident/nursing adherence to this modification. With continued use we anticipate a culture change will result in an additional and sustained decrease in calls for these issues.

DOES RESEARCH TRAINING DURING RESIDENCY PROMOTE SCHOLARSHIP? A CROSS-SECTIONAL ANALYSIS OF A TEN-YEAR COHORT OF THE UCSF-PRIME RESIDENCY PROGRAMMarion Stanley; Jeff Kohlwes; Bridget O’Brien; Rebecca L. Shunk; Denise M. Connor; Patricia A. Cornett; Harry Hollander. VAMC/UCSF, San Francisco, CA. (Tracking ID #1941144)

BACKGROUND: Professional organizations, including the Association of Program Directors in Internal Medicine (APDIM), the Accreditation Council for Graduate Medical Education (ACGME), the Alliance for Academic Internal Medicine Education Redesign Task Force as well as the Carnegie study on medical education call for individualized training approaches to ensure that residents have sufficient experience with the various roles of physicians to make informed career choices. In 2000, the Primary Medical Education (PRIME) track of the University of California, San Francisco (UCSF) categorical residency program implemented a novel clinical outcomes research immersion curriculum to enable residents to experience a research career during residency. In PRIME, residents receive a research training course, mentoring and protected time to complete a project in their area of interest. We examined whether implementation of the PRIME clinical outcomes research track led to increased resident publications, improved mentoring and/or subsequent alumni careers in outcomes research as compared to peers at UCSF.

METHODS: We performed a cross sectional analysis of an email survey of internal medicine residency alumni from the UCSF categorical residency program who graduated from 2001 to 2010. We used Pearson Chi-Square analysis and ANOVA to compare PRIME and non-PRIME alumni on categorical and continuous variables.

RESULTS: From 2001 to 2010, 316 categorical residents became alumni. 211/316 (66%) responded to the survey. A significantly higher percentage of PRIME alumni published research projects they worked on during residency compared to non-PRIME alumni (64% vs. 40%; p = 0.002). A non-significant trend showed PRIME alumni identifying clinical or translational research as their primary career role (35% of PRIME vs. 29% non-PRIME) and spending a higher percentage of their professional time conducting research compared to non-PRIME alumni (26% for PRIME, 21% for non-PRIME). Process measures that could explain these findings include adequate access to mentors (mean 4.4 for PRIME vs. 3.6 for non-PRIME alumni (p < .001)) on a 5 point Likert scale (strongly disagree to strongly agree) and agreeing that mentoring relationships affected career choice (mean 4.2 for PRIME vs. 3.7 for categorical alumni (p = 0.001)). Finally, 63% of PRIME alumni agreed that their research experience during residency influenced their subsequent career choice versus 46% amongst the non-PRIME alumni.

CONCLUSIONS: Our results support that giving residents an opportunity to try out a career in outcomes research during residency enables them to publish their residency-related research more frequently than those who did not have this opportunity. This experience significantly influenced subsequent career choice and may make housestaff more likely to continue doing outcomes research after residency. Implications: Implementation of individualized residency programs tracks that nurture academic interests along with clinical skills will foster important alternative careers within medicine for the graduates who experience them.

EARLY OUTPATIENT FOLLOW-UP AFTER STROKE REDUCES NEIGHBORHOOD SOCIOECONOMIC DISPARITIES IN MORTALITYArleen F. Brown¹; Li-Jung Liang¹; Stefanie D. Vassar¹; Sharon S. Merkin²; Jose J. Escarce¹; Bruce Ovbiagele³; W. T. Longstreth⁴. ¹UCLA, Los Angeles, CA; ²UCLA, Los Angeles, CA; ³Medical University of South Carolina, Charleston, SC; ⁴University of Washington, Seattle, WA. (Tracking ID #1939240)

BACKGROUND: Older adults who reside in disadvantaged communities have higher mortality after stroke. We examined whether the timing of post-stroke outpatient visits contributed to previously observed neighborhood disparities in post-stroke mortality.

METHODS: We used data from the Cardiovascular Health Study, a population-based, longitudinal study of adults ≥65years. Center for Medicare and Medicaid Services (CMS) data were used to obtain the number and dates of outpatient visits and pre-stroke comorbidity. Eligible participants had an incident stroke during the study period, survived through the visit interval (7, 14, 21, or 28days), were matched to CMS records, and had fee-for-service coverage during the follow up period. Neighborhood socioeconomic status (NSES) was a composite of six census variables (median household income; median value of housing units; % of households with interest, dividend, or rental income; % of residents >25years with a high school education; % of residents >25years with a college degree; and % of residents in executive, managerial, or professional specialty occupations). To examine the associations between time to first post-stroke outpatient visit, NSES, and the dependent variable of mortality at 1year post-stroke, we constructed multilevel Cox proportional hazard models that also included age, sex, race, stroke type, comorbidity score, and an interaction term between time to first visit and NSES. Separate models were constructed for each visit interval.

RESULTS: Among the 495 eligible participants (mean follow up 11.5years), 17.4% had an outpatient visit within 7days, 27.9% within 14days, 36.3% within 21days, and 43.6% within 28days. In adjusted models, the mortality hazard at 1year after stroke was lower among participants without, compared to those with, an outpatient visit within 7days (HR = 0.42; 95% CI = 0.23, 0.79), 14days (HR = 0.53; 95% CI = 0.32, 0.85), 21days (HR = 0.52; 95% CI = 0.32, 0.83), and 28days (HR = 0.59; 95% CI = 0.36, 0.95). NSES was associated with mortality only for models of outpatient visits within 28days. The visit x NSES interaction was not significant for any models.

CONCLUSIONS: Early follow up after stroke, specifically a visit within the first 3weeks, was associated with lower mortality at 1year and appeared to mitigate the association between lower NSES and stroke mortality. For those with later follow up, lower NSES remained independently associated with mortality. Early and appropriate clinical care may play an important role in reducing neighborhood socioeconomic disparities in stroke mortality. These findings suggest a role for community-level interventions to support secondary prevention after stroke.

EARLY PALLIATIVE CARE CONSULTATION IS ASSOCIATED WITH IMPROVED SUPPORTIVE CARE CANCER QUALITY FOR VETERANSAnne M. Walling^2,1; Diana Tisnado²; Susan Ettner¹; Steven Asch⁴; Sydney M. Dy⁵; Philip Pantoja²; Martin Lee²; Sangeeta Ahluwalia³; Hannah C. Schreibeis-Baum²; Karl Lorenz². ¹UCLA, Los Angeles, CA; ²VA Greater Los Angeles Healthcare System, Los Angeles, CA; ³Oregon State University, Corvallis, OR; ⁴VA Palo Alto, Palo Alto, CA; ⁵Johns Hopkins University, Baltimore, MD. (Tracking ID #1936329)

BACKGROUND: A growing evidence base suggests that palliative care improves outcomes for patients with life-limiting illness including quality of life and bereaved family members satisfaction with end of life care. Less is known about the processes of care influenced by palliative care that lead to these outcomes in real-world settings. We hypothesized that palliative care consultation is associated with improvement in specific processes of supportive care as measured by the Assessing Symptoms Side Effects and Indicators of Supportive Treatment (ASSIST) quality measures.

METHODS: We studied patterns of non-hospice palliative care consultation and the quality of supportive care using ASSIST process of care measures in a retrospective cohort of 719 patients diagnosed with advanced cancer in 2008 over the period of 3years or until death who received care in the Veterans Affairs Health System (VA) using rigorous chart abstraction methods (inter-rater agreement >90%). We also studied the association of palliative care consultation and quality of supportive care overall and by domain (information and care planning, pain, non-pain symptoms) using a multivariate regression model. Our main predictor variable was the proportion of time (from diagnosis of metastatic disease to death or end of study) a patient with advanced cancer was exposed to palliative care (range 0–1). Our main outcome variable was a patient-level overall quality score calculated with the scores of 40 process quality measures. To take into account that different patients are eligible for different quality indicators with varying pass-rates, we used an observed minus expected score. Using this methodology, a patient’s observed score is weighed against an average hypothetical patient who was eligible for the same quality indicator pattern. We conducted standard OLS regression controlling for variables pre-specified as important in our conceptual model of quality palliative care. To address the potential for treatment selection, we also conducted a sensitivity analysis using doubly robust propensity score methods and found similar results.

RESULTS: Of 719 patients, all but 54 patients died before the end of the study period and 293 received a palliative care consult at least once during the study period. On average, patients who had involvement of palliative care received their first palliative care consult towards the second half of their disease trajectory between diagnosis and death or end of study. Controlling for other variables, patients who receive a palliative care consult at time of diagnosis receive 7% higher quality overall (P < 0.001) and 11% higher quality (p < 0.001) within the information and care planning domain compared to patients who do not receive a consult, with non-significant improvements within the pain and non-pain domains.

CONCLUSIONS: Earlier palliative care leads to a higher process of care quality score in lung, colorectal, and pancreatic cancer, predominantly driven by improvements in the domain of information and care planning. This study supports the effectiveness of early palliative care in three common advanced cancers within the VA.

EARLY WINNERS AND LOSERS IN DIALYSIS CENTER PAY-FOR-PERFORMANCEMilda R. Saunders¹; Marshall Chin². ¹University of Chicago, Chicago, IL; ²University of Chicago, Chicago, IL. (Tracking ID #1938490)

BACKGROUND: In 2012, the Centers for Medicare and Medicaid Services (CMS) began the End Stage Renal Disease Quality Incentive Plan (ESRD QIP), a pay-for-performance program for dialysis facilities. Examination of QIP may be instructive as a case-study of the ability of financial incentives both to improve quality and impact disparities. Here we examine the impact of ESRD QIP for dialysis facilities overall, by facility characteristics, neighborhood demographics and region. We also examine how ESRD QIP had its impact by examining changes in its three clinical performance measures over time.

METHODS: Our data sources were the 2012 CMS ESRD QIP Facility Performance File, which contained total performance scores (TPS)—range 1–30—for most CMS certified facilities (n = 5005) in the US, linked to US Census data by zip code. Each facility’s TPS is based on weighted clinical performance measures for dialysis adequacy (urea reduction rate >65) and hemoglobin outside of the targeted range (% of patients with hemoglobin (Hgb)<10g/dL and % patients with Hgb > 12g/dL). Per QIP, facilities with a TPS less than 26 (of 30) will have their payments reduced on a sliding scale, ranging from 0.5 to 2%. We dichotomized the outcomes as ‘any payment reduction’ versus ‘no payment reduction’, and ‘large payment reduction’ (>1.5%) versus ‘other’ (<1.5%). We used logistic regression to characterize associations between QIP performance and dialysis facility characteristics, neighborhood demographics, and region. To determine how the QIP impacts facility outcomes, we used linear regression to examine changes in dialysis facility outcomes between 2007 and 2010.

RESULTS: Only 30% of facilities will have any payment reduction in 2012. In multivariable analysis, dialysis facilities with any payment reductions were more likely to have more dialysis stations (OR 1.02 per station, 95% CI 1.01, 1.03), longer operation (OR 1.03 per year, 95% CI 1.02, 1.04) and a greater proportion of African-Americans in the neighborhood (lowest versus highest quartile, OR 1.34, 95% CI 1.08,1.65) Only a small proportion of facilities (8%) had a large payment reduction, >1.5%. Facilities with large payment reductions were less likely to be for-profit (OR 0.69, 95% CI 0.53, 0.89) and less likely to be in the South and West (OR 0.65, 95% CI 0.50, 0.85 and OR 0.40, 95% CI 0.27, 0.59, respectively compared to facilities in the Midwest). Decline in the percentage of patients with a hemoglobin >12g/dL was largely responsible for improvement in clinical outcomes between 2007 and 2010. For-profit status and increasing proportion of African-Americans in the neighborhood were associated with greater reduction in percentage of patients with hemoglobin above the targeted range (>12g/dL).

CONCLUSIONS: In the first year of CMS pay-for-performance, a large proportion of dialysis facilities met or exceeded national standards and received no payment reduction. Facilities in African-American communities were more likely to receive a payment reduction despite large improvements in hemoglobin within the targeted range. As the number of outcomes increase and total performance thresholds rise, the quality chasm may widen.

ECONSULTS: CONTENT ANALYSIS OF PCP QUESTIONS, SPECIALIST ADVICE, AND PCP RESPONSESKatherine Wrenn; Nathaniel Gleason; Sereina Catschegn; Marisa Cruz; Ralph Gonzales. University of California, San Francisco, San Francisco, CA. (Tracking ID #1938360)

BACKGROUND: Electronic consults (eConsults) have been proposed as one strategy to help address the excess demand for specialty care, but little is known about the specific types of clinical questions primary care providers (PCPs) are asking through an eConsult program, and how they respond to specialist recommendations. An eConsult service was implemented at the University of California, San Francisco (UCSF) in 2012, in which PCPs can initiate an asynchronous exchange of information with a specialist, posing a clinical question and receiving a response within 3days, without the need for in-person consultation. The objectives of this study were to describe the types of clinical questions PCPs are asking in eConsults, describe the types of advice specialists are providing, and explore to what extent PCPs implement specialist recommendations.

METHODS: We conducted this study at UCSF, a multi-site academic medical center that includes eight adult primary care practice sites and uses a shared electronic health record (EHR). The following medicine subspecialties participated in the first phase of the eConsult program: Cardiology, Endocrinology, Gastroenterology/Hepatology, Pulmonary Medicine, Rheumatology, and Nephrology. We analyzed the narratives of 98 of the first 158 eConsults completed during the first 2months of the program. We categorized each eConsult into the following types of questions and responses: “Diagnosis,” “Treatment,” and/or “Monitoring.” The principal investigator (KW) reviewed all 98 eConsults. A second independent reviewer (MC) coded 15 eConsults and specialist responses using coding instructions created by the first reviewer, to check for inter-rater reliability. Kappa statistics ranged between 0.60 (PCP “Diagnosis”) and 1.0 (PCP “Monitoring” and Specialist “Monitoring” categories). Chart abstraction was performed to determine what proportion of specialists’ recommendations PCPs implemented over the 6months following completion of the eConsult. We report the results of abstractions from the first 50 eConsults.

RESULTS: In our study sample, PCPs asked questions related to diagnosis in 68.4% of cases, treatment in 49.0% of cases, and monitoring in 22.4% of cases. Among specialist responses, 74.5% related to diagnosis, 66.3% to treatment, and 36.7% to monitoring. The types of questions asked by PCPs varied based on the specialty they were consulting. Diagnosis-related questions were dominant (≥80%) for Cardiology, Nephrology, Pulmonary Medicine, and Rheumatology, whereas treatment-related questions were dominant (>70%) for Endocrinology. The majority of eConsults included only one type of question posed by the PCP, whereas specialist responses often provided additional recommendations pertaining to more than one category of response (Figure 1). Based on chart review, we found that PCPs ordered 76.5% of all recommended laboratory tests, 92.3% of recommended imaging tests or procedures, 66.7% of recommended new medications, and 85.7% of recommended medication changes. 24.0% of eConsult patients had a visit with the specialist during the 6-month period following the eConsult, 4.0% had an unrelated emergency department (ED) visit, and 8.0% had a hospitalization, of which 75.0% were unrelated to the condition in the eConsult.

CONCLUSIONS: We find that eConsults include clinical questions across the spectrum of patient care: diagnosis, treatment, and monitoring of multiple different medical conditions. We also find that PCPs implement specialists’ recommendations in the large majority of cases, and very few patients subsequently require specialty care or ED visits related to the reason for eConsult.

Figure 1. Content of eConsult Questions and Responses.

ED VISITS AFTER MEDICAL HOME IMPLEMENTATION IN VA PRIMARY CARE CLINICSKristina M. Cordasco²; Jean Yoon¹; Adam Chow¹; Lisa V. Rubenstein². ¹Palo Alto VA, Menlo Park, CA; ²Greater Los Angeles VA, Los Angeles, CA. (Tracking ID #1940650)

BACKGROUND: Implementation of the VA medical home model aimed, in part, to reduce unnecessary acute care by emphasizing access and continuity within primary care-based teams. We assessed for change in number of ED visits, by different varying types of ED visits, after medical home implementation began in the VA system, and we examined the relationships between primary care clinic-level measures of access and continuity and number of ED visits per patient.

METHODS: We obtained a cohort of 75,532 patients using VA care from fiscal years 2010–2012 in 23 clinics based in three VA medical centers. Patient information on demographics, chronic conditions, and VA ED visits were obtained from outpatient and inpatient files and linked to clinic administrative data over the 3-year period. ED visits were categorized, by discharge diagnosis, into non-emergent, primary care treatable, ED care needed but preventable, ED care needed and not preventable, and mental health-related visits, based on the NYU algorithm. We also obtained annual primary care clinic measures of access (percent of patients receiving primary care within 1day of requested date, waiting time to third next available appointment) and continuity (percent of visits with primary care provider). We compared mean number of ED visits per patient by year and by clinic access and continuity. We conducted multivariable regressions of number of ED visits per patient adjusting for patient and clinic factors with patient random effects using Poisson regression models for each ED visit type.

RESULTS: Although mean clinic measures of access and continuity improved over the 3-year period, ED visits of any categorization increased from 0.55 per patient in 2010 to 0.58 per patient in 2012 among the study cohort (all P < 0.001). In multivariable regressions, clinic measures of higher access predicted significantly fewer ED visits that were both emergent and non-emergent (all P < 0.001) as well as ED visits for psychiatric diagnoses, but was not related to number of ED visits for alcohol and drug use diagnoses. Better clinic continuity predicted fewer ED visits that were non-emergent (P = 0.001) or ED care needed but preventable (P < 0.001), but was not significantly related to mean number of ED visits for any mental health diagnoses.

CONCLUSIONS: Implementation of VA’s medical home was not associated with an overall lower ED visit rate in a primary care cohort although access and continuity measures were reported to be slightly higher by primary care clinics from 2010 to 2012. In adjusted analyses access measures consistently predicted fewer ED visits across types of visits although access did not impact ED visits for substance use diagnoses or for any mental health diagnoses. Clinic continuity was related to fewer ED visits for certain types of ED care but was not related to ED visits for any mental health diagnoses.

EFFECT OF ELECTRONIC HEALTH RECORD IMPLEMENTATION ON INPATIENT WORK PROCESSES OF INTERNAL MEDICINE RESIDENTSMarianne Zachariah^1,2; Erin L. Duffy^1,2; Michael A. Pfeffer¹; Douglas Bell^1,2. ¹UCLA, Los Angeles, CA; ²UCLA, Los Angeles, CA. (Tracking ID #1939481)

BACKGROUND: Electronic health records (EHRs) are rapidly gaining adoption in the U.S. If EHRs increase the time that providers need to spend on documentation and ordering, this might reduce the time available for patient interactions and learning activities. We sought to compare how residents spend their time on inpatient rotations before and after the implementation of a commercial EHR.

METHODS: We shadowed a total of 30 internal medicine residents to observe their usage of time on inpatient general medicine, medical intensive care unit, and coronary care unit rotations, before and after UCLA’s March, 2013 launch of the Epic EHR system. Prior to Epic, most clinical documentation was performed using a home-grown computer system and orders were written on paper. After the launch, nearly all documentation and orders were required to be entered in Epic. Fourteen subjects were shadowed prior to the Epic launch and 27 were shadowed in the period from 30 to120days after the launch (11 were shadowed both periods). Residents were shadowed only on days when they were not covering another team member or taking morning holdover admissions. Shadowing was performed by trained undergraduate students who recorded the tasks that each resident was conducting each minute between the hours of 7am and 5pm. For analysis, certain tasks were grouped together: Write Orders (on paper before Epic, in Epic or on paper afterward); Clinical Documentation (writing admit, discharge and progress notes plus viewing labs, vitals and other patient data); Patient Interaction (time in patient rooms plus hallway conversations with patient or family). Wilcoxon Rank-Sum tests were used to compare time spent on specified task categories.

RESULTS: After Epic implementation, residents spent more time on order writing but not on clinical documentation. The median time spent with patients was 23% lower after Epic implementation but the difference was not statistically significant. Among residents on general medicine, 6 of 11 (55%) attended noon conference before Epic vs. 5 of 19 (26%) afterward, and those who attended spent less time post-Epic (average 51 vs. 26min.). The median times spent receiving and giving didactic teaching were also lower post-Epic, although the difference was statistically significant only for giving didactic teaching. Time spent in social conversation was higher in the post-Epic period. Median times spent in conversation with nurses and other physicians were also higher post-Epic but the differences weren’t statistically significant. The median time spent in conversation with students was lower post-Epic. Among residents who spent any time talking with a student (12 vs. 13), the average time spent was also lower post-Epic (29.9 vs. 19.5min.).

CONCLUSIONS: After EHR implementation, residents spent more time writing orders and less on teaching activities, with additional trends toward less time with patients and more time in conversations. However, the residents were observed after only 1–4months of experience with the system, so these results might not represent their longer-term performance.

Median (IQR) Minutes Spent Before/ After EHR Implementation

EFFECT OF FINANCIAL INCENTIVE FOR COLORECTAL CANCER SCREENING ADHERENCE ON APPROPRIATENESS OF COLONOSCOPY ORDERSThomas B. Morland¹; Marie Synnestvedt²; Steven Honeywell²; Feifei Yang²; Katrina Armstrong³; Carmen E. Guerra². ¹Geisinger Medical Center, Danville, PA; ²Hospital of the University of Pennsylvania, Philadelphia, PA; ³Massachusetts General Hospital, Boston, MA. (Tracking ID #1935049)

BACKGROUND: There is some evidence that financial incentives may help physicians achieve higher rates of preventive health screenings among their patients. However, it is unclear whether these incentives affect the appropriateness of screening tests physicians order. In July of 2010 the University of Pennsylvania Health System implemented a performance incentive for general internists based upon achieving target screening adherence rates for several cancers, including colorectal cancer (CRC). Providers were eligible for $1,000 for achieving a 50% adherence rate and an additional $2,000 for achieving an 80% adherence rate for all applicable tests. The primary objective of our study was to determine whether implementation of the performance incentive was associated with an increase in potentially inappropriate screening colonoscopy orders for patients with life expectancies <4years. We also assessed whether providers with high rates of CRC screening adherence had a higher proportion of colonoscopy orders for patients with life expectancies <4years vs. providers with low rates of screening adherence.

METHODS: Electronic records of visits with participating providers were queried for screening colonoscopy orders during the last year prior to the incentive program (pre-intervention period) and the first year of the incentive program (post-intervention period). Using a previously validated mortality prediction model, orders were classified as “inappropriate” if patients’ 4-year expected mortalities were >50%. A chi-square test was conducted to compare the proportion of orders that were “inappropriate” during the pre-intervention period vs. the post-intervention period. A t-test was also performed comparing the mean risk scores of patients receiving colonoscopy orders during the pre-intervention period vs. the post-intervention period. Logistic and linear regressions were also performed, controlling for age, race, marital status, and gender. In a second analysis, we compared the proportion of “inappropriate” orders for providers with the highest and lowest proportion of screening colonoscopy orders (defined as the top 20% and bottom 20%), respectively.

RESULTS: The study population included screening colonoscopy orders for 1057 patients in the pre-intervention period and 1021 patients in the post-intervention period across 23 providers participating in the financial incentive. Patients were on average 58.03years of age and 61% were female. Only 0.6% (n = 6/1057) of screening colonoscopy orders in the pre-intervention period and 0.6% (n = 6/1021) of screening colonoscopy orders in the post-intervention period were deemed “inappropriate.” There was no significant difference in the mean risk scores or the proportions of “inappropriate” orders between the pre- and post-intervention periods. Linear regression found no effect of time period upon risk score. There was no significant difference between the proportions of orders that were “inappropriate” among orders by high rate providers vs. low rate providers.

CONCLUSIONS: We found no evidence that a performance incentive based upon colonoscopy adherence rate led to a significant increase in inappropriate orders for screening colonoscopies. Our model is limited in that it only identifies orders that are inappropriate due to patients’ age, functional status, smoking, body mass index, and multiple comorbidities.

EFFECT OF THERAPEUTIC SUBSTITUTION ON MEDICATION RECONCILIATION AT HOSPITAL DISCHARGEJessica S. Wang¹; Boback Ziaeian^2,3; Osama Abdelghany⁴; Robert L. Fogerty^5,6; Grace Jenq⁷; Sandhya V. Kanade⁶; Leora I. Horwitz^5,8. ¹Yale University School of Medicine, New Haven, CT; ²David Geffen School of Medicine at UCLA, Los Angeles, CA; ³Jonathan and Karin Fielding School of Public Health at UCLA, Los Angeles, CA; ⁴Yale-New Haven Hospital, New Haven, CT; ⁵Yale University School of Medicine, New Haven, CT; ⁶Yale-New Haven Hospital, New Haven, CT; ⁷Yale University School of Medicine, New Haven, CT; ⁸Yale-New Haven Hospital, New Haven, CT. (Tracking ID #1927862)

BACKGROUND: Although therapeutic substitution, or the substitution of a chemically different but therapeutically equivalent drug for the one originally prescribed, is widely used by hospitals to control inpatient pharmaceutical costs through hospital formularies, the impact of this practice on patients’ discharge medication regimens has not been adequately studied. We determined the frequency at which inpatient therapeutic substitution led to changes in discharge medications and assessed the extent to which it contributed to medication reconciliation errors.

METHODS: We analyzed data from the DIagnosing Systemic failures, Complexities and HARm in GEriatric discharges (DISCHARGE) study, a prospective, observational cohort study of patients 65years or older admitted to a tertiary care hospital for acute coronary syndrome, heart failure, or pneumonia between May 2009 and April 2010 who were discharged home. We examined patients’ medications from six commonly substituted drug classes: proton pump inhibitors, H2 blockers, HMG-CoA reductase inhibitors, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, and inhaled corticosteroids. The occurrence of therapeutic substitution was determined by comparing admission to inpatient medication lists. Medications that were changed to a different drug within the same class at discharge were noted by comparing admission to discharge medication lists. Any medication reconciliation changes at discharge that did not appear intentional based on review of the medical record were classified as suspected errors. We calculated the relative risk of a home medication being switched to a different drug within the same class at discharge among medications that were therapeutically substituted during the hospital stay compared to those that were not. We also calculated the relative risk of a medication reconciliation error between these two groups.

RESULTS: A total of 377 patients enrolled in the DISCHARGE study, collectively taking 2,583 admission medications. We analyzed the 555 (21.5%) admission medications taken by 303 of the patients (80.4%) that were within the six drug classes of interest. Of these, 244 (44.0%) were therapeutically substituted during the hospitalization while the remaining medications were continued unchanged or were held for the duration of the hospital stay. Twenty-eight of the two hundred forty-four therapeutically substituted medications (11.5%), as compared to 8 of the 311 non-therapeutically substituted drugs (2.6%), were changed at discharge to a different medication within the same drug class as the patient’s original home medication (relative risk (RR) 4.46, 95% confidence interval (CI) 2.07–9.61). A total of 41 (7.4%) of the 555 medications of interest had a suspected medication reconciliation error at discharge. 28 (68.3%) of these were medications that were therapeutically substituted while inpatient. Suspected medication reconciliation errors were more likely to occur among medications that had been therapeutically substituted than those that had not been (RR 2.75, 95% CI 1.45–5.19).

CONCLUSIONS: Therapeutic substitution during hospitalization was associated with a significantly higher rate of suspected medication reconciliation errors at discharge as well as a considerable increase in changes to medication type at discharge. This creates a risk for patient misunderstanding and adverse drug events. In light of these findings, the widespread practice of therapeutic substitution may warrant reevaluation.

EFFECTIVENESS OF A BRIEF PEER SUPPORT INTERVENTION FOR VETERANS WITH CHRONIC PAINMarianne S. Matthias; Alan B. McGuire; Marina Kukla; Joanne Daggy; Laura J. Myers; Matthew J. Bair. Roudebush VAMC, Indianapolis, IN. (Tracking ID #1932711)

BACKGROUND: Chronic pain is prevalent, disabling, and costly. Pain reduces quality of life and is associated with emotional distress when it interferes with work, social and recreational activities, and family life. This pilot study tested a peer support program, involving peer delivery of pain self-management strategies, for veterans with chronic pain.

METHODS: This was a pre-test/post-test design with a 4-month intervention period. Measures were administered at baseline and at 4months. Ten peer coaches and 20 patients were recruited. All had chronic musculoskeletal pain. Peer coaches had participated in a prior chronic pain study involving pain self-management. Patients were recruited from a VA medical center and had at least moderate pain severity, defined by pain ≥5 on a 0 (no pain) to 10 (worst pain imaginable) scale. After a 3-hour training session, peer coaches were each assigned two patients. Coaches and patients were matched on age and pain location to the extent possible and were instructed to meet face-to-face or via phone biweekly for 4months. Coaches and patients were given a study manual that included sections on chronic pain basics, activity pacing, relaxation skills, and self-care skills. Peer coaches were encouraged to discuss different topics, guided by the manual, at each meeting, as well as to work on goal setting and provide support, motivation, and encouragement as appropriate. Coaches participated in regular supervision calls, which reinforced training, taught additional motivational strategies, and provided a forum to discuss questions or concerns. Pain was the primary outcome and was assessed with the PEG, a 3-item version of the Brief Pain Inventory, and the PROMIS Pain Interference Questionnaire. The following secondary outcomes were assessed: depression (PHQ-9), anxiety (GAD-7), self-efficacy (Arthritis Self-Efficacy Scale), patient activation (PAM), perceived social support (Multi-Dimensional Perceived Social Support Scale), and two measures of negative pain cognitions: pain catastrophizing (Pain Catastrophizing Scale), and pain centrality (Centrality of Pain Scale). A linear mixed model with a random effect for peer coaches was applied to the change scores (post - pre measures). This accounts for the intra-class correlation (patients nested within peer coaches). Outcomes are reported for patients only.

RESULTS: Nine peer coaches and 17 patients completed the study. All were male veterans. Patients’ ages ranged from 35 to 66 (M = 58, SD = 8) years. Peer coaches’ ages ranged from 50 to 71 (M = 60, SD = 7) years. Patients’ pain conditions were as follows: low back (n = 8), neck (6), knees (1), shoulders (1), “everywhere” (1). Patients’ pain improved but did not reach statistical significance (p = .33, ICC = .28, Cohen’s d = −.25 for PEG; p = .17, d = −.35 for PROMIS). Depression showed little improvement (p = .47, d = −.17). Anxiety (p = .11, d = −.36), self-efficacy (p = .16, ICC = .56, d = .60), patient activation (p = .12, ICC = .40, d = .49), perceived social support (p = .11, d = .37), pain catastrophizing (p = .12, d = −.42), and centrality of pain (p = .06, ICC = .32, d = −.62) all improved.

CONCLUSIONS: This study suggests that peers can effectively deliver pain self-management strategies and support to other veterans with pain. Although this was a pilot study with a relatively short intervention period, patients showed improvement in several outcomes. In particular, effect sizes were moderate for improved self-efficacy, patient activation, and pain centrality. Although pain did not significantly improve, there was a modest change in the expected direction, suggesting that a longer intervention, which allows patients more time to adopt and benefit from the self-management strategies learned, may result in a larger effect on pain, as well as greater effects on psychosocial outcomes.

EFFECTIVENESS OF A HEART FAILURE EDUCATION INTERVENTION IN A LOW LITERACY PATIENT POPULATIONShing-Yu Lin; Lori Randall; Mary E. Wiles; Yoo Mee Shin; Michelle Edwards; Diane Wirth; Nurcan Ilksoy; Robin Klein. Emory University, Atlanta, GA. (Tracking ID #1923870)

BACKGROUND: High disease burden and low literacy among patients can make educational interventions important and challenging. The objective was to evaluate the effectiveness of a heart failure education intervention in a low literacy patient population.

METHODS: We conducted a prospective study to evaluate the effectiveness of a heart failure education intervention in a low literacy patient population. Patients completed a pretest prior to the education while inpatient and a subsequent posttest at follow-up clinic visit 1 to 2weeks later. We included all adult inpatients with new or worsening heart failure newly referred to the Grady Heart Failure Program from November 2012 to April 2013. Outcomes included self-reported education level and employment status, knowledge of and confidence with heart failure self-management, and 30-day and 6month readmissions.

RESULTS: Fifty-one patients agreed to participate and complete data including pretest and posttest was available for 34 patients. Heart failure education led to a significant increase in patients’ mean knowledge scores (77.6% to 82.8%, p = 0.04). Knowledge scores did not correlate with employment status or education level. Confidence scores increased significantly with the educational intervention. Highly confident patients had significantly higher knowledge scores than less confident patients on the pretest and posttest (p = 0.04 and p = 0.01 respectively). There was no significant difference in knowledge or confidence scores between patients readmitted within 30days or 6months of discharge and those not readmitted. There was an overall decrease in heart failure readmissions after the implementation of the heart failure program, including the education program.

CONCLUSIONS: Educational intervention geared towards a low literacy population is effective in improving patients’ knowledge of and confidence with managing heart failure. This suggests there is benefit to expanding education programs in patient populations with low literacy and high disease burden.

EFFECTIVENESS OF TARGETED PHONE CALLS TO REDUCE NO-SHOWS IN A HOSPITAL-BASED PRIMARY CARE CLINIC: A RANDOMIZED CONTROLLED TRIALSachin J. Shah¹; Clemens S. Hong¹; Patrick R. Cronin²; Benjamin I. Bearnot¹; Calvin A. Richardson³; Blair W. Fosburgh¹; Alexandra Kimball⁴. ¹Massachusetts General Hospital, Boston, MA; ²Massachusetts General Hospital, Boston, MA; ³Massachusetts General Hospital, Boston, MA; ⁴Massachusetts General Hospital, Boston, MA. (Tracking ID #1934774)

BACKGROUND: “No-shows”, or missed outpatient appointments, fragment continuity of care, effectively decrease access to primary care for all patients within the practice, and decrease productivity for primary care clinicians. Patient Service Coordinators (PSCs) are the first contact for patients with the office both in person and via phone or electronic communication. They schedule appointments, facilitate referrals, and perform outreach for patients overdue for appointments or screening tests and may be important sources of contact in preventing no-shows. We aimed to assess the effectiveness of a targeted phone call to a population predicted to be at high risk of “no-shows.”

METHODS: We conducted a randomized controlled trial to reduce no-shows at the Internal Medicine Associates (IMA), a large, academic, hospital-based primary care practice staffed by 89 residents and 131 attending physicians. Approximately 320 patients visit the IMA daily. We included all patients older than 18years old predicted to be at high risk to no-show for an appointment scheduled between April and October 2013. We estimated risk that a patient would no-show using a predictive model that included age, insurance status, prior history of no-shows, appointment type and wait days (days between the scheduling and appointment date) as predictors. We randomized high-risk patients (defined as those with >15% absolute risk) to usual practice (control) or usual practice plus a phone call from a PSC (intervention) using their medical record number. To encourage patients to arrive for appointments, routine practice for the IMA is to mail a letter 2–6weeks prior to the appointment and make an automated phone call 3days prior to the appointment. Patients randomized to the intervention appeared in the queue of an online calling tool 7days prior to their appointment. PSCs completed calls in the queue daily and recorded outcomes of the call in the online calling tool. Patients would remain in the calling queue until the call was recorded as complete or until the date of their appointment passed. We trained PSC, over 4weeks, to use the web-based calling tool in both group sessions and one-on-one sessions. PSCs also underwent a brief training in a conversational technique designed to engage patients in concrete planning. We instructed PSCs to remind the patient of the appointment, ask if they will make the appointment and problem solve any barriers that would prevent them from making the appointment. We compared our main outcomes: 1) no-show rate 2) percentage of cancelled appointments slots in which another patient was seen using chi-square test. We performed analyses on an intention-to-treat basis.

RESULTS: We randomized 4055 high-risk patients into intervention (n = 2117) and usual practice (n = 1938) groups. PSC completed 1622 of the 2117 (76.6%) assigned calls prior to patients’ appointments. In the intervention arm 325 (15.4%) of 2117 patients no-showed for their appointment, and in the control arm 395 (20.4%) of 1938 patients no-showed for their appointment (RR 0.75; 95% CI 0.66–0.86; p < 0.001). The percent of canceled appointment slots in which another patient was scheduled and seen, was no different between the intervention and control groups (24.4% vs 22.6%, respectively; RR 1.08; 95% CI, 0.87–1.33; p = 0.48).

CONCLUSIONS: A targeted program of phone calls from a patient service coordinator designed to engage the patient in concrete planning significantly reduced no-show rates among patients at high risk for no-shows. However, when appointments were canceled, the intervention did not lead to a significantly higher rate of successfully rescheduled appointments. Phone interventions designed to address barriers and reduce no-shows among patients at high-risk, may reduce no-show rates while improving continuity of care, access to primary care, and productivity for primary care clinicians.

EFFECTIVENESS OF USING NON-CLINICIANS IN DELIVERING A BRIEF SMOKING CESSATION INTERVENTION IN THE EMERGENCY DEPARTMENTEllie Grossman¹; Maria I. Duenas¹; Ashley Colucci²; Renee Fruchter³; Binhuan Wang³. ¹New York University, New York, NY; ²New York University, New York, NY; ³New York University, New York, NY. (Tracking ID #1937034)

BACKGROUND: Tobacco remains a leading preventable cause of morbidity and mortality. Prior studies have shown that Emergency Department (ED) patients typically smoke at rates exceeding that of the general population, are interested in quitting, and often have limited access to primary care. However, a busy ED can be a difficult setting to implement preventive-health interventions. In this study, we describe a pilot initiative using non-clinician volunteers to deliver a brief smoking cessation intervention to ED patients who smoke. We focused on providing minimal counseling in the ED and instead encouraging post-visit Quitline contact for high-quality telephone counseling and delivery of cessation medication.

METHODS: We trained a cohort of college-aged volunteers to assess adult patients for use of tobacco products in the Bellevue Hospital ED and proactively refer interested smokers to the NYS Quitline. All patients were offered printed information about resources to help them quit smoking. We collected information about the number of patients approached, the number who reported using tobacco and nicotine products, and acceptance of referral to the NYS Quitline. We collected follow-up reports from the NYS Quitline and also attempted to contact all smokers at least once via phone approximately 1month after ED visit. These follow-up reports provided information about use of smoking cessation pharmacotherapy, quit attempts, and smoking status. We performed bivariable and multivariable analyses to explore for predictors of quitline referral, successful access of quitline services, and smoking status at 1month follow-up.

RESULTS: During the two-month intervention period in summer 2013, we screened 1619 adult patients for tobacco use. 428 (26.4%) had smoked a cigarette in the past 30days. Mean age was 42.6years (SD 12.4), and 72% were male. English (85%) and Spanish (8%) were the preferred languages. Other nicotine product use in the past 30days included electronic cigarettes (8.3%), cigars (4.6%), and pipes (1.7%). Among cigarette users, 40% reported smoking <5 cigarettes/day, 32% 5–10 cigarettes/day, 21% 11–20 cigarettes/day, 5% 21–30 cigarettes/day, and 3% >30 cigarettes/day. Two-hundred fifty patients accepted a flyer with information about quit-smoking resources, and 139 (56% of the smokers with phones) agreed to participate in the Quitline proactive outreach program. The NYS Quitline reported follow-up data for 99 patients, and of these, 25 (25%) had completed a Quitline telephone encounter. The NYS Quitline reported that 15 (60%) had already quit smoking, 9 (36%) wanted to quit smoking, and 2 (8%) were not ready to quit yet. Our program reached 78 patients for 1-month follow-up, and 25 (32%) reported abstinence from tobacco for the previous 7days. Forty-five patients (58%) reported a quit attempt lasting longer than 24h since their ED visit, and 11 (14%) had used a smoking cessation medication. Smokers who smoked more cigarettes per day (11–20 as compared to <5) were more likely to accept quitline referral (OR 2.53, 95% CI 1.16–5.53); there were no significant associations between age, gender, language, or e-cigarette use and acceptance of quitline referral. Non-English speakers were more likely than English speakers to successfully complete a Quitline outreach call (OR 3.07, 95%CI 1.00–9.42, p = .05). There were no significant associations between age, gender, use of e-cigarettes, or amount of cigarettes smoked and likelihood of successful contact by Quitline. At one-month follow-up, participants who reported smoking 5–20 cigarettes/day at baseline were less likely to have quit than those smoking <5 cigarettes/day (p < .05).

CONCLUSIONS: Trained volunteers are able to deliver a brief smoking cessation intervention in the ED and can feasibly refer patients to the Quitline. This brief intervention, consistent with evidence-based guidelines for tobacco dependence treatment, is acceptable to patients undergoing treatment in a busy urban safety-net hospital ED.

EFFECTS OF 2010 ACGME REGULATIONS ON INTERNAL MEDICINE RESIDENCY PROGRAMS IN THE NORTHEASTOmar Mousa; Rushikesh Shah; Samana Zaidi; Amit S. Dhamoon. SUNY - Upstate Medical University, Syracuse, NY. (Tracking ID #1939563)

BACKGROUND: To study the variables that influence the perceptions of the 2010 Accreditation Council for Graduate Medical Education (ACGME) work hour regulations (WHR) by the housestaff (HS), program directors (PD) and faculty physicians (FP) within the residency training programs of internal medicine.

METHODS: An online questionnaire was distributed among 131 residency training programs in the northeast of the US. Study subjects included the HS, PD and FP in the department of internal medicine. Questionnaires were completed anonymously. Our endpoint focused on analysis of the variables that influence how the 2010 ACGME/WHRs are perceived by HS, PD and FP. Data analysis was conducted using SPSS Statistics v19.

RESULTS: 203 internal medicine HS, PD and FP were included (158 residents, 24 FP, 11 PD). 87.2% were of ages 20–40years; 57.1% Males; 41.9% US medical graduates (US-MG). Many respondents agreed that the 2010 ACGME/WHR negatively affected the continuity of patient care (48.8%), and had a positive impact on the residents’ quality of life while not at work (69%). Among the different levels of training, there was a significant difference in the perceptions regarding patient care (p < 0.05), residency education (p 0.004) and continuity of care (p 0.001). Those of ages >40years feel that residency education and the care of patients is impacted adversely (p 0.012) compared to their younger colleagues. The younger age groups (<=40years) were more likely to feel that their programs do not strictly follow the ACGME regulations (p 0.016). However, they were more satisfied with the rules (p 0.033). Compared to international medical graduates (IMG), the US-MGs believe more that residency education (36.5% vs 28%) and patient care (40% vs 30.5%) worsened. More US-MGs believe that the continuity of care is affected (p 0.011). 58.1% feel that the residents and interns work under pressure to leave the hospital on time due to the WHRs. Higher rates of IMGs (51.7%) feel that residents do not work under pressure to leave the hospital on time, while 71.8% of the US-MG feel that they do (p 0.001). Satisfaction among HS, PD and FP is significantly variable (p 0.003).

CONCLUSIONS: The 2010 ACGME WHRs were implemented to enhance patient safety by improving the quality of life of the resident. Several unintended consequences of this change, including difficulties with continuity of care, patient safety, and residency education have resulted. The perception of these changes is influenced by the age, gender, and previous medical training of the respondents to our survey. These variables should be addressed by the ACGME when future changes to the resident work hour regulations are made. Such changes have a great impact on the current clinical practice.

EFFECTS OF MINIMAL VS. INTENSIVE INTERVENTION TO ENHANCE MOTIVATIONAL INTERVIEWING IN HIV CAREMary Catherine Beach¹; Michael B. Laws²; Gary S. Rose¹; Debra L. Roter¹; Richard D. Moore¹; Geetanjali Chander¹; Ira Wilson². ¹Johns Hopkins University, Baltimore, MD; ²Brown University School of Public Health, Providence, MA. (Tracking ID #1938857)

BACKGROUND: Behavior change counseling has become an essential component of primary care, but many clinicians lack the skills to counsel patients effectively. Motivational interviewing (MI), a client-centered counseling method of eliciting motivation for change, has been widely endorsed as a framework for clinicians to use in counseling patients. MI may be particularly helpful for clinicians providing HIV care, who counsel patients frequently on issues such as adherence, safe sex, and substance use treatment. As MI programs are rolled out, it is important to evaluate how to efficiently and effectively deliver MI training to busy clinicians. We conducted a randomized trial comparing the effect of two different levels of MI training on clinician communication behaviors and patient experiences.

METHODS: We enrolled 12 HIV clinicians (eight physicians, three nurse practitioners, and one physician assistant) at a single academic medical center. All clinicians attended a one-day workshop with an experienced MI trainer, focusing on behavior change counseling skills. After the workshop, we randomized clinicians to receive (or not) 3 to 5 rounds of personalized one-on-one feedback from the MI trainer, using audio-recordings of the clinician’s own visits with patients as the basis for feedback. We measured clinicians’ attitudes towards and use of MI counseling techniques at baseline and again after completion of the intervention. We also administered post-visit Healthcare Climate Questionnaires assessing patient perceptions of the degree to which the clinician adhered to the spirit of MI. Audio recordings were analyzed using the Roter Interaction Analysis System to evaluate MI-consistent communication behaviors. We compared outcome measures before and after the interventions and between the two intervention groups (workshop alone vs. workshop plus feedback). To assess whether one intervention was more effective than the other, we tested time-by-study arm interactions to determine whether one group improved more than the other. For all analyses, we used generalized estimating equations to account for clustering of patients within clinicians, with Gaussian or negative binomial distributions as appropriate.

RESULTS: Clinicians in both intervention groups reported greater use of MI-consistent behaviors after vs. before the intervention, without differences between groups in the magnitude of the effect. Patients also rated their visits as more MI consistent (6.86 vs. 6.65, p = 0.004), and audio recorded analysis revealed that visits were more patient-centered (1.34 vs. 0.96, p = 0.003), with a more positive patient affect (22.36 vs. 20.84, p < 0.001), after vs. before the intervention, without differences between intervention groups. Analysis of audio-recorded visits revealed increases in clinician behaviors such as empathic statements and asking patient permission, and reductions in disapproval, after vs. before the intervention, again without differences by intervention group. Some clinician behaviors such as asking patient opinions and the ratio of open to closed-ended questions improved to a greater extent in the workshop plus feedback vs. the workshop only intervention arm.

CONCLUSIONS: The workshop alone was as effective as the workshop plus feedback intervention in improving patient experiences and overall communication measures. Certain communication behaviors improved to a greater extent with the more intensive intervention, but these additional improvements may not be substantial enough to warrant the extra financial and logistical resources required.

EHR USABILITY BURDEN AND ITS IMPACT ON PRIMARY CARE PROVIDERS WORKFLOWZia Agha^1,2; Alan Calvitti³; Shazia Ashfaq³; Neil J. Farber²; Richard L. Street⁴; Kristin Bell¹; Lin Liu¹; Mark Gabuzda¹; Yunan Chen⁵; Barbara Gray³; Steven Rick³. ¹VA San Diego Healthcare System, San Diego, CA; ²Univ. of California San Diego, San Diego, CA; ³Veterans Medical Research Foundation San Diego, San Diego, CA; ⁴Texas A&M, College Station, TX; ⁵Univ. of California Irvine, Irvine, CA. (Tracking ID #1940994)

BACKGROUND: Electronic Health Records (EHRs) often integrate poorly with clinical workflow and suffer from poor usability. In this paper, we describe how primary care clinicians’ EHR use patterns relate to usability and how these patterns suggest multiple inefficiencies and burdensome workflow for primary care providers.

METHODS: This time-motion study is based on video and EHR activity capture. We observed 21 clinicians and 111 established patients during primary care office visits at 4 Veterans Administration (VA) clinics. Data were coded for specific EHR tasks and clinical workflow. Hierarchical and sequential analysis of EHR clickstream and clinical workflow were integrated to provide objective baseline use patterns.

RESULTS: The 111 outpatient follow-up visits (~60h of observation) were analyzed in terms of time-at-task in clinical workflow (Figure 1A). Clinicians spent 42% of visit time with EHR tasks compared to 35% with patients. We observed a median 158 mouse clicks per visit and for every 100 mouse clicks (Fig 1B), visit duration increases by 6min (95% CI = 0.05–0.07, p < 0.0001). Clinicians frequently multitask when using EHRs and navigate across multiple functions (median 19 times per visit). Notes (40%) and Orders (27%) were the most frequently used functions. Menu and form driven functions like Consultation (15.8 clicks/unit), imaging (14.5 clicks/unit), and medication orders (9.4 clicks/unit) required burdensome user input.

CONCLUSIONS: EHR activity consumes a majority of visit time and shifts the clinician’s focus away from the patient. EHRs also introduce multitasking, which has been associated with medical errors, clinician dissatisfaction, and missed opportunities for physician-patient communication. While some time-and-motion studies have reported little or no change in visit duration for EHR-based versus paper-based visits, we observed that higher EHR activity is associated with longer visits. These associations while not conclusive, provide partial validation of physicians’ concerns that poorly designed and implemented EHRs can lower clinical productivity. Our study highlights unintended inefficiencies introduced in clinical work due to poorly designed EHR user interfaces and emphasizes the need to address EHR usability as a key area for improvement to support meaningful use.

ELECTRONIC HEALTH RECORDS IMPROVE CLINICAL NOTE QUALITYHarry B. Burke¹; Albert Hoang¹; Dorothy Becher¹; Paul Fontelo²; Ronald W. Gimbel¹. ¹Uniformed Services University of the Health Sciences, Bethesda, MD; ²National Library of Medicine, Bethesda, MD. (Tracking ID #1938123)

BACKGROUND: Electronic health records (EHR) are being adopted throughout the United States yet it is not known whether they improve the quality of clinical notes. We hypothesized that electronic health records would improve the quality of clinical notes.

METHODS: This five and one-half year blinded longitudinal, retrospective multicenter study compared the quality of handwritten and electronic outpatient clinical notes for the same cohort of 100 patients with type II diabetes mellitus at three time points: 6months prior to the introduction of the EHR (before-EHR), 6months after the introduction of the EHR (after-EHR), and 5years after the introduction of the EHR (5-years-EHR). QNOTE is a quantitative validated instrument that assesses the quality of outpatient clinical notes. (JAMIA online 1/2/14) Its 12 evaluative elements are: chief complaint, history of present illness, problem list, past medical history, medications, adverse drug reactions & allergies, social & family history, review of systems, physical findings, assessment, plan of care, and follow-up information. QNOTE scores can range from a low of 0 to a high of 100. Sixteen primary care physicians with active practices used QNOTE to determine the quality of the 300 clinical notes (100 patients with 3 notes each).

RESULTS: All 12 element quality scores significantly improved in the interval between before-EHR and 5-years-EHR. The grand mean QNOTE scores were: 6months before the introduction of the EHR (handwritten), 52.0 (SD 18.4), 6months after EHR, 61.2 (SD 16.3), and 5years after EHR, 80.4 (SD 8.9). The grand mean QNOTE scores significantly improved after the introduction of the EHR, comparing before-EHR to after-EHR, p < 0.0001, and before-EHR to 5-years-EHR, p < 0.0001.

CONCLUSIONS: The introduction of the electronic health record significantly improved the quality of clinical notes.

QNOTE scores (mean, SD)

ELECTRONIC RESOURCE USE AND PRIMARY CARE CAPACITY AS INSURANCE COVERAGE EXPANDSRenuka Tipirneni^1,2; Hwajung Choi^1,2; Matthew M. Davis^3,1. ¹University of Michigan, Ann Arbor, MI; ²University of Michigan, Ann Arbor, MI; ³University of Michigan, Ann Arbor, MI. (Tracking ID #1934522)

BACKGROUND: As millions of Americans gain health insurance coverage offered by the Patient Protection and Affordable Care Act (ACA), newly insured individuals will face the challenge of finding a primary care physician (PCP). Given the limited supply of PCPs in the U.S., state and local health planners must consider how to increase PCPs’ capacity to accept new patients. Use of electronic resources, or “meaningful use”, has been widely viewed as providing potential savings in practice costs and time. However, it remains unclear whether electronic resource use can improve practice efficiency and, ultimately, expand capacity of primary care physicians to accept the newly insured into their panels. We examine whether increased use of electronic resources is associated with change in primary care capacity.

METHODS: We analyzed data from the Center for Healthcare Research & Transformation 2012 Survey of Michigan Physicians, a cross-sectional survey of Michigan PCPs conducted between October and December 2012. Surveys were mailed to 1500 PCPs, 500 from each of three specialties (pediatrics, internal medicine and family medicine). Survey items included questions about the primary predictor variable, PCPs’ use of electronic resources; this variable was operationalized for each of several resources individually (electronic health records, patient registry, electronic prescribing, web portal for appointment scheduling or prescription refills, reminder system, care improvement registry and electronic access to admitting hospital records) and in aggregate (total count of electronic resources in use). The primary outcome was PCPs’ assessment of their future capacity (yes/no) to accept new patients overall, and by payer (e.g. private insurance, Medicaid, and Medicare). We conducted bivariate analyses between primary predictors and outcomes. We then performed multivariable logistic regression analysis to examine the relationship between use of electronic resources and primary care capacity, after controlling for physician characteristics (gender, specialty, number of years in practice, panel volume; all by self-report), practice factors (practice size, current insurance payer mix; all by self-report), and urbanicity of the PCPs’ practice setting (urban vs. suburban vs. rural county; by linking Federal Information Processing Standard county codes obtained from self-reported zip code to the U.S. Department of Agriculture Economic Research Service 2013 Urban Influence Codes).

RESULTS: The study sample included 739 physicians (response rate = 49%). Overall, 83% of PCPs reported they would have capacity to accept new patients in the future. The most common electronic resource in use at the time of the survey was electronic prescribing (89% of PCPs); the least common resource in use was a web portal for patients to schedule their own appointments (21%). In aggregate, PCPs had a mean of 5.1 electronic resources in their practices. In bivariate analyses, with each incremental increase in the number of electronic resources in use, there were significantly lower odds of future primary care capacity (OR = 0.87, 95%CI 0.78–0.97). After adjusting for physician-, practice- and community-level covariates, incrementally greater numbers of electronic resources in use remained significantly inversely associated with anticipated capacity (adjusted OR = 0.86, 95%CI 0.76–0.97). Among individual electronic resources examined, electronic health records (EHR) were the only resource significantly associated with anticipated capacity (adjusted OR 0.53, 95%CI 0.30–0.94). In analyses of future capacity by payer type, increased use of electronic resources overall (adjusted OR = 0.87, 95%CI 0.77–0.98), and use of EHR specifically (adjusted OR = 0.50, 95%CI 0.29–0.88), were associated with significantly decreased capacity to accept privately insured patients. There were no associations with future capacity to accept Medicaid or Medicare patients.

CONCLUSIONS: In this study of primary care capacity in a large state in the era of insurance coverage expansion, PCPs who were most likely to implement EHR and use a greater number of electronic resources were significantly less likely to accept new patients. The implication of these findings for patients obtaining coverage under the ACA is that access to primary care may be somewhat impeded, rather than facilitated, by factors associated with electronic resource implementation.

ELEMENTS OF TEAM-BASED CARE IN A PATIENT-CENTERED MEDICAL HOME ARE ASSOCIATED WITH LOWER BURNOUT AMONG VA PRIMARY CARE EMPLOYEESChristian Helfrich¹; Joseph Simonetti¹; Robert Reid⁶; Sandra Joos³; Bonnie J. Wakefield⁴; Ian A. Randall¹; Gordon Schectman⁵; Richard Stark⁵; Stephan D. Fihn^2,1; Henry B. Harvey¹; Karin M. Nelson¹. ¹VA Puget Sound Health Care System, Seattle, WA; ²Veterans Health Administration, Seattle, WA; ³Department of Veterans Affairs, Portland, OR; ⁴Department of Veterans Affairs, Iowa City, IA; ⁵Veterans Health Administration, Washington, DC; ⁶Group Health, Seattle, WA. (Tracking ID #1938477)

BACKGROUND: A high proportion of the US primary care workforce reports burnout, which is associated with negative consequences for clinicians and patients. Many protective factors from burnout are characteristics of patient-centered medical home (PCMH) models, though even positive organizational transformation is often stressful. The existing literature on the effects of PCMH on burnout is limited, with most findings based on small-scale demonstration projects with data collected only among physicians, and the results are mixed. To determine if components of PCMH related to team-based care were associated with lower burnout among primary care team members participating in a national medical home transformation, the VA Patient Aligned Care Team (PACT).

METHODS: We conducted a web-based, cross-sectional survey linked to administrative data from May 2012. A total of 4,539 VA primary care personnel from 588 VA primary care clinics. The dependent variable was burn-out, and the independent variables were measures of team-based care: team functioning, time spent in huddles, team staffing, delegation of clinical responsibilities, working to top of competency, and collective self-efficacy. We also included administrative measures of workload and patient comorbidity.

RESULTS: Overall, 39% of respondents reported burnout. Participatory decision making (OR 0.65, 95% CI 0.57, 0.74) and having a fully staffed PACT (OR 0.79, 95% CI 0.68, 0.93) were associated with lower burnout, while being assigned to a PACT (OR 1.46, 95% CI 1.11, 1.93), spending time on work someone with less training could do (OR 1.29, 95% CI 1.07, 1.57) and a stressful, fast-moving work environment (OR 4.33, 95% CI 3.78, 4.96) were associated with higher burnout. Longer tenure and occupation were also correlated with burnout. Neither spending time on work for which one has too little training, nor measures of workload were associated with burnout.

CONCLUSIONS: Lower burnout may be achieved by medical home models that are appropriately staffed, emphasize participatory decision making, and increase the proportion of time team members spend working to the top of their competency level. Ensuring that primary care team members are tasked with appropriately challenging work appears to be a greater issue for burnout than work that exceeds one’s training or the overall workload.

EMPATHY, CAREER INTERESTS, AND FACTORS INFLUENCING CAREER CHOICE AMONG MEDICAL STUDENTSEmma C. Jeffries; Charlotte E. Ward; Marya J. Cohen. Massachusetts General Hospital, Boston, MA. (Tracking ID #1938468)

BACKGROUND: An initial study indicates that high levels of physician empathy correlate with positive patient outcomes. However, in previous literature, medical student empathy scores have been shown to drop over the course of education despite increasing exposure to patient care, particularly in the inpatient setting. Empathy scores in students exposed to mentorship, educational opportunities, and longitudinal patient care in the outpatient setting have not been studied. At the Crimson Care Collaborative (CCC), five student-faculty collaborative practices across the Harvard teaching hospitals, students have the opportunity to care for and develop long-term relationships with primary care patients. Our goal was to examine the relationship between student empathy, future career interests, and factors influencing career choice.

METHODS: Prior to joining CCC, Harvard Medical School (HMS) students completed a survey about career interests, and factors important in career choice. Empathy was evaluated using the Jefferson Empathy Scale, a validated 20-question survey. A single empathy composite score was then generated for each respondent. The relationship between career interest and career choice factors, and empathy, was assessed using generalized linear models, controlling for year in medical school, gender, race, marital status, undergraduate major (science vs. non-science vs. both), and hometown type (rural vs. urban vs. suburban).

RESULTS: The study population included 166 students who completed both an initial survey and the Jefferson Empathy Scale. The average score was 118.8 (range 89–140). Out of 166 students, 75% were first year medical students, 59% were female, 60% were white, and the average age was 25. Students who were interested in general medicine had a higher empathy score than those who were not (120.3 vs. 114.8; p = 0.04). Students who prioritized working with the under-served had higher empathy scores than those who did not (120.5 vs. 114; p = 0.0003). Students who prioritized future income had a lower empathy score than those who did not (117.4 vs. 120.2; p = 0.004). Students who prioritized the ability to work part-time had higher empathy scores than those who did not (122 vs. 117.3; p = 0.02). Students who wanted to teach students and residents had higher empathy scores than those who did not (120 vs. 115.8; p = 0.0003).

CONCLUSIONS: Among medical students participating in a student-faculty collaborative practice, higher empathy scores were correlated with interest in general medicine, working with the under-served, the ability to work part-time, and teaching. The importance of income was negatively correlated with empathy score. In the future, we hope to examine empathy scores of students before and after participation in the Crimson Care Collaborative. Further research is needed on the extent to which empathy levels can be changed in medical students and what experiences might influence empathy levels. Findings around empathy and career choice may have implications for medical school admissions and primary care recruitment.

EMR-BASED PHYSICAL ACTIVITY COUNSELING AND REFERRAL IN PRIMARY CARE: AN EVALUATION STUDYJulien J. Dedier¹; Julie A. Wright²; Timothy Heeren³; Robert H. Friedman¹. ¹Boston University, Boston, MA; ²University of Massachusetts, Boston, MA; ³Boston University, Boston, MA. (Tracking ID #1939345)

BACKGROUND: Physical activity (PA) is beneficial in the prevention and management of several diseases, and linking primary care patients to effective PA programs may improve health outcomes for exercise-responsive conditions such as diabetes, hypertension and obesity. The purpose of this study was to evaluate the implementation of an electronic medical record (EMR) tool designed to help primary care providers conduct PA counseling and refer patients to an automated telephone PA coach.

METHODS: Content of the EMR PA tool was based on the 5A’s approach to PA counseling and elements of motivational interviewing. Providers reviewed a prototype of the tool, and their feedback was used to design the final version. The tool was publicized at business meetings and by e-mail announcements. Providers could use the tool to do 1 or more of the following: (1) determine the patient’s current level of PA and motivation to increase PA, and counsel accordingly; (2) create an exercise prescription; and (3) refer patients to use an automated telephone PA coach (TLC-PA). TLC-PA was developed by our group, and was shown to significantly increase self-reported PA of moderate-or-greater intensity in an RCT of hypertensive primary care patients. Patients referred to TLC-PA were scheduled to receive 12 weekly calls of roughly 10min each. We collected implementation information from providers and patients. We captured the electronic footprint providers created whenever they used the tool, and surveyed users’ opinions of the tool each time they used a new feature. We surveyed a random subset of non-users to determine their reasons for non-use. We also surveyed patients who were referred to TLC-PA on their impressions of the referral encounter.

RESULTS: 50 unique providers accessed the EMR tool at least once and were sent a survey about its functions. Of respondents (n = 13), most agreed or strongly agreed the 3 EMR tool functions were easy to use (66%–88%), increased the quality of PA counseling they provided (72%–88%), were worth the time spent to use them (57%–75%), and should be made a permanent part of the EMR (82%–100%). Providers accessed the exercise prescription function most often (104 instances), then the stage-based counseling function (62 instances), followed by the TLC-PA referral function (36 instances). Providers’ greatest reservation about PA referral was not knowing enough about the Coach to confidently refer their patients. Among non-user survey respondents (n = 24), only 37.5% had heard of the EMR PA tool. Ninety-two percent were not sure how to access it. Seventy-eight percent of patients selected for exercise counseling or referral had an exercise-responsive diagnosis of obesity, DM, PVD, or CAD. Fourteen of the 36 patients referred to use TLC-PA completed a survey about the office referral encounter. All recalled discussing PA and TLC-PA, and 86% recalled receiving printed information about TLC-PA. All referred patients went on to use TLC-PA.

CONCLUSIONS: PA interventions such as our automated PA coach have been shown to increase PA. We successfully linked a PA counseling and referral intervention to the EMR. Primary care providers and their patients considered it helpful, but low-intensity advertising of the tool resulted in only modest use. Comprehensive implementation efforts are required to fully assess the feasibility of linking PA interventions to the EMR in primary care settings.

ENGAGING VETERANS IN CONVERSATIONS ABOUT HEALTH: A COMMUNITY PARTNERSHIP AND ASSESSMENT TO PROMOTE VETERAN HEALTHAnita D. Misra-Hebert^1,2; Laura Santurri³; Richard DeChant⁴; Brook Watts⁵; David Aron⁵. ¹Cleveland Clinic, Cleveland, OH; ²Louis Stokes Cleveland Veterans Affairs Medical Center, Cleveland, OH; ³Weber State University, Ogden, UT; ⁴Cuyahoga Community College, Cleveland, OH; ⁵Louis Stokes Cleveland Veterans Affairs Medical Center, Cleveland, OH. (Tracking ID #1923869)

BACKGROUND: As veterans of recent conflicts in Iraq and Afghanistan return to civilian life, and as veterans of previous conflicts continue to age, the importance of health promotion as related to chronic disease prevention and management in the veteran population will be a required area of focus for clinical providers. The high prevalence of mental health disease in veterans of recent military conflicts may affect risk for chronic disease development and the self-efficacy/coping mechanisms required for optimal chronic disease management. Veteran students pursuing higher education may be particularly motivated to engage in health promotion. The study objective is to explore motivations to improve health and to assess health among student veterans, using a shared forum, through partnership between a Veterans Affairs (VA) medical center and a community college.

METHODS: This study utilized a sequential mixed methodological design with an equally weighted approach for the qualitative and quantitative portions. The study participants were student veterans enrolled at a community college in October 2012–April 2013. Veteran students were recruited to participate in focus groups. Participants discussed views of health, motivations to improve health, and perceptions of medical care. Discussions were audio-recorded and transcribed. Other students then participated in a cross-sectional health assessment survey. Qualitative data analysis utilized a conventional content analysis approach to coding and thematic development. Survey data analysis included descriptive and logistic regression analyses.

RESULTS: A total of 17 veteran students participated in 6 focus groups. Emergent themes included changing views of health after transition from active duty, generational differences among veterans, barriers to accessing health services, and health care costs. Concerns about basic needs such as job/ housing and providing/caring for family, stigma and privacy, and preferences related to health communication were also important themes. The motivation to improve health for many of our study participants was perceived as secondary to the need to obtain stable housing or work, and concerns about privacy and stigma were identified as possible barriers to seeking healthcare. Two hundred four students (21% response rate) participated in the health assessment survey. Physical and emotional limitations (45% and 35%, respectively), and pain interfering with work (42%) were reported. Self-reported depression and unhealthy behaviors were high.

CONCLUSIONS: Engaging veteran students in conversations about health and conducting a health assessment survey identified factors influencing motivation to improve health and priority areas to inform interventions for chronic disease prevention. The impact of social determinants on health is substantial in this group. A significant proportion of student veterans are limited in daily activities as a result of physical as well as emotional issues and specifically from pain. A partnership between a VA medical center and a community college integrates primary care with public health as a step to improving the health of veteran students in the community.

ENGAGING YOUTH THROUGH A SPOKEN WORD TYPE 2 DIABETES PREVENTION SOCIAL MARKETING CAMPAIGNElizabeth A. Rogers^1,2; Sarah Fine⁴; Margaret A. Handley^3,4; Purba Chatterjee⁴; James Kass⁵; Dean Schillinger^3,4. ¹University of Minnesota, Minneapolis, MN; ²University of Minnesota, Minneapolis, MN; ³University of California San Francisco, San Francisco, CA; ⁴University of California San Francisco, San Francisco, CA; ⁵Youth Speaks, San Francisco, CA. (Tracking ID #1939606)

BACKGROUND: Type 2 diabetes (DM2) prevalence is rising among minority young adults. There is an unmet need to engage youth in identifying solutions to decrease this risk. Some media campaigns such as the successful truth® antismoking campaign used a socio-environmental perspective to shift social norms, and this approach may be one avenue toward engagement around DM2. Objective: To evaluate the effectiveness and reach of The Bigger Picture, a youth-generated, youth-targeted DM2 prevention social marketing campaign.

METHODS: Campaign Development: Through a partnership between UCSF and Youth Speaks, a youth spoken-word nonprofit, we used a participatory approach engaging youth poets to create 11 video public service announcements (PSAs) addressing the socio-environmental conditions that increase DM2 risk. Our dissemination focus was regional low-income and minority-serving high schools identified through collaboration with school districts, and we developed a scripted assembly and a writing workshop. We created a campaign website (www.thebiggerpicture.org) with information, PSA videos, and informational toolkit, and used Facebook and Twitter social media. Evaluation: Using the RE-AIM framework, we evaluated the campaign through efficacy measures capturing changes in knowledge and attitude, and those that characterized the reach of the campaign. We gathered data using audience response system handheld devises distributed to students (up to 80 per assembly) to answer 3 questions gauging knowledge and attitude embedded at the beginning and end of the assembly, and through paper pre/post surveys (closed- and open-ended questions) of all students in the workshops. To assess change, we used McNemar’s chi-square test and a mixed model (adjusted for demographics) accounting for clustering at the school level. We employed counts, means, and proportions of state-reported school data and web analytics to assess reach of the school-based dissemination and web-based media.

RESULTS: PSA themes included the association of DM2 with the built environment, industry advertising, and racial and socioeconomic disparities. In 13 high schools over 6months, 1900 students participated in the poet-led assemblies and 263 in the workshops. The participating schools enrolled on average 982 students, 66% of whom qualified for free or reduced lunches. Seven hundred eighty-one students responded to assembly surveys. Fifty-six percent were female with an average age of 15.7years, and 28% were Latino, 27% Asian, 18% African American, and 14% multi-ethnic. Ninety-two percent agreed that DM2 is preventable after the assembly compared to 70% before (McNemar’s p < 0.001, adjusted mixed model OR 5.3), and 83% after identified environmental factors in addition to individual behaviors as influencing DM2 risk compared to 34% before (p < 0.001, OR 13.1). Using a Likert scale, 59% after responded “I care a lot” about preventing DM2 compared to 29% before (p < 0.001, OR 9.1). In the workshops, students identified the DM2 epidemic as a social justice issue. To date, the 11 web-based video PSAs combined had 72,232 views. The campaign’s Facebook page has 383 “likes” and the Twitter account has 342 followers.

CONCLUSIONS: A youth-targeted, youth-generated social marketing campaign improves short-term knowledge and attitudes about DM2 risk. Through high schools, we reached our target population, and web-based media has potential for extending this reach. This type of campaign may be one avenue toward engaging diverse, high-risk youth to change the conditions that increase DM2 risk.

EREFERRALS AND ECONSULTS: DOWNSTREAM IMPACT ON ACCESS, UTILIZATION, AND COST IN A FEE-FOR-SERVICE SETTINGNathaniel Gleason¹; Jennifer J. Monacelli⁶; Chanda Ho²; Sara Ackerman³; Priya A. Prasad⁴; Michael Wang⁵; Don Collado¹; Delthia McKinney⁵; Ralph Gonzales¹. ¹UC San Francisco, San Francisco, CA; ²UC San Francisco, San Francisco, CA; ³UC San Francisco, San Francisco, CA; ⁴UC San Francisco, San Francisco, CA; ⁵UCSF Medical Center, San Francisco, CA; ⁶Michigan State University, Grand Rapids, MI. (Tracking ID #1939845)

BACKGROUND: Demand for specialty care services exceeds supply at many academic referral centers. Delays in access to specialty care are common, and are anathema to patient-centered care. Electronic referrals (eReferrals) and non-face-to-face consultations (eConsults) are mechanisms associated with significant improvements in the specialty care referral process in integrated delivery systems with aligned incentives. Translating these innovations to academic medical centers (AMCs) has been challenging; reimbursement remains largely fee-for-service, and Medicare and most commercial payers do not pay for eConsults. With support from the Medicaid waiver program (DSRIP), UCSF Medical Center implemented a new referral platform. The system is designed to improve information exchange at the point-of-referral and to provide an eConsult option, allowing PCPs to request timely input from specialists for data-oriented referral problems that the PCP does not perceive to require in-person evaluation of the patient by the specialist. In this study, we analyzed the downstream impact of this program on PCP referral rate, specialty care access, and costs.

METHODS: eReferral/eConsult Platform: In May 2012, UCSF introduced a structured referral platform within the electronic health record (EHR) (Epic Systems) with over 90 problem-specific, structured templates providing clinical decision support at the point-of-referral. An eConsult option was added in September 2012. The expected eConsult response time is 3 business days. Upon review, if the specialist determines the case is not appropriate for eConsult, the specialist converts the eConsult to a standard new-patient visit. Specialists receive a payment corresponding to 0.5 wRVU per completed eConsult. PCPs receive 0.5 wRVU credit toward productivity targets. in recognition of the fact that the PCP maintains management responsibility for the referral problem. Study populations: The program included 8 primary care sites and 12 medicine subspecialty practices. The study population includes all eligible PCPs (n = 178) and all referrals to participating practices for patients assigned to an eligible PCP during baseline (9/1/11–8/31/12) and study periods (9/1/12–4/30/13) (n = 13,738 referrals representing 11,597 unique patients). (To account for seasonal effects on ED visits and hospitalizations, we used baseline and intervention periods of Oct–April 2012 vs Oct–April 2013 for the utilization and cost analysis). We obtained data on referral rates from the EHR scheduling database. Data on utilization and costs were obtained from the University Health System Consortium. Analysis: We calculated the total referral rate per 100 primary care visits per month, and the proportion of these sent as eConsults. To compare the trends in referral rate, we modeled the effect of time using a linear spline analysis with two knots to determine whether there were differences in the trajectory following the introduction of the structured referral system and eConsults. To measure impact on access to specialty care, the proportion of UCSF primary care patients who received specialty care input (office visit or eConsult) within 14 business days in the baseline and eConsult periods were compared using chi-square tests. Mean monthly ambulatory, emergency department, hospital utilization and professional fee costs occurring within 120days following each referral or eConsult were log transformed to minimize effects of outlier data. Average costs were compared between baseline (n = 6months) and eConsult (n = 6months) periods with unpaired t-tests.

RESULTS: Of eligible PCPs, 69% placed at least one eConsult. Based on the spline analysis, the referral rate declined from 12.42 per 100 primary care visits during the baseline period to 10.66 per 100 primary care visits in August 2012 (11% decrease; p = 0.011). eConsults were added as an option in September 2012, and the referral rate remained stable though the remainder of the study. eConsults were used for 8.2% of total referrals. The proportion of patients who received specialty care input (office visit or eConsult) within 14 business days improved from 39 to 49% (p = 0.001) across all participating specialties. Specialties with greatest improvements were pulmonary (from 20 to 50% (p = 0.004)), GI (from 19 to 40% (p = 001)) and nephrology (from 50 to 74% (p = −.05)). The proportion of referrals with an ED visit during the 120-day period following referral or eConsult decreased from 9.8 to 8.6%, and there was a significant reduction in total ED pro-fee costs (p ss0.016). Mean ambulatory and hospital pro-fees did not change between periods.

CONCLUSIONS: This combined eReferral (enhanced referral platform) and eConsult program showed robust adoption and had significant impact on referral rate and specialty care access time. The significant impact on ED costs may represent a downstream benefit of improved access to care.

ESTABLISHING GUIDELINES FOR REFERRAL TO ORTHOPAEDIC SPECIALTY CARE: CONSENSUS USING THE MODIFIED DELPHI METHODMaria E. Otto; Carlin Senter; Christy K. Boscardin; Ralph Gonzales; Nathaniel Gleason. University of California San Francisco, San Francisco, CA. (Tracking ID #1939439)

BACKGROUND: Standardized electronic referrals can improve physician communication, provider experience, and wait times. The Council for Subspecialty Societies, sponsored by the American College of Physicians, defines the features of high-quality care coordination between primary care and specialty care in a Patient Centered Medical Home-Neighborhood (PCMH-N). Specifically, high-quality referrals in the PCMH-N are defined as referrals that facilitate effective transfer of clinical information, elicit a clinical question, convey expectations about management roles, and ensure appropriate and timely consultation. The University of California San Francisco (UCSF) implemented an eReferral system, integrated into the Electronic Health Record, in which structured templates facilitate referrals to specialty care that are in keeping with PCMH-N recommendations and provide decision support at the point-of-care. Expanding this program to orthopaedics, we found limited available guidelines for orthopaedic problems regarding referral timing, management steps to be taken prior to referral, imaging in advance of referral, and clinical questions for which a non-face-to-face visit (eConsult) with an orthopaedist might be appropriate. This study describes our experience using the modified Delphi method to inform development of standardized orthopaedic referral guidelines by identifying consensus between referring PCPs and orthopaedic clinicians on specific recommendations.

METHODS: To represent a broad range of expertise, all UCSF Physicians, Physician Assistants, and Nurse Practitioners who provide care to adult patients in the UCSF orthopaedic clinic were asked to participate (Foot & Ankle n = 2, Hand = 3, Sports = 5, Arthoplasty = 6, Spine = 8). Primary care providers (PCPs) volunteered in response to a recruitment email sent to all 8 UCSF adult primary care practices (n = 22 volunteers). The expert panel completed two rounds of anonymous electronic surveys over a period of 4weeks. PCPs answered all questions, while clinicians in the orthopaedic practice answered only questions related to the joints each treated in practice. Surveys were composed of clinical scenarios followed by questions about appropriate pre-referral evaluation, conservative management, and the utility of a non-face-to-face consultation (eConsult) with an orthopaedist. Scenarios were written for the most commonly referred orthopaedic diagnoses and constructed to differ by one variable at a time (e.g. chronic vs. acute, a sedentary vs. athletic patient). The degree of importance for each possible intervention was ranked on a 5-point Likert scale (strongly disagree to strongly agree). The second round survey included the median response for each group (PCPs and orthopaedic clinicians) from the first round for each question, allowing panelists to factor in these initial responses when re-visiting the scenarios in round two. Consensus was defined as ≥70% of PCPs and orthopaedic clinicians giving a score of 1–2 or 4–5.

RESULTS: All 22 PCPs completed the Round 1 & 2 surveys. Nineteen orthopaedic clinicians completed Round 1 and 17 completed Round 2 (Response Rate: 79% & 70% respectively). Consensus was reached in 142 of 214 items. Most notably, PCPs and orthopaedists agreed that 1) X-ray should be performed in advance of referral for most clinical scenarios while pre-referral MRI is recommended in select scenarios: acute knee pain concerning for ligamentous injury and/or meniscal tear, spinal stenosis, chronic neck pain with arm pain in a dermatomal distribution, and intractable neck pain. 2) Pre-referral physical therapy is recommended for the following conditions: Achilles tendinitis, groin pain concerning for labral tear or impingement in a patient ≥65years old, trochanteric bursitis, osteoarthritis of the hand, shoulder, and knee, meniscal tear in a patient ≥65years old patellofemoral knee pain, shoulder impingement syndrome, frozen shoulder, medial and lateral epicondylitis, and spinal stenosis. 3) PCP management guided by electronic consultation by an orthopaedist, in place of a patient visit, could be appropriate for the following conditions: chronic pain after ankle sprain, plantar fasciitis, stress fracture of the foot, osteoarthritis of the hand, shoulder, knee, and hip, trochanteric bursitis, degenerative meniscal tear, patellofemoral pain, shoulder impingement syndrome, medial epicondylitis, intractable low back and neck pain.

CONCLUSIONS: This consensus is the first step in improving the quality of orthopaedic referrals. Through this effort, we hope to have established meaningful referral guidelines that represent the collective opinions of experts with the premise that ‘pooled intelligence’ enhances individual judgment. Our future efforts will aim to measure whether or not standardized referral templates developed using the Delphi method in consultation with a panel of experts and following the current PCMH-N guidelines improve referral efficiency and quality.

ESTIMATING TOBACCO-, ALCOHOL-, AND DRUG-ATTRIBUTABLE DEATHS AND THEIR CONTRIBUTION TO MORTALITY DISPARITIES AMONG HOMELESS ADULTSTravis P. Baggett^1,2; Yuchiao Chang¹; Bianca Porneala¹; Jessie M. Gaeta^2,3; James J. O’Connell^2,1; Nancy A. Rigotti¹. ¹Massachusetts General Hospital, Harvard Medical School, Boston, MA; ²Boston Health Care for the Homeless Program, Boston, MA; ³Boston University School of Medicine, Boston, MA. (Tracking ID #1939356)

BACKGROUND: Homeless individuals have a high prevalence of substance use disorders and experience excess mortality rates in comparison to non-homeless people. Our objective was to estimate the burden of deaths attributable to tobacco, alcohol, and drug use in a large cohort of homeless adults, and to assess the contribution of substance-attributable mortality to the disparity in all-cause mortality with the general population.

METHODS: We assembled a cohort of 28,033 adults aged ≥18years who were seen at Boston Health Care for the Homeless Program in 2003–08. We used probabilistic methods to cross-link this cohort with Massachusetts mortality files spanning the same years. For decedents, we based causes of death on ICD-10 underlying cause codes in the death occurrence files. We used accepted epidemiologic methods to estimate population attributable fractions (PAFs) for causes of death that high-quality studies have demonstrated to be causally related to the use of tobacco, alcohol, or drugs. For each cause of death, the PAF represents the proportion of these deaths that would not have occurred in the absence of tobacco, alcohol, or drug use. We used Monte Carlo simulation methods to generate 95% confidence intervals for our PAF estimates. For each condition, we multiplied the PAF by the count of deaths to estimate the number of deaths attributable to tobacco, alcohol, or drug use. For conditions related to more than one substance, we used standard methods to account for overlapping attribution. We divided the number of attributable deaths by the person-time at risk to generate attributable rates. As a basis for comparison, we used the same methods to estimate substance-attributable mortality rates for the 2003–08 Massachusetts general population.

RESULTS: Of 1302 deaths, 236 (95% CI 215–254) were tobacco-attributable, 215 (95% CI 199–232) were alcohol-attributable, and 286 (95% CI 285–287) were drug-attributable. After accounting for overlap, about 52% (N = 676, [95% CI 655–698]) of all deaths were attributable to any of these substances. Drug use was the predominant contributor to substance-attributable deaths under the age of 50years (63%), while tobacco use was the major contributor to substance-attributable deaths over the age of 50years (55%). In comparison to the Massachusetts general population, age- and sex-stratified mortality rates in the homeless cohort were 3–4 times higher for tobacco-attributable causes, 6–12 times higher for alcohol-attributable causes, 8–17 times higher for drug-attributable causes, and 5–11 times higher for all substance-attributable causes combined. Disparities in substance-attributable deaths accounted for nearly 60% of the age- and sex-standardized mortality rate difference between the homeless cohort and the Massachusetts general population.

CONCLUSIONS: Over half of all deaths among homeless people are attributable to tobacco, alcohol, or drug use. Substance-attributable deaths are a major contributor to mortality disparities between homeless adults and the general population, but they do not fully explain the mortality gap between these groups of people. Efforts to reduce mortality among homeless individuals should focus on the expansion of substance abuse treatment services in conjunction with comprehensive health care and social policy interventions to address non-addiction sources of excess mortality in this vulnerable population.

EVALUATING INPATIENT HOSPITAL CHARGE VARIABILITY USING LOCAL HEALTH AND MARKET FACTORSJames D. Park¹; Edward Kim²; Rachel M. Werner³. ¹Rutgers, Robert Wood Johnson Medical School, New Brunswick, NJ; ²The College of New Jersey, Ewing, NJ; ³Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA. (Tracking ID #1937089)

BACKGROUND: In May 2013, the Center for Medicare and Medicaid Services released previously undisclosed information about the price of hospital care. Within a diagnosis related group, inpatient hospital charges to Medicare demonstrated a wide range of charges across US hospitals. This variation in hospital charges remains unexplained, and this study examines its relationship to the local market environment in which hospitals operate.

METHODS: This descriptive study evaluated the CMS Inpatient Provider Charge data of over 3000 US hospitals using clustered, multivariate linear regression. This analytic approach examined how county health, socioeconomic, and market factors relate to hospital charges. Six common inpatient conditions—cellulitis, chronic obstructive pulmonary disease, congestive heart failure, myocardial infarction, orthopedic surgeries, and pneumonia—were chosen for the analysis. Related DRGs for each condition were grouped together, and a weighted average charge for each hospital was calculated based on the proportion of discharges per DRG. Based on a conceptual model, a common set of covariates was used to create a single regression model for each condition. The analysis used various county-level measures of health status (years of potential life lost per 100,000 population, percent of adults who report fair or poor health, percent days reported as mentally unhealthy per month, diabetes prevalence, obesity prevalence); health behavior (prevalence of smoking, no leisure time activity, or heavy alcohol drinking); clinical access and quality (prevalence of uninsured status, preventable hospital stays rate, percent of diabetics who received a hemoglobin A1c test, number of primary care physician per 100,000 population); socioeconomics status (percent of a 9th grade cohort that graduates in 4years, prevalence of unemployment, of children living in poverty, and of children living in single-parent homes, median household income, violent crime rate per 100,000 population); the built environment (number of accessible recreational facilities per 100,000 population); demographics (percent of the population that is African American, Asian, Latino, or other race, and percent of the population over the age of 65); and the hospital market (for-profit status, proportion of hospital discharges within the county) from every US county and hospital. All analyses controlled for other market factors such as the hospital wage index, disproportionate share hospital index, hospital cost-to-charge ratio, and the county-level Herfindahl-Hirschman Index. Covariates were extracted from multiple datasets including the US Census Bureau, Behavioral Risk Factor Surveillance System, National Center for Chronic Disease Prevention and Health Promotion, CMS Impact, CMS Provider of Service, County Business Patterns, American Community Survey, Dartmouth Atlas of Health Care, Bureau of Labor Statistics, Small Area Income and Poverty Estimates, Federal Bureau of Investigation Uniform Crime Reporting, National Center for Health Statistics, and the National Center for Educational Statistics. All regression analyses were clustered by county, and statistical testing was deemed significant with p-values <0.05.

RESULTS: The mean number of hospitals included across six regressions was 2603. Across all six conditions, hospital charges were associated with uninsured status and resulted in $210.28–$623.26 higher charges for every 1% increase in the prevalence of uninsured status (p < 0.01). In addition, hospital charges were associated with for-profit hospital status across all six conditions and resulted in $1230.47–$7772.10 higher charges than not-for-profit status (p < 0.05). In all conditions except myocardial infarction, hospital charges were associated with unemployment status and resulted in $295.83–$752.53 higher charges for every 1% increase in unemployment prevalence (p < 0.05). Health behavioral factors such as smoking, physical inactivity, and excessive alcohol consumption, or health status factors such as premature death, fair or poor health status, poor mental health days, or diabetes were largely unassociated with hospital charges. Of all health factors utilized in this study, only obesity was repeatedly associated with hospital charges in four of the six conditions, resulting in $232.57–$513.92 higher charges for every 1% increase in obesity prevalence (p < 0.05).

CONCLUSIONS: Non-health related factors, such as uninsured, unemployment, and for-profit hospital status, were consistently associated with higher hospital charges. In contrast, nearly all health factors had no association to inpatient hospital charges. This research raises concerns about a regressive pricing strategy that is indifferent to health status but charges vulnerable and uninsured populations higher prices.

EVALUATING THE UTILITY OF OSTEOPOROSIS SCREENING TOOLS (OST, FRAX) AND VIDEO EDUCATION IN HOSPITALIZED PATIENTSEve Thau; Eileen Hennrikus. Penn State Milton Hershey Medical Center, Hershey, PA. (Tracking ID #1938794)

BACKGROUND: Approximately eight million women and two million men in US suffer from osteoporosis, with the cost of fragility fractures an estimated $10–15 billion annually. To calculate osteoporotic fracture risk, the FRAX (Fracture Risk Assessment Tool), takes into account age, weight, height, gender, parental fractures, tobacco and alcohol use and secondary causes for osteoporosis. A simpler risk assessment measurement is the OST (Osteoporosis Self-Assessment Tool), a calculation using only age and weight. We proposed that osteoporosis risk assessment and education can be accomplished in patients hospitalized on the inpatient internal medicine service. Our objectives include (1) assess if OST compared to FRAX is a useful tool to screen inpatients for osteoporosis, (2) determine if osteoporosis video education is well received and (3) determine if patients hospitalized for other illnesses will act upon knowledge and education for a preventive health measure.

METHODS: Three days per week for 2months, an OST was calculated on all internal medicine inpatients >60years old. One hundred twelve patients with an OST > 2, deemed not at risk for osteoporotic fracture then had a FRAX calculated by chart review. The 72 patients with an OST < 2, deemed at risk for osteoporotic fracture, were consented for further study. Patients were excluded if they were not interested in participating, had altered mental status or were too sick to participate. Fifteen eligible patients were excluded. Patients were given the FRAX questionnaire, a 7-question osteoporosis pre-test, followed by an NIH-sponsored educational osteoporosis video, and the same test post-video. Afterwards, patients were asked their opinion of the video education, and if they would take any new measures following hospital discharge to help improve bone strength and prevent osteoporosis.

RESULTS: 184 patients were evaluated, mean age 73years old, and mean weight 91kg. Causes for admission were primarily cardiovascular, pulmonary, and renal disease with only two patients admitted with bone fractures. Forty percent of internal medicine inpatients were found to be at osteoporotic fracture risk by the OST calculation. Of those 72 patients, 68 (94% p = 0.001) were also found to be at risk for both major fracture and hip fracture by FRAX calculation. Of the 112 patients not found to be at risk by OST, 92 (84%, p = 0.01) correlated with FRAX risk. Of the 20 patients where the FRAX found risk and the OST did not, 10 (50%) had an obvious secondary cause for osteoporosis such as a history of chronic renal failure or prednisone usage. Of the other 10 patients who did not have a secondary cause for osteoporosis, eight of them were women over the age of 68, median age 74years old. 30% (22) of the at risk patients had been previously counseled by their physician for osteoporosis, and 40% (29) of these patients were already taking calcium supplements, vitamin D and 12% (9) were taking bisphosphonates before hospital admission. 36% (26) patients previously had a DEXA screen before participation in the study. With regards to patient education, we found an improvement of 20.6% when comparing quiz scores conducted before and after exposure to the NIH-sponsored educational video p = 0.001. Patients found this video very helpful in learning more about osteoporosis and preferred it to written material. After being informed of their risk, and after watching the video, 28% (20) patients said that they would address osteoporosis, 16 would make mild lifestyle changes and only four planned to speak with their doctor and pursue treatment. Patients who opted to not alter lifestyle or seek further medical advice regarding osteoporosis noted that they were too sick or too old.

CONCLUSIONS: The OST is an effective risk assessment tool for osteoporosis. OST does not capture those patients with obvious secondary causes for osteoporosis. Of those patients at risk, 60% had not had osteoporosis addressed by their primary care physician. Video education was well received in the hospital and revealed a significant improvement in knowledge about osteoporosis. However, after being informed of their fracture risk and receiving education only 28% of patients would pursue treatment. A significant number of patients were found to be at risk and untreated for osteoporosis. Although the education was well received, preventive measures for osteoporosis were not a priority compared to the acute issues for which patients are hospitalized. Optimally, screening should be done in the outpatient setting, but perhaps inpatient screening and education can be useful if translated to the primary care physician to address when acute issues recede.

EVALUATION OF AN ELECTRONIC HEALTH RECORD PROMPT FOR HEPATITIS C ANTIBODY SCREENING OF BABY BOOMERS IN PRIMARY CARE - A CLUSTER RANDOMIZED CONTROL TRIALKatherine Krauskopf¹; Natalie Kil¹; Anastasia Sofianou²; Wilma Toribio¹; Joanne Lyons³; Mark Singer³; Joseph Kannry¹; Bryce Smiith⁵; David B. Rein⁴; Alex Federman¹. ¹Icahn School of Medicine at Mount Sinai, New York, NY; ²Icahn School of Medicine at Mount Sinai, New York, NY; ³Mount Sinai Health System, Huntington, NY; ⁴NORC at University of Chicago, Atlanta, GA; ⁵Centers for Disease Control and Prevention, Atlanta, GA. (Tracking ID #1938222)

BACKGROUND: The CDC recommends one-time hepatitis C virus (HCV) testing for all persons born between 1945 and 1965 (birth cohort testing). Electronic Heath Records (EHR) may represent an effective tool to increase testing uptake. We sought to assess use and acceptance of an EHR alert to promote birth cohort testing in primary care, and determine subsequent prevalence of positive HCV tests among previously undiagnosed individuals.

METHODS: We conducted a cluster randomized controlled trial of three adult primary care practices (teaching and faculty) within the Mount Sinai Medical System, NY, beginning 4/2013 and ongoing to 3/2014. Ten clusters were identified based on provider type, geography, and patient population, and randomized to intervention (EHR screening prompt) or control (usual care). A member of the research team gave a brief intro to all physicians (MD), nurse practitioners (NP), medical assistants (MA), and registered nurses (RN), regardless of study arm assignment, about HCV, birth cohort testing recommendations, and the study. In intervention clusters, the EHR (Epic Systems, Madison, WI) triggered a screening prompt when eligible patients (those without prior HCV antibody [Ab] testing or HCV diagnosis born during 1945–1965) registered for a visit. When the MA/RN opened the electronic chart before the MD/NP, the EHR generated a prompt to pend an HCV Ab order for the MD/NP to discuss with the patient, then sign or delete. When the MD/NP opened the chart first, the EHR generated an alert to suggest HCV testing, and provided an HCV Ab order for acceptance or deletion. At study completion, differences in characteristics of intervention and control providers, and among the patients seen in intervention vs. control clusters, will be measured using t test and X2 test. We will compare testing and HCV positivity prevalence between intervention and control arms, as well as process outcomes, using generalized estimating equations.

RESULTS: After 6months of data collection, 146 of 154 MD/NP (95%) approached have consented to participate, as have 73 of 78 (93%) RN/MA. There have been 10,857 eligible intervention-arm visits and 7,372 control-arm visits. Of these, 2,176 HCV Ab tests were ordered in the intervention arm (20%) versus 225 ordered in the control arm (3%). There were 20 positive tests in the intervention arm (0.9% of those tested) versus five in the control arm (2.2% of those tested).

CONCLUSIONS: Preliminary data show that the EHR intervention results in a higher rate of testing and consequently finds more cases than usual care. While targeted methods of usual care detect a higher prevalence, testing occurs too infrequently to match the aggregate case detection of the EHR prompt. Although successful in detecting more cases, only 20% of eligible patients were tested via the intervention, suggesting that modification of the intervention may be needed to achieve HCV testing among all birth cohort adults.

EVALUATION OF STEPPED CARE FOR CHRONIC PAIN (ESCAPE) IN VETERANS OF IRAQ AND AFGHANISTAN: A RANDOMIZED TRIALMatthew J. Bair^1,2; Dennis Ang⁴; Jingwei Wu⁵; Samantha D. Outcalt¹; Christy Sargent¹; Amanda Gerwig¹; Arlene Schmid⁶; Teresa Damush^1,3; Zhangsheng Yu⁵; Louanne W. Davis⁷; Kurt Kroenke^1,3. ¹VA HSR&D Center for Health Information and Communication, Indianapolis, IN; ²Indiana University School of Medicine, Indianapolis, IN; ³Regenstrief Institute, Inc, Indianapolis, IN; ⁴Wake Forest School of Medicine, Winston-Salem, NC; ⁵Indiana University School of Medicine, Indianapolis, IN; ⁶Colorado State University, Fort Collins, CO; ⁷Roudebush VA Medical Center, Indianaplis, IN. (Tracking ID #1937595)

BACKGROUND: Despite the prevalence and substantial functional, psychological, and economic impact of chronic musculoskeletal pain on military personnel and veterans, there have been few intervention studies to treat chronic pain in these populations. We conducted a randomized controlled trial to determine if a stepped-care intervention is more effective than usual care in reducing pain-related disability, pain interference, and pain severity.

METHODS: We recruited patients from post-deployment and five general medicine clinics at a Veteran Affairs Medical Center. Participants included 242 Operation Enduring Freedom and Operation Iraqi Freedom Veterans with chronic, disabling musculoskeletal pain of the cervical or lumbar spine or extremities (shoulders, knees, and hips). The intervention involved 12weeks of analgesic treatment and optimization according to an algorithm coupled with pain self-management strategies (Step 1) followed by 12weeks of brief cognitive behavioral therapy (Step 2). All aspects of the intervention were delivered by nurse care managers. The primary outcome was pain-related disability (Roland Morris Disability Scale) at 9months. We also assessed pain interference (Brief Pain Inventory) and pain severity (Graded Chronic Pain Scale). The primary analysis compared the change from baseline to 9months between study groups. To estimate treatment effects, we used random-effect linear regression adjusted for age, sex, severity of pain, and baseline value of outcome measure. Analyses were conducted with SAS version 9.13.

RESULTS: There were no significant baseline differences between stepped-care intervention and usual care arm. The mean age of ESCAPE participants was 36.7years (range = 21 to 73years) and 88.4% were men. In terms of racial mix, 77.7% were white, 13.0% were black, and 9.2% classified as other. The primary site of pain was the low back (57%) followed by knee pain (21.6%), neck pain (7.5%), shoulder pain (7.1%), and hip pain (6.6%). Two-thirds (66.4%) had been in the Army, 74.9% had been deployed to Iraq, 8.8% deployed to Afghanistan, and 16.3% deployed to both conflicts. One hundred twenty-one intervention participants and 120 usual care participants were included in the primary analysis. At 9months, the mean decrease from baseline in the Roland Morris Disability Scale score was 1.6 points (95% confidence interval [CI] 0.72–2.6) in the usual care group and 3.4 points (95% CI 2.5–4.3) in the stepped-care intervention group (between group difference of −1.8 points, −3.1 to −0.51; p = 0.0063). The mean decrease from baseline in the Brief Pain Inventory interference score was 0.86 points in the usual care group and 1.6 points in the intervention group (between group difference of −0.69 points, −1.3 to −0.12; p = 0.0170). The Graded Chronic Pain Scale severity score was reduced by 7.1% (95 CI’s 2.7% to 11.5%) and 20.4% (95% CIs 14.3% to 26.6%), respectively (between group difference 13.3%, 11.6% to 15.1%; p = 0.0013).

CONCLUSIONS: A stepped-care intervention that combined analgesics, self-management strategies, and brief cognitive behavioral therapy resulted in a significant reduction in pain-related disability, pain interference, and pain severity in Veterans of Iraq and Afghanistan with chronic, disabling musculoskeletal pain of the spine and extremities.

EVALUATION OF THE MODIFIED HEART SCORE IN PREDICTING OUTCOMES IN PATIENTS WITH LOW-RISK CHEST PAINCarlos Calle-Muller; David Paje; Matthew Cerasale; Iani Patsias. Henry Ford Hospital, Detroit, MI. (Tracking ID #1940326)

BACKGROUND: The HEART (History, ECG, Age, Risk factors and Troponin) score was recently shown to be potentially useful in identifying patients presenting with chest pain with a very low pretest probability of acute coronary syndrome (ACS) and where further testing may not be beneficial. However, all of the previous studies relied on the investigators’ retrospective appraisal of the patient’s chest pain history and used a nonstandard definition for classifying chest pain. We adopted the American Heart Association’s clinical classification of chest pain to modify the HEART score and prospectively evaluated its ability to predict outcomes in patients with low-risk chest pain.

METHODS: Patients presenting with chest pain were considered at low risk for ACS and were referred to the observation unit of an 805-bed hospital if they had none of the following features: ongoing chest pain, dynamic ST-segment changes and cardiac troponin I ≥ 0.20ng/mL. As part of the initial evaluation, clinicians filled out a standard data collection tool that included clinical variables necessary to calculate the modified HEART Score (Table 1). The occurrence of the composite endpoint of death from any cause, acute myocardial infarction and coronary revascularization within 30days of the visit were recorded through follow-up phone call and medical records review. Sensitivity, specificity, and positive and negative predictive values were calculated.

RESULTS: Of the 420 patients that were enrolled, 298 (71%) had complete follow-up data and were included in the analysis. The composite endpoint occurred in 9 (3%) patients within 30days (1 died, 3 had nonfatal acute myocardial infarction and 5 had significant disease requiring coronary revascularization). A cardiac endpoint occurred in 3 of 155 (1.9%) patients with a HEART score 0 to 3, and in 6 of 143 (4.2%) with a score of >3 (p = 0.25). The sensitivity, specificity, and positive and negative predictive values of a HEART score >3 in predicting a cardiac endpoint are shown in Table 2.

CONCLUSIONS: The incidence of significant coronary events is very low among patients presenting with chest pain who do not have ongoing chest pain, dynamic ST-segment changes or abnormal cardiac troponin. The modified HEART score does not appear to be useful in further identifying those who are at very low risk.

EVALUATION OF THE MODIFIED NORTH AMERICAN CHEST PAIN RULE TO IDENTIFY PATIENTS AT VERY LOW RISK OF ACUTE CORONARY SYNDROMEMatthew Cerasale; David Paje; Carlos Calle-Muller; Iani Patsias. Henry Ford Hospital, Detroit, MI. (Tracking ID #1940296)

BACKGROUND: The North American Chest Pain Rule (NACPR) was developed to identify patients presenting with chest pain who are at very low risk for a cardiac event and who may be suitable for discharge from the emergency department without additional investigations. However, this tool defines chest pain as typical for acute coronary syndrome based on the judgment of the attending emergency physician, which makes it difficult to standardize. We adopted the American Heart Association’s definition of typical angina to modify the chest pain component of the NACPR and we prospectively evaluated its ability to predict outcomes among patients with low-risk chest pain.

METHODS: Patients were considered to have low-risk chest pain and were referred to the observation unit of an 805-bed hospital if they had no ongoing chest pain, no dynamic ST-segment changes and no abnormal cardiac troponin value. During their initial evaluation, clinicians filled out a standard data collection tool that included clinical variables necessary to capture the components of the modified NACPR (Table 1). Primary outcome was the composite of all-cause mortality, acute myocardial infarction and coronary revascularization within 30days of the visit. Follow-up was done through phone call and medical records review. Test characteristics of the modified rule were calculated.

RESULTS: Baseline data was recorded on 419 patients, one patient was excluded because of ST-segment elevation in the initial ECG. A total of 298 patients had complete follow-up data (72% with phone follow-up) and were included in this analysis. The composite endpoint occurred in 9 (3%) patients within 30days (1 died, 3 had nonfatal acute myocardial infarction and 5 had significant disease requiring coronary revascularization). All 9 events occurred in patients that had at least one component of the modified NACPR. The sensitivity, specificity, and positive and negative predictive values of the presence of any component of the modified NACPR in predicting the primary outcome are presented in Table 2.

CONCLUSIONS: The modified NACPR appears useful in identifying chest pain patients who are at very low risk of acute coronary syndrome and who may benefit from early discharge from the emergency department.

EXAMINING PERCEIVED BARRIERS TO GOOD NUTRITION IN A SUBSET OF HOMELESS WOMEN IN URBAN BIRMINGHAM, ALABAMASandhya L. Kumar¹; Jeremey B. Walker¹; Dennis J. Pillion²; Amanda Willig²; April Agne²; Andrea Cherrington². ¹University of Alabama at Birmingham, Birmingham, AL; ²University of Alabama at Birmingham, Birmingham, AL. (Tracking ID #1939137)

BACKGROUND: The relationship between homelessness and poor health is well-recognized. Many of today’s food-insecure homeless suffer from obesity-related complications, including hypertension and diabetes. Homeless individuals who regularly access meals provided by charitable services are at the mercy of well-meaning food-providers. The lack of control over quality and quantity of meals is often viewed as an insurmountable barrier, and can be particularly frustrating for patients and their health educators/providers as they attempt to discuss dietary modifications as an avenue for chronic disease management. The resulting apathy further perpetuates disregard for nutrition. In this study, we used a mixed-methods approach to further explore the various factors that influence dietary choice in the homeless population. In partnership with three community organizations, we interviewed and surveyed women who sought services from an urban Birmingham charitable meal program with the aim of exposing the perceptions and attitudes they held regarding good nutrition, as well as to assess if any relationships existed between factors such as health status, extent and accuracy of nutrition knowledge, food security, and determinants of healthy food choices.

METHODS: A convenience sample of 49 women was recruited over a 6week period (Summer 2013) based on participation in meal services provided by a partner agency’s charitable organization. In this mixed-methods study, 11 women in a transitional housing program took part in a semi-structured interview focused on perceived barriers to good nutrition and an 83-item questionnaire on demographics, health status, extent and accuracy of nutrition knowledge, food security, and determinants of healthy food choices. Another 38 women recruited from the same institution’s “Day Center” participated in the survey alone. Quantitative analysis was performed using SPSS. Qualitative content analysis was conducted to identify major categories and substantive themes.

RESULTS: Mean age category was 41–50years. Diabetes and hypertension prevalence were 20% and 53%, respectively. Sixty-three percent of participants express a desire for a “special diet”, 81% citing for health concerns. Thematic analysis of interview scripts reveals food concepts (good versus bad foods, utility versus consequences), desired nutrition knowledge (food budgeting, how to shop, components of food), and influences on food choices (availability, access, competing demands, social reciprocity).

CONCLUSIONS: This subset of homeless experiences a high rate of hypertension and diabetes. Many in the homeless population display an interest in learning more about nutrition and having more control over dietary intake as a way of health management. However, for those on the streets or in temporary housing, meeting immediate physiological needs is the highest priority and dietary control is a secondary or tertiary concern. For these individuals, it seems most appropriate to impact diet by increasing the dietary adequacy and nutrition density of charitable meals. Nutrition-based education and lifestyle modification is a more practical and a desired avenue of chronic disease prevention and management in the transitional population. Individuals in transitional programs are focused on conquering higher order needs, such as security of employment, housing, and health. Given this consideration, this group seems the most amenable to productive discussion of dietary choices and nutrition education. Permanent incorporation of nutrition-based life-skills classes that focus on practical application of nutrition knowledge (i.e. food budgeting, how to shop, etc.) is a long-term investment that will help redefine concepts of food and benefit health status, financial status, and future generation’s health-promoting behaviors.

EXAMINING THE POTENTIAL FOR CPOE SYSTEM DESIGN AND FUNCTIONALITY TO CONTRIBUTE TO MEDICATION ERRORSArbor J. Quist¹; Alexandra Robertson¹; Thu-Trang Thach¹; Lynn A. Volk²; Adam Wright^1,2; Shobha Phansalkar^1,3; Sarah P. Slight^1,4; David W. Bates^1,5; Gordon D. Schiff^1,5. ¹Brigham and Women’s Hospital, Boston, MA; ²Partners HealthCare, Wellesley, MA; ³Wolters Kluwer Health, Indianapolis, IN; ⁴University of Durham School of Medicine, Durham, United Kingdom; ⁵Harvard Medical School, Boston, MA. (Tracking ID #1937921)

BACKGROUND: The Institute of Medicine has estimated that at least 1.5 million people are harmed by medication errors each year, thus efforts to reduce medication errors are an essential component to improving patient safety. While computerized prescriber order entry (CPOE) systems have reduced medication errors by creating legible and structured prescriptions and incorporating effective clinical decision support, they have also introduced new types of errors and must be optimally designed if medication errors are to be minimized. The FDA Brigham and Women’s Hospital Computerized Prescriber Order Entry Medication Safety Project is an FDA-funded study conducted to better understand 10 CPOE systems, their functionality, features, interfaces, and safety, at 6 sites.

METHODS: We examined 10 different CPOE systems that varied in providers, vendors, and care settings across 6 different healthcare facilities. CPOE functionality data were collected through usability testing and detailed interviews with key CPOE system users and managers at each site. Test patients were created in each system and a CPOE Medication Safety Assessment Tool was developed to assist in eliciting and collecting structured information regarding the systems and their functions. These interviews and system evaluations were recorded, transcribed, and organized to highlight the similarities and differences between functionality in the systems.

RESULTS: We found a high level of inconsistency in the functionality of the CPOE systems evaluated, both within and across systems. Differences were found based on the system, care setting, care provider, and user role. Inpatient and outpatient systems at the same site were often unable to communicate, causing confusion and burdening clinicians. Users with different roles, such as physicians, nurses, and pharmacists, were found to have different system capabilities. Differences in workflow and medication prescribing across specialties led to some CPOE functionality that varied according to specialty. We found variation in discharge modules, override reasons, and hard stops. In searching institutional “paper trails” for CPOE system change reasons, we found a lack of system documentation of modifications and features. Some unanticipated findings included frequent user unawareness of obvious screen items and features as well as the surprising degree of local customization of various vendor-based systems (including customized interfaces, clinical decision support, order capabilities, and drug name presentation). We found studying CPOE system design features and rationale to be challenging as we attempted to connect the local patient safety logs. We observed differences in the mental models between CPOE system designers and users, complications with creating and using test patients to evaluate functionality, and difficulties in sharing and comparing useful, but proprietary, information. For example, it was often difficult or impossible to create test patients that were pregnant to simulate scenarios that would trigger pregnancy-related alerts.

CONCLUSIONS: A detailed and comprehensive understanding of the functionality of CPOE systems is important in building better systems that will reduce errors due to inefficiencies and inconsistencies in CPOE systems. Design features and tracking of issues arising from experience with these systems need to be more systematically catalogued. To more effectively prevent CPOE frustrations and errors, functionality needs to be improved through more standardized and less confusing interfaces and design features.

EXCESSIVE PERSPIRATION IS A PREDICTIVE FACTOR FOR AGGRAVATION OF RHABDOMYOLYSISMasaki Tago¹; Naoko E. Furukawa^3,1; Yuka Naito¹; Akiko Eiri¹; Hiroki Aihara¹; Motoshi Fujiwara¹; Jun Tokutomi¹; Motosuke Tomonaga¹; Yuta Sakanishi²; Tsuneaki Yoshioka¹; Norio Fukumori²; Masaki Hyakutake¹; Itaru Kyoraku¹; Takashi Sugioka²; Shu-ichi Yamashita¹. ¹Saga Medical School Hospital, Saga, Japan; ²Saga Medical School, Saga, Japan; ³Saga Medical School, Saga, Japan. (Tracking ID #1912342)

BACKGROUND: Rhabdomyolysis is a serious syndrome in which skeletal muscle fibers are damaged and release breakdown products such as creatinine kinase, or myoglobin into circulation. Some of the products are harmful and may lead to acute kidney injury, multiple organ failure, and even death. However, which factors are valuable in predicting the prognosis of this condition remains to be clarified. The objective of the study was to elucidate the predictive factors for the prognosis of this syndrome.

METHODS: We analyzed consecutive 53 cases of rhabdomyolysis hospitalized at the department of general medicine of the Saga University Hospital in Japan by retrospective chart review from 2006 to 2013. The diagnosis of rhabdomyolysis was made by the presence of increased amount of serum creatinine kinase (CK) (>200IU/L). At first, we divided the patients into two groups, one with serious consequences of death or acute kidney injury requiring blood purification therapies such as hemodialysis (HD) or continuous hemodialysis and filtration (CHDF) (Group A) and the others without ones (Group B). Then, we compared the characteristics of the patients of both groups about age, sex, number of days from onset to admission, main reason of admission, symptoms and signs (fever, muscle rigidity, extrapyramidal signs, the state of consciousness, excessive perspiration, dehydration), blood chemistry on admission (CK, blood urea nitrogen, creatinine, albumin, sodium and potassium), and maximum levels (CK, blood urea nitrogen, creatinine).

RESULTS: All of the 53 patients who were enrolled were Japanese (men, 60.4%; women, 39.6%; age 58.6 +/− 17.2). Group A consisted of 5 cases, in which 2 cases required blood purification therapies (HD/CHDF), and 4 cases deceased. According to univariate analysis, sex, age, fever, muscle rigidity, consciousness, CK, creatinine level on admission, and maximum level of CK were not significantly related to serious consequences. Only the finding of excessive perspiration on admission was related to the serious consequences (odds ratio 31.33, 95% confidence interval 2.17–450.07, p-values < 0.05).

CONCLUSIONS: The presence of excessive perspiration on admission is a useful factor to predict the serious prognosis of the patients who are admitted because of rhabdomyolysis.

EXPENSIVE DREAMS: SLEEP DURATION AND INCOME LEVELS IN GENEVA, SWITZERLANDIdris Guessous^2,1; Cédric Gubelmann¹; Jean-Marc Theler²; José Haba-Rubio³; Jean-Michel T. Gaspoz²; Pedro Marques-Vidal⁴. ¹Lausanne University Hospital, Lausanne, Switzerland; ²Geneva University Hospitals, Geneva 14, Switzerland; ³Center for Investigation and Research in Sleep, Lausanne University Hospital, Lausanne, Switzerland; ⁴Institute of Social and Preventive Medicine (IUMSP), Lausanne, Switzerland. (Tracking ID #1935443)

BACKGROUND: Sleep has been shown to depend on socioeconomic status (SES), but the associations remain controversial. We aimed to assess associations between sleep duration and several markers of SES (education, monthly income, and nationality).

METHODS: Adults of a representative sample of the Geneva population were recruited. Self-reported sleep duration was assessed by questionnaire, where the participants indicated the usual time they went to bed and they woke up.

RESULTS: The analysis included 3853 adults (50% women, 51.7 ± 10.9years). After adjusting for survey year, sleep duration was higher among elderly subjects (adjusted mean ± SE: 7.67 ± 0.02, 7.82 ± 0.03 and 8.41 ± 0.04h/day for age groups [35–50], [50–65] and [65+ years, respectively, p < 0.001), and lower in participants who were foreigner (7.77 ± 0.03 vs. 7.92 ± 0.03h/day for Swiss nationals, p < 0.001), with higher education (7.92 ± 0.02 for non-university vs. 7.74 ± 0.03h/day for university, p < 0.001) or higher income (8.10 ± 0.04, 7.84 ± 0.03 and 7.70 ± 0.03h/day for <5′000 SFr; 5′000–9′500 SFr and >9′500 SFr, respectively, p < 0.001). Similar associations were found for prevalence of sleep <8h/day or after multivariate adjustment, except that the association with education was borderline significant (p = 0.065). Finally, no significant trend was found both for reported sleep time (7.7 ± 1.1 in 2005 vs. 7.8 ± 1.1h/day in 2011, p = 0.36) or the prevalence of sleep <8h/day, and no interaction was found between SES markers and study year.

CONCLUSIONS: Income, more than education, is negatively and positively associated with sleep duration and short sleep in the Geneva population. No change in sleep duration and short sleep was observed from 2005 to 2011.

EXPLORING THE ASSOCIATION BETWEEN PATIENT-CENTERED MEDICAL HOME TEAM-BASED CARE ELEMENTS AND EMPLOYEE JOB SATISFACTION AT THE VHAIan A. Randall^1,2; Henry B. Harvey²; David Mohr^3,4; Emily D. Dolan^1,2; Karin M. Nelson^1,5; Paul Hebert^1,5; Stephan D. Fihn^5,6; Christian Helfrich^1,2. ¹University of Washington, Seattle, WA; ²VA Puget Sound Healthcare System, Seattle, WA; ³VA Boston Healthcare System, Boston, MA; ⁴Boston University, Boston, MA; ⁵VA Puget Sound Healthcare System, Seattle, WA; ⁶Veterans Health Administration, Seattle, WA. (Tracking ID #1938409)

BACKGROUND: The Veterans Health Administration (VHA) implemented a patient-centered medical home (PCMH) model, termed Patient Aligned Care Teams (PACT), in 2010. We explored how various team-based care elements of this organization-wide implementation and other employee- and clinic-level covariates were associated with employee job satisfaction.

METHODS: We analyzed a cross-sectional, web-based survey of VHA employees from May 2012 and VHA administrative data. Overall job satisfaction was scored on 5-point scale from “Not at all satisfied” to “Very satisfied” and modeled using multivariate linear regression. PACT team-based care elements that are characteristic of either the presence or absence of an implemented PCMH model were tested for associations with employee job satisfaction.

RESULTS: 4,565 VHA primary care employees, including primary care providers (physicians, physician assistants and nurse practitioners), nurse managers, clinical assistants and clerks from 536 different primary care clinics responded to the survey. Results presented below are coefficients representing the estimated linear change on the 1–5 job satisfaction outcome scale (e.g., a β of 0.5 represents an estimated job satisfaction outcome that is 0.5 points higher). Among PACT characteristics, participatory decision-making (i.e., giving PACT team members a voice in decisions about how the team operates) was associated with higher job satisfaction (s = 0.36, 95% CI: 0.21, 0.52), while a chaotic or stressful work environment (β = −0.49, 95% CI: −0.59, −0.40) and spending time on tasks someone with less training could do (β = −0.15, 95% CI: −0.24, −0.05) were associated with lower job satisfaction. Among employee-level covariates, being a Team Leader (β = 0.13, 95% CI: 0.04, 0.21), Supervisor (β = 0.36, 95% CI: 0.16, 0.55) or Manager (β = 0.57, 95% CI: 0.39, 0.76) were associated with higher job satisfaction compared to having no supervisory responsibility. Being a nurse (β = 0.17, 95% CI: 0.06, 0.28), clinical associate (β = 0.46, 95% CI: 0.34, 0.57), or administrative clerk (β = 0.22, 95% CI: 0.08, 0.35) were associated with higher job satisfaction compared to PCPs. Spending more than 80% of one’s time in primary care (β = −0.21, 95% CI: −0.39, −0.03) was associated with lower job satisfaction. Having VA tenure of 1–2years, 2–5years, 5–10years, 10–15years, 15–20year or more than 20years was associated with lower job satisfaction, compared to having VA tenure less than 6months. Among clinic-level covariates, working in a clinic with a higher mean patient comorbidity (as measured by mean DCG score) was associated with higher job satisfaction (β = 0.14, 95% CI: 0.06, 0.22), while working in a VA Medical Center-affiliated clinic was associated with lower job satisfaction (β = −0.11, 95% CI: −0.20, −0.03).

CONCLUSIONS: Some PACT characteristics hypothesized to correlate with job satisfaction were associated with overall job satisfaction after adjusting for employee- and clinical-level covariates. Participatory decision-making was associated with higher job satisfaction while working in a chaotic or stressful environment and working below one’s competency were associated with lower job satisfaction. Results indicated that job satisfaction may be lower among PCPs, longer-tenured VA employees and primary care personnel, posing potential long-term challenges to the sustainability of the PACT model and retention of key personnel.

EXTENT AND PREDICTORS OF TRAINING IN OFFICE-BASED TREATMENT OF OPIOID ADDICTION IN U.S. RESIDENCY PROGRAMS: A NATIONAL SURVEY OF PROGRAM DIRECTORSLello Tesema; Jeffrey Marshall; Rachel Hathaway; Christina Pham; Camille Clarke; Genevieve Bergeron; James Yeh; Michael Soliman; Danny McCormick. Cambridge Health Alliance and Harvard Medical School, Cambridge, MA. (Tracking ID #1938929)

BACKGROUND: Opioid addiction (OA) is a public health epidemic and is associated with a high mortality rate. Highly effective pharmacotherapies to treat OA exist, but less than 15% of those in need of treatment receive it. While methadone and buprenorphine are the gold standard treatments for OA, buprenorphine alone can be prescribed in office-based settings, potentially providing greater access. Prescription of buprenorphine in ambulatory medical practice is uncommon, however, potentially due to limited instruction of physicians during residency training. However, little is known about the extent to which residency programs provide training in the prescribing of buprenorphine or the treatment of OA.

METHODS: We conducted an email based survey of all U.S. residency program directors (RPD) in Family Medicine, Internal Medicine and Psychiatry that were listed on the website of the American College of Graduate Medical Education (ACGME). Survey items assessed the extent of program training in the use of buprenorphine, in addictions medicine generally, and whether or not residents were required or encouraged to obtain a Drug Enforcement Administration buprenorphine waiver (required to prescribe this medication). We also assessed program directors’ views and attitudes regarding the use of buprenorphine and potential barriers to buprenorphine training. We produced descriptive statistics for all items and also compared responses of RPD across the three program types. We also examined the relationship between RPD views of the value of buprenorphine and the extent of program training in buprenorphine by calculating odds ratios of 95% confidence intervals.

RESULTS: Of the 947 RPD we surveyed, 471 responded (49% response rate). The majority of RPD report that their residents frequently manage patients with opioid addiction (77%), believe buprenorphine is an important treatment option for opioid dependence (88%), and believe increased residency training in buprenorphine would expand access to care for OA patients (73%). Yet a minority of programs train residents in the use of buprenorphine (36%), require or encourage residents to obtain a waiver (23%) or dedicate greater than 12h (over 3years) to formal training to addiction medicines (23%), although psychiatry programs were more likely to do so (data not shown) A minority also believed that prescribing buprenorphine should be a core competency (37%) of their discipline. We also found that RPD belief that buprenorphine is an important treatment option for OA was associated with a greater odds of the program providing training in buprenorphine (3.5 [CI95% 1.6–7.4)]. The most frequently cited barriers to implementing buprenorphine training in residency programs included: shortage of licensed preceptors (77%), competing curricular priorities (64%), and lack of support services (54%).

CONCLUSIONS: Most residency programs frequently care for OA patients and most RPD believe in the effectiveness of buprenorphine and its ability to expand access to treatment for OA. Yet only a minority of training programs offer formal training in the treatment of OA generally, or with buprenorphine specifically. Increasing the exposure of physicians during residency training to ambulatory treatment of OA using buprenorphine could promote greater access to this treatment option.

EXTERNAL VALIDATION OF A SCORING SYSTEM THAT PREDICTS READMISSION OF PATIENTS WITH ACUTE PANCREATITIS TO THE HOSPITAL WITHIN THIRTY DAYS OF DISCHARGEMohammed Aljasmi; Aymen Bukannan; Yathreb Alaali; Ahmed Saeed; Gregory Buran. Henry Ford Hospital, Detroit, MI. (Tracking ID #1939127)

BACKGROUND: Acute pancreatitis (AP) is one of the medical conditions that cause a heavy financial burden on the American health care system, in addition to causing significant morbidity to patients. Studies have estimated that the average cost of hospitalization for acute pancreatitis ~$10k, and acute pancreatitis is responsible for about $2.2 billion in US health care expenditure annually. Studies have shown that the rate for early readmission for acute pancreatitis was found to be close to 20%. Our aim was to validate a recently published scoring system to predict readmission of patients with acute pancreatitis to the hospital within 30days of discharge.

METHODS: A Retrospective review of patients with AP admitted to Henry Ford Hospital from October 2008 to December 2010. Inclusion and exclusion criteria reflected that of the original study. Diagnosis of AP required two of the following: characteristic abdominal pain, serum amylase/lipase more than or equal to three times the upper limit of normal, or CT of the abdomen within the first 7days of hospitalization demonstrating changes consistent with AP. Factors of the scoring system were analyzed at time of discharge including eating less than solid diet; GI symptoms; pancreatic necrosis; antibiotics; pain. Results were analyzed using Logistic regression modeling and a ROC curve to evaluate the strength and performance of the scoring system.

RESULTS: There was 380 patient admitted during the study period with 47 being readmitted within the 30day period for a readmission rate of 12%. An odd ratio of 1.381 (p-value <000.1) indicate that each single unit increase in the risk score corresponds to approximately a 1.381 times increase in the likelihood of 30-day readmission within both settings. The resulting ROC curve areas 0.66 indicate that the risk score is a moderate predictor of 30-day readmission.

CONCLUSIONS: Our findings suggest that the “SNNAP” scoring system was predictive of 30-day readmission.

Comparison in 30day readmission in patients who are low, moderate and high risk for readmission

ROC Curve to predict readmission within 30days

FACTORS ASSOCIATED WITH LOSING AT LEAST 75% OF GESTATIONAL WEIGHT GAIN BY EARLY POSTPARTUM IN WOMEN WITH RECENT GESTATIONAL DIABETESJacinda M. Nicklas^1,2; Chloe A. Zera³; Sue E. Levkoff^4,5; Ellen W. Seely². ¹University of Colorado School of Medicine, Aurora, CO; ²Brigham and Women’s Hospital, Boston, MA; ³Brigham and Women’s Hospital, Boston, MA; ⁴Brigham and Women’s Hospital, Boston, MA; ⁵University of South Carolina, Columbia, SC. (Tracking ID #1937893)

BACKGROUND: Women with gestational diabetes (GDM) have a 30–70% risk for developing type 2 diabetes (T2DM) later in life. Postpartum weight retention is highly predictive for future obesity, and further increases risk for T2DM. We sought to identify factors associated with losing at least 75% of gestational weight gain by early postpartum in women with recent GDM.

METHODS: We recruited women with GDM during pregnancy or just after delivery. Pre-pregnancy weight was self-reported at recruitment; gestational weight gain, mode of delivery, and insulin use were extracted from medical records. At a mean of 7.2 (±2.1) weeks postpartum we measured weight, height, and administered questionnaires, including: demographics, breastfeeding, sleep, Edinburgh Postnatal Depression Scale, Harvard Food Frequency, and the International Physical Activity Questionnaire. After conducting an analysis of univariates, we used backward selection to fit a multivariable logistic regression model for univariates with p < .2 to identify factors associated with losing 75% or more of gestational weight gain by early postpartum.

RESULTS: Our study included 75 women (mean 33 ± 5years; pre-pregnancy BMI 31.4 (±5.6) kg/m2; 57% White, 29% African-American, 15% Asian, with 20% identifying as Hispanic and 34% low income). Mean pregnancy weight gain was 12.5kg (±7.8). Fifty-two percent of participants lost at least 75% of their pregnancy weight gain by the study visit. Dietary composition, fiber intake, added sugars, and glycemic index were similar between those who demonstrated early weight loss and those who did not, and consequently were not included in the multivariate model. Physical activity, sedentary time, race/ethnicity, education, income, mode of delivery, postpartum depressive symptoms, and breastfeeding were also similar between groups and not included in the model. In a multivariate model adjusting for weeks postpartum, only increased age (OR 1.31; 95% CI 1.06 to 1.76 per year) and less gestational weight gain (OR 0.63; 0.48 to 0.76 per kg) were associated with at least 75% early postpartum weight loss. Insulin use, nulliparity, and sleep were not retained in the model.

CONCLUSIONS: A substantial proportion of women with recent GDM lost at least 75% of their gestational weight gain by early postpartum. Older women and those who had gained less weight during pregnancy were significantly more likely to have lost at least 75% of gestational weight. Limiting gestational weight gain may help women with GDM return to pre-pregnancy weight quickly during the postpartum period.

FACTORS ASSOCIATED WITH PATIENT’S VIEW ABOUT DISAPPEARENCE OF THEIR CHRONIC PAINMichel P. Kossovsky¹; Valerie Piguet²; Christine Cedraschi¹; Anne-Francoise Allaz¹. ¹Geneva University Hospitals, Geneva, Switzerland; ²Geneva University Hospitals, Geneva, Switzerland. (Tracking ID #1935509)

BACKGROUND: Chronic pain is a frequent complaint among internal medicine patients. One important aspect in the care of such patient is to understand how they apprehend the manifestation of the pain and what they expect for the future. The aim of our study was to investigate the variables associated with chronic pain patients’ expectations about the possibility of pain disappearance.

METHODS: A self-administered questionnaire was given to all patients referred to our pain centre over a 2year period. Patients were asked about the likelihood of pain disappearance (dichotomic answer: possible/probable vs. unlikely/impossible). Multivariate logistic linear regression analysis evaluated which variables were associated with the likelihood of pain disappearance. The model was controlled for age and gender, the presence of depression (evaluated with the Beck Depression Inventory), anxiety (evaluated by the Hospital anxiety and depression scale), present pain intensity (evaluated on a visual analog scale (0–10cm)), duration of chronic pain, attribution of pain origin (extrinsic, intrinsic or undetermined) and perceived level of illness on a 4 level scale (from considerably to not at all).

RESULTS: We collected information on 437 patients. Mean age was 53; women were 284 (65%). Table represents the odds-ratios associated with patients’ estimation of the likelihood of pain disappearance.

CONCLUSIONS: Neither age nor sex were associated with likelihood of pain disappearance according to patients. Interestingly, pain intensity, anxiety or causal attribution were not either associated with likelihood of pain disappearance. Depression, a longer duration of chronic pain and a higher perceived level of illness were significantly associated with a reduced likelihood of pain disappearance. Practitioners confronted with chronic pain patients should be aware that these three variables are linked with patients’ views about the future of their pain and taken into account when planning the treatment.

FACTORS ASSOCIATED WITH THE DECISION TO UNDERGO CONTRALATERAL PROPHYLACTIC MASTECTOMY AT THE TIME OF A BREAST CANCER DIAGNOSISKyungsuk Jung^1,2; Kala Visvanathan². ¹Presence Saint Francis Hospital, Evanston, IL; ²Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD. (Tracking ID #1921712)

BACKGROUND: Increasing number of breast cancer patients are electing for contralateral prophylactic mastectomy (CPM) to prevent contralateral breast cancer. CPM was shown to be associated with a survival advantage in women with BRCA1/2 mutation carriers, thus at high risk of developing breast cancer with a life expectancy gain of 0.6–2.1years. This study sought to identify factors that may affect the decision of breast cancer patients to undergo CPM.

METHODS: A cohort of >1200 women who are at high risk of developing breast cancer were recruited primarily from the high risk clinic. Women with either positive family history of breast or ovarian cancer, positive BRCA1/2 gene or <40years of age were eligible for inclusion. From this cohort, 370 women diagnosed with breast cancer were identified. All study participants had completed a Questionnaire that included—demographic, disease history, surgical history, lifestyle factors extensive family history of cancer, and genetic testing. Pearson’s χ2 test, univariable analysis and multivariable analysis were used to compare characteristics in both groups.

RESULTS: 81/370 patients with unilateral breast cancer, chose to have a CPM. Univariable analyses showed that younger patients were significantly more likely to undergo CPM (OR, 0.93; 95% CI, 0.91–0.96). Patients were also significantly more likely to under go CPM if they had bilateral oophorectomy (OR, 2.18 95% CI, 1.22–3.87), deleterious result from BRCA test (OR, 1.87, 95% CI, 1.00–3.49), had breastfeeding experience (OR, 2.96, 95% CI, 1.28–3.20) and had a positive family history of breast cancer in 1st and 2nd degree relative (OR, 2.75, 95% CI, 1.22–6.17). Multivariable logistic regression analysis with significant variables from the above outcome revealed that patients who were younger (OR, 0.92, 95% CI, 0.89–0.97), had bilateral oophorectomy (OR, 2.63, 95% CI, 1.18–5.89), had a positive family history of breast cancer in 1st & 2nd degree relative (OR, 4.36, 95% CI, 1.40–13.60) were more like to choose to undergo CPM. BRCA testing, BRCA test result, marital status, menopausal status and breastfeeding experience were not significantly associated with the use of CPM.

CONCLUSIONS: Younger age is the strongest determinant of undergoing CPM in women at high risk for breast cancer which is consistent with their higher risk of a second breast cancer. For women with family history of breast cancer, it is important to identify their genetic risks to help with their decision. History of oophorectomy was also associated with CPM. However, oophorectomy is reported to be a negative factor on the risk of developing CBC. Evidence-driven models are needed to better inform patients.

FACTORS ASSOCIATED WITH UNCONTROLLED HYPERTENSION IN AN URBAN ACADEMIC PRIMARY CARE CLINIC POPULATIONYaakov Y. Liss; Jonathan Arend; Georgia Giebel; Lauren Peccoralo. Mount Sinai School of Medicine, New York, NY. (Tracking ID #1922682)

BACKGROUND: Despite innumerable anti-hypertensive medication treatment options, only 70% of Americans receiving treatment have controlled hypertension. Emerging data demonstrates that using a panel-based approach can dramatically improve hypertension control for a clinic population. However, little data exists on factors associated with uncontrolled hypertension within an urban academic primary care clinic population. Identifying such factors is crucial for designing a hypertension panel management program that will succeed in improving hypertension control for this population. This pilot study aims to identify factors related to uncontrolled hypertension in an urban academic primary care clinic population including patient demographics, co-morbid disease status, specific medication use, and disease control measures, as well as to identify visit specific factors that may contribute to hypertension control.

METHODS: Data included in this study is from a convenience sample of all patients with a diagnosis of hypertension seen in an urban academic primary care clinic on October 1, 2013. Data was obtained using the reporting workbench function of the clinic’s Electronic Medical Record (EMR) system. Predictor variables included patients’ sex, ethnicity, race, age, preferred language, marital status, last blood pressure, BMI, presence of diabetes, CHF, CAD, or asthma, ace inhibitor use, beta blocker use, hemoglobin A1C, LDL, and GFR. The outcome variable of uncontrolled hypertension was dichotomized and defined as uncontrolled if the last blood pressure reading performed in an ambulatory setting exceeded 140/90. For categorical non-numeric predictor variables, a chi square test was used to determine an association with blood pressure control, and for continuous predictor variables, a one-way anova test was used. Using the same methods as above, data for all hypertensive patients (>5000) seen in the clinic over a 1-year period will be recorded and analyzed. Additionally, a chart review was performed of 32 randomly selected patients with uncontrolled hypertension to identify the management plan. Eventually, 200 charts will be reviewed by two investigators, with a 3rd investigator to resolve discrepancies.

RESULTS: Based on the convenience sample described above, 261 patients were identified, 40% of whom had uncontrolled hypertension. Baseline demographic data of the 261 patients identified can be found in Table 1. There was a statistically significant association with gender and uncontrolled hypertension (73% males uncontrolled versus 27% females uncontrolled, p = 0.0001), and with race-ethnicity and uncontrolled hypertension (Non-Hispanic African American descent versus Hispanic Non-African American descent versus Non-Hispanic Non-African American Descent, 51% versus 35% versus 25%, p = 0.008). There were no other statistically significant associations. The results of the chart review demonstrated seven common scenarios and management plans (or lack thereof) during the visit when the last (uncontrolled) blood pressure measurement was taken. These are described and quantified in Table 2.

CONCLUSIONS: This study demonstrated significantly worse hypertension control in African American patients and male patients, results that differ significantly from recent national trends where Hispanic and female patients were noted to have worse hypertension control. It is unclear why this discrepancy exists and whether these results are specific to this clinic population or are reflective of a broader trend in urban academic primary care clinics. Additionally, the identification of seven management plans occurring during the patient-doctor interaction presents several different possibilities of interventions that may improve overall hypertension control in this clinic population. Among these are instituting a policy encouraging providers to schedule patients with uncontrolled hypertension for follow up within 1month; EMR alerts to a patient’s primary doctor when an elevated blood pressure is recorded by a consultant; and a system to remind patients of upcoming appointments.

Table 2. Patient Doctor Interaction Scenarios

BP = blood pressure; f/u = follow up; Pt = patient

Table 1. Patient Demographics. Total 261 Patients.

AA = African American

FACTORS INFLUENCING MEDICAL STUDENT USE OF ELECTRONIC HEALTH RECORDS: A NEEDS ASSESSMENTDylan Sherry; Megan S. Mcnamara. Case Western Reserve University, Cleveland, OH. (Tracking ID #1934496)

BACKGROUND: Physicians are increasingly using Electronic Health Records (EHRs) in their daily practice, but there is a lack of formalized training about EHRs in most medical school curricula. Education in the use of EHRs for medical students has typically been unofficial “on the job” training by superiors or “how-to” online modules which focus on how to accomplish a specific task using the EHR, such as documenting a history and physical, or writing a progress note. These narrow approaches may lead students to consider the EHR as merely a repository of information, rather than a system which can potentially improve physician workflow and patient care. In order to design effective curricula for medical students, we conducted a needs assessment to understand how students are currently using the EHR and what factors influence their use.

METHODS: We developed a seven-item electronic questionnaire which surveyed students about how they would use the EHR in a particular clinical scenario. Each item was associated with a list of potential responses, and students were asked to justify their choice through a written explanation. The clinical scenarios were developed using the RIME educational framework (Reporter-Interpreter-Manager-Educator) approach to EHR education. We mapped each question to both a RIME category and to ACGME core competencies. For example, a question about whether or not a student might copy and paste a previous day’s progress note to save time would be mapped to the RIME category “Reporter” and to the ACGME core competency “Professionalism.” In order to ensure content validity we sent the questionnaire to 10 national informatics experts for recommendations and made suggested changes. All Case Western Reserve University students were eligible for participation and were recruited via email invitation.

RESULTS: The questionnaire was sent to 495 medical students (Preclinical, Core Clerkships, Elective/Acting Internship) with a 39% return rate. Using a framework analysis approach we identified 11 specific factors which influence students’ use of the EHR. These included efficiency, communication tool, accuracy of documentation, influence on physician/patient communication, trusting the EHR, intrusive, role modeling, familiarity with the EHR, habit, advancing learning, and impediment to learning. For example, efficiency was a factor that drove students to use templates or copy and paste. Influence on physician/patient communication was a factor which determined how, if at all, the EHR was used during a clinical encounter. These factors could be broadly grouped into how students view the positive and negative aspects of the EHR as it relates to patient care, how students learn about the EHR, and how the EHR influences their medical education overall. The following exemplar quotes help to illustrate why medical students chose to use the EHR in certain ways: “Flipping through fill-in-the-blanks is fastest, but it makes me feel like a robodoctor. I would rather write it out to really make myself learn it, but when you don’t have much time, [templates] are the best.” Core Clerkship Student “I like to be able to shape my differential in my head without a pre-conceived notion of what is important. When I am entering it [in the EHR] afterwards, I am happy that there are templates available so I can be reminded of what I may have missed.” Core Clerkship Student “I [copy and paste] for efficiency in tracking the patient, not just to save time. I like having my note in the same essential format from 1day to the next so that my eye can scan quickly to pertinent changes. I can quickly see how many days an antibiotic has been taken, which consults have returned recommendations, etc. If I wrote each note from scratch I would have to read the entire thing the next day to find what I wanted to know.” Core Clerkship Student “Copy/pasting is dangerous and can lead to incorrect information in the chart. Always start from scratch and you’ll never be inaccurate. Reference old notes to ensure you’re not missing anything, but don’t copy paste.” Elective/Acting Internship Student

CONCLUSIONS: Medical students are quite aware of the positive and negative impacts that the use of EHRs can have on patient care and on medical education. These factors could be used as a springboard for EHR education. Alongside the current how-to focus, education could be enhanced by addressing student’s concerns about the potential negative impacts of the EHR, such as copy and pasting or interrupting the clinical encounter. Student decision-making regarding EHR use suggests that future EHR education should be placed more in the context of clinical medicine to help students learn to mitigate negative effects on medical education and on patient care.

FACULTY PERCEPTIONS OF COMPETENCY-BASED EVALUATION OF MEDICAL STUDENTSElizabeth Karwowski^1,2; Elizabeth Sandman²; Jane M. Liebschutz¹; Bruce Fraser³; Jay D. Orlander^2,1. ¹Boston University Medical Center, Boston, MA; ²Boston VA Healthcare System, Boston, MA; ³Boston University School of Education, Boston, MA. (Tracking ID #1938972)

BACKGROUND: The Accreditation Council for Graduate Medical Education (ACGME) transitioned to competency-based evaluation with its Outcome Project in 1999. The Liaison Committee on Medical Education (LCME) adopted the same core competencies to evaluate medical students: medical knowledge, patient care, interpersonal communication, professionalism, practice-based learning and improvement, and systems-based practice. In May 2012 Boston University School of Medicine changed all student clerkship evaluation forms to a standardized, competency-based format. Little is known about faculty perception of the use of the core competencies to evaluate medical students or their process of student evaluation.

METHODS: Semi-structured, one-on-one interviews were conducted with Department of Medicine physician faculty. Eligibility criteria included evaluating six or more third and/or fourth year clerkship students per year in any clinical setting. Interview questions explored types of faculty interaction with students, faculty attitudes toward student evaluation, and faculty attitudes towards evaluating medical students using the ACGME core competencies. All interviews were transcribed verbatim. Co-investigators reviewed transcripts and identified common themes using Grounded Theory. Two separate investigators coded interviews and achieved consensus on the themes present, resolving discrepancies through discussion. The remaining three co-investigators, with more qualitative experience, checked the final coding schemes for accuracy.

RESULTS: Twelve faculty members with a range of clinical responsibilities (primary care, hospitalist, specialist) and experience participated in the study. All faculty acknowledged that their evaluations were limited due to a lack of direct observation for many competencies. The majority primarily observed case presentations and used these as a proxy for areas they were unable to observe. One faculty commented: “the majority of my evaluation is gonna be based on how they presented the patient to me… so, even if I’m not observing how they’re getting the data, if they come back and they’ve got pretty good data… I can technically check these off [evaluation questions on global assessment form].” For competencies, such as systems based practice, faculty had few, if any, objective or measurable activities to evaluate. Faculty identified multiple barriers to student evaluation including time, other learners, patient care responsibilities, and lack of training in evaluation methods. Faculty expressed significant tension over the evaluation process. They contrasted a sense of inadequacy in their ability to reliably evaluate students with a heightened awareness of the significance of their evaluations in determining student grades. In particular, faculty felt they could identify and evaluate students at the extremes of performance but struggled with discriminating between students in the middle. One faculty commented: “I feel like… there’s not enough ability to distinguish [between students], just using this scoring system. And… presumably, that’s the intent.” The purpose and ramifications of medical student evaluation were not well understood: does evaluation certify that a student is competent or does it provide a means of stratification between students? Often this resulted in self-reported grade inflation.

CONCLUSIONS: Competency-based evaluation on the clinical clerkships presents a challenge to many experienced faculty members. Faculty development around student evaluation that explicitly addresses student milestones and expectations, purpose of evaluation, and feedback on an individual’s grading practices may be beneficial in improving the process. Given the frequency at which it seems to occur, further work needs to elucidate the association of presentation skills as a proxy for other clinical skills.

FAILING TO CHOOSE WISELY: LACK OF TREATMENT DE-INTENSIFICATION IN OLDER PATIENTS WITH DIABETESEve A. Kerr^1,2; Rob Holleman¹; Mandi L. Klamerus¹; Sameer D. Saini^1,2; Lillian Min^1,2; Sandeep Vijan^1,2; Jeremy Sussman^1,2; TImothy Hofer^1,2. ¹VA Center for Clinical Management Research, Ann Arbor, MI; ²University of Michigan, Ann Arbor, MI. (Tracking ID #1936536)

BACKGROUND: It is increasingly recognized that efforts to intensively improve hemoglobin A1c (A1c) and blood pressure (BP) control in patients with diabetes sometimes unintentionally result in overtreating patients who are unlikely to benefit. Recent guidelines and initiatives like Choosing Wisely recommend less aggressive treatment for older patients with limited life expectancy (LE), since such treatment is unlikely to improve outcomes and may cause harm. This suggests that medication de-intensification is indicated among many patients with limited LE. Yet, we know little about how often providers decrease treatment intensity among patients who are unlikely to benefit. We examined how often the number or dose of medications are decreased in older patients with diabetes and lower than recommended levels of A1c or BP, and whether medication de-intensification is more common in patients with lower LE.

METHODS: Using the Veterans Health Administration’s (VHA) Corporate Data Warehouse, we identified all patients with diabetes, age 70 and older, receiving primary care in VHA during 2012. We then identified the patients’ last A1c and BP during 2012, and their prescribed medications at the time of this index A1c or BP. Patients with A1c < 6.5% and on hypoglycemic medications (other than metformin alone) were considered eligible for de-intensification. Patients with BP < 130/65 and on BP medications (other than low dose angiotensin converting enzyme inhibitor (ACEI) alone or low dose angiotensin receptor blockade (ARB) alone) were similarly eligible. We determined the proportion of patients who had medications discontinued or had the dose decreased in the 6months after the index A1c or BP. Using logistic regression, we examined the association between limited LE, based on the Charlson comorbidity index, and de-intensification, controlling for number of medications the patient was taking at the time of the index A1c or BP.

RESULTS: 500,742 patients with diabetes, age 70+, were receiving primary care in VHA during 2012. Among the 114,411 with A1c < 6.5%, 51% were on no medications, 15% were on metformin only, 8% were on insulin only, 14% were on sulfonylureas only, and 12% were on other or combination therapy. Among the 36,686 eligible for de-intensification, 8,466 (23%) were de-intensified within 6months. Patients with LE < 5years were modestly more likely to be de-intensified than those with LE > 10years (predicted probability 29% vs 19%, p < 0.001). Among the 66,750 patients with a BP < 130/65, 10% were on no medications, 6% were on low dose ACEI or ARB only, 22% were on 1 BP medication, and 62% were on 2 or more BP medications. Among those eligible for de-intensification, 9,883 (21%) were de-intensified. Patients with LE < 5years were somewhat more likely to be de-intensified than those with LE > 10years (predicted probability 24% vs 17%, p < 0.001).

CONCLUSIONS: Despite increasing awareness that overtreatment of diabetes may be harmful in older adults, less than one quarter of patients treated to levels significantly lower than recommended had medications de-intensified. While patients with limited LE were more likely to be de-intensified than those in good health, the majority of eligible older patients, even those with limited LE, continued to receive care that is of low value or potentially harmful. Future initiatives that seek to help patients and providers choose wisely will need to address this significant clinical inertia for de-intensification of treatment.

FEAR OF INDEPENDENCE LOSS (FOIL): OLDER ADULT PERCEPTIONS ON REFUSAL OF HOME CAREVanessa Ramirez-Zohfeld¹; Kenzie A. Cameron¹; Priya Sunkara¹; Chris Forcucci²; Megan Huisingh-Scheetz³; Lee Lindquist¹. ¹Northwestern University, Chicago, IL; ²Aging and In-Home Services of Northeast Indiana, Fort Wayne, IN; ³University of Chicago, Chicago, IL. (Tracking ID #1935965)

BACKGROUND: Remaining in one’s own home is often one of the highest priorities of older adults. As their physical, functional, cognitive, and social needs increase, older adults may require more care in their home which they may illogically refuse. We sought to understand seniors’ perceptions, in particular fears that they faced related to the potential loss of independence and refusal of accepting care in the home.

METHODS: We conducted eight focus groups with older adults (age 65 and older) living in rural, suburban and urban areas. Following a semi-structured interview guide, participants discussed advanced planning, decision-making, barriers, and resources that might have an impact on their ability to remain in their own home. We elicited participant’s thoughts regarding losing independence. Three independent coders used content and constant comparative analysis to interpret results.

RESULTS: Sixty-eight older adults participated in the focus groups with a mean age of 73.9years (SD = 6.53); 50 (73.5%) were female. Several major themes related to fear of independence loss (FOIL) as well as strategies to overcome FOIL emerged. FOIL was associated with loss of control, resentment of depending on others, annoyance with working around other people’s schedules, as well as guilt of being burdensome. One participant stated: “I had to depend on friends and neighbors to fix food or shop or do anything, and I thought, I don’t want to be in that position.” A feeling of burden, especially related to their children was a prominent theme across the focus groups with one participant stating: “I don’t want to be a hinder. I don’t want to stop their living. Because I have enjoyed my life and I want them to enjoy their life.” Lack of trust in others and concerns about being exploited were also identified as reasons for FOIL: “At times, it seems like trying to get across a mine field avoiding exploitation from caregivers, from relatives and it is very disconcerting. Who can you trust?” Conversely, multiple participants re-framed the concept of independence and saw the actions as more inter-dependent. Someone noted that even if one becomes more limited over time, the recognition that one is still contributing something meaningful to society is important to overcoming FOIL: “I think we forget what the other person is contributing… the person who is able to see his grandchildren to smile at them; I mean he’s giving them something as well.” Another strategy presented to overcome FOIL was recognition that others may engender positive emotions by helping someone who is in need once they are less able to do so themselves: “When I’m on the receiving end, I’ve got to remember that that’s giving somebody else some joy.”

CONCLUSIONS: Older adults perceive fear of independence loss, which can impact their willingness to accept help. Addressing and promoting strategies to overcome this fear may aid in understanding how to assist seniors to obtaining and accepting needed home care, enabling them to remain safely in their own homes.

FINANCIAL BARRIERS TO CARE AMONG THE INSURED IN MASSACHUSETTS AFTER HEALTH CARE REFORM: DISPARITIES BY RACE, INCOME, AND HEALTH STATUSSonali Saluja^1,2; Leah Zallman^4,2; Steffie Woolhandler³; David Himmelstein³; David Bor^2,1; Danny McCormick^2,1. ¹Harvard Medical School, Boston, MA; ²Cambridge Health Alliance, Cambridge, MA; ³City University of New York School of Public Health, New York, NY; ⁴Institute of Community Health - Cambridge Health Alliance, Cambridge, MA. (Tracking ID #1936234)

BACKGROUND: Massachusetts’ 2006 Health Care Reform, the model for the Affordable Care Act (ACA), aimed to increase coverage and access to health care by making care more affordable, particularly for minorities, persons with low incomes and persons in poor health. Little is known about the post-reform prevalence of financial barriers to care among the insured in Massachusetts.

METHODS: We analyzed data from Massachusetts Behavioral Risk Factor Surveillance System, a representative statewide cross-sectional population-based survey of adults, pre-reform (2004–2005) and post-reform (2009–2010). The sample included adults ages 18 to 65 who reported having health insurance (2004–2005: N = 11,876; 2009–2010: N = 21,782). We considered respondents to have a financial barrier to care if they indicated that within the last 12months they needed to see a doctor but could not because of cost. We compared rates pre and post-reform, overall and for racial/ethnic, income and health status groups. We used multivariate logistic regression models to assess the independent effect of race and income on financial barriers to care after adjusting for other covariates. Finally we examined the association between financial barriers to care and potential markers of need for care and utilization of care.

RESULTS: Among all non-elderly insured adults in Massachusetts, the proportion reporting a financial barrier to care did not change after state health care reform (pre-reform:5.8%, post-reform:6.0%; p = 0.60). However, the proportion citing such a barrier rose 37.9% (from 9.5 to 13.2%; p = 0.004) among low-income individuals (incomes $15,000–$34,999)—many of whom acquired subsidized insurance requiring copayments for physician visits under the reform. Post-reform, financial barrier rates varied by race and income group—with the highest rates in low-income minority individuals. Among persons with incomes below $15,000, 24.9% of Blacks and 24.2% of Hispanics reported financial barriers vs. 9.8% of Whites (p = 0.011). In multivariate analyses, lower income individuals reported more financial barriers than those with incomes over $50,000 (<$15,000: OR = 5.68 95%CI = 5.61–5.74, $15,000–$34,999: OR = 4.91; 95%CI = 4.87–4.95, $35,000–$49,999: OR = 3.43; 95%CI = 3.40–3.47) as did minorities compared to Whites (Black: OR = 1.18; 95%CI = 1.16–1.19, Hispanic: OR = 1.45; 95%CI = 1.43–1.46). Individuals reporting financial barriers were less likely to have a primary care provider (RR = 0.91; 95%CI = 0.86–0.95) and to have a routine health care visit in the last year (RR = 0.89; 95%CI = 0.83–0.95), but were more likely to have two or more chronic health conditions (RR = 1.26; 95%CI = 1.10–1.44) and much more likely to be in poor or fair health (RR = 3.13; 95%CI = 2.63–3.77).

CONCLUSIONS: After full implementation of health reform in Massachusetts, financial barriers to care among Black, Hispanic and low-income residents are common and strongly related to worse health status. These findings raise concern about the likely persistence of financial barriers to care nationwide after implementation of the ACA.

FINANCIAL INCENTIVES FOR COMPLETION OF FECAL OCCULT BLOOD TESTS AMONG VETERANS: A 2-STAGE PRAGMATIC CLUSTER RANDOMIZED, CONTROLLED TRIALJeffrey T. Kullgren¹; Tanisha N. Dicks²; Xiaoying Fu²; George L. Tzanis^2,3; Martin Tobi^2,3; Steven C. Marcus^2,3. ¹Ann Arbor VA Healthcare System and University of Michigan, Ann Arbor, MI; ²Philadelphia Veterans Affairs Medical Center, Philadelphia, PA; ³University of Pennsylvania, Philadelphia, PA. (Tracking ID #1940020)

BACKGROUND: Rates of primary care patient completion of prescribed fecal occult blood tests (FOBTs) are often low. A promising tool for improving rates of FOBT completion is financial incentives, which capitalize on insights from behavioral economics that suggest patients disproportionately weight near-term relative to distant costs and benefits. The objective of this study was to identify the dose and design of modest financial incentives that could increase rates of initial completion of prescribed FOBTs among Veterans Health Administration (VHA) primary care patients.

METHODS: We conducted a 2-stage pragmatic cluster randomized, controlled trial in the primary care clinics of the Philadelphia Veterans Affairs Medical Center. All 1,549 patients who were prescribed an FOBT during the study period were enrolled (i.e., there were unique samples of 713 patients in stage 1 and 836 patients in stage 2). In stage 1, patients were assigned to usual care or $5, $10, or $20 for FOBT completion within 30days. In stage 2, different patients were assigned to usual care or $5, a 1 in 10 chance of $50, or a $500 raffle for FOBT completion within 30days. The primary outcome was FOBT completion within 30days. Secondary outcomes included time to FOBT completion and, among patients who had failed to complete a prescribed FOBT in the previous year, 30-day FOBT completion and time to FOBT completion.

RESULTS: In stage 1, $5 (40.4%, 95% CI, 30.4% to 51.3%; P = 0.96), $10 (44.9%, 95% CI, 37.6% to 52.3%; P = 0.44), and $20 (45.3%, 95% CI, 38.4% to 52.4%; P = 0.38) did not lead to a statistically significant increase in FOBT completion compared to usual care (40.7%, 95% CI, 33.3% to 48.6%). In stage 2, a 1 in 10 chance of $50 led to a statistically significant increase (47.8%, 95% CI, 39.5% to 56.3%; P = 0.002) but $5 (38.9%, 95% CI, 29.5% to 49.3%; P = 0.15) and a $500 raffle (41.5%, 95% CI, 32.4% to 51.2%; P = 0.05) did not lead to a statistically significant increase in FOBT completion compared to usual care (30.7%, 95% CI, 25.0% to 37.0%). In each stage there were no significant differences in time to FOBT completion and, among patients who had failed to complete a prescribed FOBT in the previous year, no significant differences in 30-day FOBT completion or time to FOBT completion.

CONCLUSIONS: Fixed payments of $5, $10, and $20 did not increase rates of VHA patient completion of prescribed FOBTs within 30days. However, when we tested 3 different financial incentive designs that shared a common upfront per patient allocation of $5, we found that a 1 in 10 chance of $50 was effective at increasing rates of FOBT completion within 30days. Our finding serves as an example of a low-cost, scalable intervention that can improve health care quality by promoting patient adherence to a high-value clinical preventive service and adds to a growing evidence base supporting the effectiveness of financial incentives in improving health behaviors among VHA patients.

FINANCIAL ISSUES AND ADULT IMMUNIZATION: MEDICARE COVERAGE AND THE AFFORDABLE CARE ACTLaura Hurley^1,2; Megan C. Lindley⁴; Mandy Allison^1,5; Lori A. Crane^1,6; Michaela Brtnikova¹; Brenda Beaty^1,3; Megan Snow¹; Carolyn Bridges⁴; Allison Kempe^1,5. ¹Children’s Hospital Colorado, Aurora, CO; ²Denver Health, Denver, CO; ³University of Colorado Anschutz Medical Campus, Aurora, CO; ⁴Centers for Disease Control and Prevention, Atlanta, GA; ⁵University of Colorado Anschutz Medical Campus, Aurora, CO; ⁶Colorado School of Public Health, University of Colorado, Denver, CO. (Tracking ID #1935920)

BACKGROUND: Financial barriers are often reported for adult vaccinations. The Affordable Care Act addresses some of these barriers; most notably by mandating private health plans to cover Advisory Committee on Immunization Practices recommended vaccinations with first-dollar coverage when delivered by in-network providers. The law does not change coverage of vaccines by Medicare: Medicare Part D, a pharmaceutical benefit, covers some vaccines, and Medicare Part B, a medical provider office-based benefit, covers others. Medicare part D has been seen as a barrier to physician-based vaccination. Our objectives were to assess among general internists (GIM) and family physicians (FM): 1) how often vaccines are deferred or refused by adult patients for financial reasons 2) how often physicians reported not recommending an adult vaccine for financial reasons 3) knowledge of Medicare coverage of vaccines, and 4) awareness of vaccine specific provisions of the Affordable Care Act.

METHODS: We administered an Internet and mail survey from June to October 2013 to a national network of 438 GIM and 401 FM physicians representative of the American College of Physicians and American Academy of Family Physicians memberships. Descriptive analyses were performed.

RESULTS: Response rates were 72% (317/438) for GIM and 59% (236/401) for FM. Depending on the vaccine, 2–23% of GIM and 5–30% of FM reported adult patients defer or refuse vaccines due to financial reasons the majority of the time. The vaccines most commonly deferred or refused by patients for financial reasons were zoster and human papillomavirus vaccine for both specialties. Physicians in many instances reported they do not recommend vaccines when they thought a patient could not afford it (GIM 4% ‘frequently’, 16% ‘sometimes;’ FM 5% ‘frequently’, 20% ‘sometimes’), when they thought a patient’s insurance would not cover the vaccine (GIM 9% ‘frequently’, 24% ‘sometimes’; FM 12% ‘frequently’, 24% ‘sometimes’) and when they thought a patient could receive a vaccine more affordably elsewhere (GIM 9% ‘frequently’ 30% ‘sometimes’; FM 11% ‘frequently’, 26% ‘sometimes’). The table shows physicians’ knowledge regarding Medicare coverage for vaccines. Over a third of physicians were unaware of provisions in the Affordable Care Act that promote vaccinations (GIM-39%, FM-34%) and one-fifth were only ‘a little aware’ (GIM-20%, FM 17%) of these provisions.

CONCLUSIONS: Patients are refusing and physicians are not recommending adult vaccinations for financial reasons. Increased knowledge of Medicare coverage would better enable physicians to direct older patients to the best venue to receive vaccines, particularly if their practice does not bill Medicare Part D. Heightened awareness of the favorable financial implications of the Affordable Care Act on vaccinations might result in physicians being more likely to recommend them.

FIRST YEAR EXPERIENCE OF HCV SCREENING OF 2,765 BABY BOOMERS ADMITTED TO A SAFETY NET HOSPITAL IN SOUTH TEXASBarbara J. Turner^2,1; Barbara S. Taylor^1,2; Joshua T. Hanson^1,2; Mary E. Perez²; Ludivina Hernandez². ¹University of Texas Health Science Center at San Antonio, San Antonio, TX; ²University of Texas Health Science Center at San Antonio, San Antonio, TX. (Tracking ID #1935853)

BACKGROUND: Baby boomers account for over 75% of the 3 to 5 million HCV-infected persons in the U.S., and 75% of these HCV-infected persons are undiagnosed. The Centers for Disease Control and Prevention and the U.S. Preventive Services Task Force now both endorse one-time testing of all baby boomers (born 1945 to 1965) for hepatitis C virus (HCV) infection. Objective: To implement a baby boomer HCV screening and linkage to care program in the largest safety net hospital in South Texas.

METHODS: The program has 4 key features: 1) Education in English and Spanish about HCV screening of baby boomers using posters throughout the hospital and flyers in admission packets that offer an opt out option; 2) Electronic medical record algorithm that identifies eligible baby boomers and, for most, places an order for HCV antibody (HCV Ab) with a reflex HCV RNA test for positive HCV Ab [exclusion criteria: record of prior HCV test or diagnosis, poor prognosis (e.g. metastatic cancer), psychiatric admission]; 3) Personalized counseling for HCV Ab+ patients by a bilingual inpatient counselor who reviews a mobile app-based HCV educational program in English and Spanish that was developed by our team; and 4) Navigation by a bilingual community health worker to facilitate follow-up primary and HCV specialty care after hospital discharge by addressing barriers to care, poor health literacy, substance abuse, and lack of social support.

RESULTS: From 12/1/2012 to 11/30/2013, 5,572 unique baby boomers were hospitalized of whom 45% were women, 59% Hispanic, median birth year 1956, and 46% uninsured. Of these, over 1,400 patients had a prior HCV test or diagnosis and were excluded. Of 2,765 unique baby boomers screened for HCV infection, 230 (8.3%) tested HCV Ab+. HCV Ab+ was more likely for men (75% vs. 55% of HCV Ab negative, p < .001) but did not differ by birth year, Hispanic ethnicity, or insurance status. Follow-up HCV RNA testing was performed for 205 (89%) of HCV Ab+ patients. 125 (61%) were HCV RNA+ while 80 (39%) were HCV RNA negative. HCV RNA positive and negative patients did not differ by birth year, gender, Hispanic ethnicity, or insurance status. Of the 125 patients diagnosed with chronic HCV infection, 110 (88%) have been counseled, 100 (80%) have had follow-up ambulatory care addressing HCV and 43 (34%) have received HCV specialty care.

CONCLUSIONS: To our knowledge, this is the first baby boomer HCV screening program that has been broadly implemented in a civilian, safety-net hospital. The 8.3% HCV Ab+ prevalence in persons without evidence of prior HCV testing or diagnosis supports the public health value of screening in similar inpatient settings. This program has newly identified 125 patients with chronic HCV infection and linked the majority to follow-up ambulatory care. The high rate of apparent clearance of HCV (39%) needs to be confirmed by repeat HCV RNA testing after 3–6months. If confirmed, cost-effectiveness models may need to be adjusted to account for a lower proportion of HCV Ab+ baby boomers with persistent infection.

FOLLOW-UP OF ABNORMAL CT RESULTSLinda J. Canty¹; Paul Visintainer¹; Mihaela S. Stefan¹; Philip Henneman¹; Sean Stanhope¹; Reva Kleppel¹; Annie Lagoy¹; Jessilia Santiago¹; Michael B. Rothberg². ¹Tufts University School of Medicine / Baystate Medical Center, Springfield, MA; ²Cleveland Clinic, Cleveland, OH. (Tracking ID #1939790)

BACKGROUND: CT scans often include abnormal findings, and follow-up is frequently recommended. However, when a CT scan ordered in the emergency department (ED) or the inpatient unit contains an abnormal finding, it may not always be followed up by the primary care provider. Mechanisms in place to ensure the appropriate follow-up of tests may miss these types of handoffs. The objective of this study was to quantify the proportion of abnormal findings with follow-up and to assess factors associated with timely follow-up.

METHODS: We conducted a retrospective cohort study to assess the proportion of abnormal findings that had documented follow-up. The CT scans were performed at three hospitals within Baystate Health, a large health system in western Massachusetts. For the two community hospitals, we looked at all scans performed in 2010 on patients who also had a visit during 2010 to the primary care office associated with the hospital. For the academic medical center, we used a sample of two-thirds of the scans from January to mid-July 2010 performed on patients who had a visit during 2010 to one of six nearby primary care practices. All providers in the system have access to radiology readings through a shared electronic health record, but only for scans ordered in outpatient settings does the ordering provider receive the scan results in their electronic medical record inbox. CT scans involved the chest (including angiograms,) abdomen, pelvis or a combination of these. An abnormal result was defined as any finding for which the radiologist’s dictated final report recommended follow-up using wordings such as “recommend,” “suggest,” or “should be” followed, or cited guidelines outlining a timeframe for follow-up. For each of these findings, we used the electronic health record to assess whether the patient received appropriate follow-up, defined as either adherence to the radiologist’s suggestion or documentation that the finding was discussed and a conscious decision was made not to pursue the finding further. We compared the proportion of findings that received follow-up if they were ordered by outpatient providers to those ordered in the ED and inpatient wards.

RESULTS: We reviewed 1922 scans performed on 1524 patients. Overall, there was documentation of follow-up within the recommended time period in 176 out of 236 findings (75%.) Of these, documentation of follow-up was seen in 41 of 64 (64%) ED findings, 37 of 52 (71%) inpatient findings, and 98 of 120 (82%) outpatient findings. Follow-up was more common for findings on CT scans ordered by outpatient providers than for those ordered in the ED (p = 0.01.) Findings on inpatients were not statistically more likely to be followed when compared to either the ED or the outpatient setting. Radiologists stating in the dictated report that the results were “communicated” was not associated with greater follow-up (44 of 56 findings; 79%) compared to when this was not in the report (132 of 180 findings; 73%; p = 0.49.)

CONCLUSIONS: Abnormal findings on CT scans ordered in the outpatient setting were more likely to have documentation of follow-up than those ordered through the emergency department. For patients scanned in the ED, more than 1/3 of abnormal findings went without evidence of timely follow-up. Improved communication systems appear needed for radiological handoffs between the ED and the outpatient setting.

FOLLOW-UP PRIMARY CARE VISITS IN A RESIDENT PRACTICE AFTER HOSPITAL DISCHARGE - HOW ARE WE DOING?Sanjai Sinha¹; Thomas Uttaro³; Daniel M. Stein²; Amanda S. Carmel¹. ¹Weill Cornell Medical College, New York, NY; ²Weill Cornell Medical College, New York, NY; ³Weill Cornell Medical College, New York, NY. (Tracking ID #1935099)

BACKGROUND: Patients cared for in internal medicine teaching clinics often have poor primary care follow up after hospital discharge. Such discontinuity can result in medical errors, suboptimal care, and re-hospitalization. In a Medicare study, more than half of patients who were re-admitted within 30days of hospital discharge had not had an outpatient encounter between admissions. Based on a 6-week period of inpatient and emergency department (ED) discharges from New York Presbyterian/Weill Cornell Medical Center (NYP/WCMC), we aimed to assess the percentage of Weill Cornell Internal Medicine Associates (WCIMA) patients that completed a post discharge appointment at WCIMA within 30days of discharge. Secondarily, we examined the percentage of patients who utilized the hospital or ER more than once in the study period, and who saw their PCP for follow up.

METHODS: WCIMA is a large internal medicine practice affiliated with Weill Cornell Medical College in New York City. It is the main practice site for 33 faculty physicians and 136 internal medicine residents. Through data extraction of our electronic health records, we identified all WCIMA discharges from NYP/WCMC from 9/1/2013 to 10/12/2013. Information for each discharge including basic patient identifiers, age, payer, hospital department/ service, assigned primary care provider, completed visit date and provider were obtained. Interval between discharge and the follow up visit were calculated. The unit of analysis was a hospital discharge event, some patients had multiple hospitalizations and ED/urgent care visits. IBM SPSS software version 22 procedures for data manipulation and descriptive statistics were used for data analysis. IRB approval is pending.

RESULTS: We identified 1,908 unique discharges for 1,513 WCIMA patients from NYP/WCMC during the study period. Two hundred ninety-three patients (19.4%) had more than one discharge. In WCIMA patients discharged from a Medicine admission, 51.2% (N = 259/506) had a post discharge visit at WCIMA, 37.2% (N = 188/506) had the visit within 30days of hospital discharge, and 37.7% (N = 190/506) saw their assigned PCP. In WCIMA patients discharged from the ED/Urgent care, 35.8% (N = 375/1,047) had a post discharge visit at WCIMA, 24.1% (N = 242/1,047) had the visit within 30days of discharge, and 19.2% (N = 201/1,047) saw their PCP. The mean follow up interval was 23.01days (SD 22.360) after a Medicine discharge and 25.40days (SD 23.619) after ED/Urgent care. The mean age of all discharged patients was 52.62years (SD 18.468). Patients who completed post-discharge appointments were significantly older than those who did not (57.2 vs. 49.3years, P < 0.001), however, this difference was lost when examining Medicine patients alone.

CONCLUSIONS: Only half of WCIMA patients discharged from Medicine during the study period, and fewer from the ED/Urgent care had a WCIMA follow up visit, indicating significant missed opportunities to coordinate care from hospital to home. Fewer than half had visits within 30days or saw their assigned PCP. Overall, older patients were more likely to follow up. Other resident teaching practices likely have similar challenges in hospital-outpatient continuity. We have designed an intervention using a transitions registrar and care manager to ensure timely and appropriate post discharge care. We hope the intervention will identify one strategy to improve follow up rates and clinical outcomes among patients of a large internal medicine teaching clinic.

FOOD INSECURITY AND COST-RELATED MEDICATION UNDERUSE ARE INDEPENDENT BARRIERS TO DIABETES CONTROL: RESULTS FROM THE MEASURING ECONOMIC INSECURITY IN DIABETES (MEND) STUDYSeth A. Berkowitz¹; James B. Meigs¹; Steven J. Atlas¹; Darren A. DeWalt⁴; Hilary K. Seligman³; Deborah J. Wexler². ¹MGH, Boston, MA; ²MGH, Boston, MA; ³UCSF, San Francisco, CA; ⁴UNC, Chapel Hill, NC. (Tracking ID #1934690)

BACKGROUND: Food insecurity, cost-related medication underuse (CRMU), and housing instability may be significant, yet modifiable, barriers to diabetes management. However, whether these barriers are independently associated with poor diabetes control, accounting for other aspects of low socioeconomic status and comorbidity, is unclear. To inform future population management interventions, we tested the hypothesis that they are independently associated with poor diabetes control.

METHODS: We contacted a stratified random sample of adult (age >20years) diabetes patients in 4 clinic sites (2 community health centers, 1 academic internal medicine practice, and a diabetes center) from June 15-Sept 15th 2013. Participants completed validated instruments assessing food insecurity, CRMU, and housing instability, in English or Spanish. We also collected information on other social circumstances, including educational attainment, limited English proficiency (LEP), health literacy, and nativity, along with age, gender, race/ethnicity, insurance, diabetes duration, Charlson comorbidity score, diabetes medications, Hemoglobin A1c (HbA1c) tests/year and number of outpatient visits. The primary outcome was poor diabetes control (most recent HbA1c > 9.0% or LDL cholesterol >100mg/dL). We estimated prevalence using inverse probability weighting. We performed multivariable logistic regression analysis with generalized estimating equations to account for clustering by clinic, and used pseudo-R² statistics to evaluate explained variation in diabetes control.

RESULTS: Overall, 412 patients were included (response rate: 62%). Of these, 19% reported food insecurity, 28% CRMU, and 11% housing instability. Patients reporting food insecurity were more likely to be younger (mean age 56 vs. 63years, p = .01), non-white (35% vs 22%, p = .01), and have Medicaid (22% vs. 11%, p = .03). In unadjusted analyses, the prevalence of poor diabetes control was higher in those with, vs. without, food insecurity (53% vs. 28%, p = .01), CRMU (53% vs. 22%, p = .01) and housing instability (56% vs. 30%, p = .01). Those with, vs. without, food insecurity, CRMU, or housing instability had similar comorbidity and HbA1c tests/year (an indicator of engagement with care), but more outpatient visits/year (Table 1). In a multivariable regression analysis, adjusted for the covariates listed above, food insecurity and CRMU were associated with worse diabetes control (Table 2). Using Pseudo-R² statistics, our model explained 28% of the variation in diabetes control; food insecurity, CRMU, and housing accounted for 4% of total variation. By comparison, education, health literacy, LEP, and nativity (other indicators of social disadvantage) together explained only 2%.

CONCLUSIONS: Food insecurity and CRMU are common, and independently identify patients at increased risk of poor diabetes control despite similar comorbidity and engagement with care. Because these patients already have higher healthcare utilization, interventions targeting food insecurity and cost-related medication underuse may have greater utility than increasing usual care.

Table 1

Table 2

FOOD INSECURITY IS ASSOCIATED WITH DIABETES SELF-CARE BEHAVIORS AND GLYCEMIC CONTROLWilliam J. Heerman¹; Ken Wallston²; Chandra Y. Osborn³; David Schlundt⁴; Shari D. Barto¹; Russell L. Rothman¹. ¹Vanderbilt University Medical Center, Nashville, TN; ²Vanderbilt University Medical Center, Nashville, TN; ³Vanderbilt University Medical Center, Nashville, TN; ⁴Vanderbilt University, Nashville, TN. (Tracking ID #1928444)

BACKGROUND: Food insecurity is the “limited or uncertain availability of nutritionally adequate and safe foods,” and has been associated with worse glycemic control among patients with type 2 diabetes (T2DM). Less is known about food insecurity’s relationship with diabetes self-care behaviors, and whether its effects on glycemic control can be attributed to that relationship. Therefore, we examined the relationships between food insecurity, self-care, and glycemic control, and whether any effect on glycemic control was due to relationships with self-care.

METHODS: We conducted a cross-sectional analysis of baseline data from patients with T2DM enrolled in a randomized trial evaluating a health literacy-focused diabetes intervention at 10 county health departments across middle Tennessee. Food insecurity was assessed with 3 items from the U.S. Household Food Security Survey and patients who answered “sometimes” or “often” to at least one of these items were considered “food insecure.” Diabetes self-care behaviors were assessed with the Summary of Diabetes Self Care Activities Scale (SDSCA), Personal Diabetes Questionnaire (PDQ-11), and Adherence to Refills and Medication Scale (ARMS). Glycemic control was assessed with A1C, collected prospectively unless available in medical record within 4weeks of study enrollment. Student’s t-tests or Wilcoxon rank-sum tests were used to evaluate the bivariate relationships between food insecurity and each self-care behavior, food insecurity and A1C, and each self-care behavior and A1C. Significant bivariate relationships were re-examined using linear regression models adjusted for gender, ethnicity, education, and income. Mediational analyses were performed as appropriate.

RESULTS: The sample consisted of 401 participants, of whom 61% were female, 57% self-identified as non-Hispanic White, 17% as non-Hispanic Black, 24% as Hispanic, and 2% as non-Hispanic other. The majority (88%) were uninsured, and 37% had less than a high school education. Nearly three-quarters (73%) reported some level of food insecurity. Food insecurity was significantly associated with self-care behaviors (Table) including less adherence to a general diet (p = 0.03), routine physical activity (p = 0.05), and medications (p = 0.01), and a greater occurrence of poor eating (p = 0.02) and skipping meals (p = 0.03). Food insecurity was also associated with worse glycemic control (adjusted β co-efficient 0.6, p = 0.015). The median A1C for food insecure participants was 9.8% (IQR 8.2, 11) compared to 9.3% for food secure participants (IQR 7.7, 10.6). There was no relationship between any of the self-care behaviors assessed and A1C, so tests for mediation were not performed.

CONCLUSIONS: There was a high rate of food insecurity in this sample of diabetes patients with low socioeconomic status. Food insecurity was associated with less adherence to recommended self-care behaviors, a greater occurrence of undesirable behaviors, and worse glycemic control. Further work is needed to understand the mechanisms by which food insecurity impacts self-care, including behaviors not directly related to food (e.g., physical activity and medication adherence) as well as the mechanisms by which it also impacts glycemic control. While our measures of self-care were not associated with A1C, limiting our ability to explore self-care as a mechanism, other more sensitive measures of behavior may hold promise. Nonetheless, our findings contribute to a limited literature on the relationship between food insecurity and diabetes behaviors.

Food Insecurity and Diabetes Self-Care Behaviors

t-test with unequal variance

FOOD INSECURITY, COPING STRATEGIES, AND GLYCEMIC CONTROL IN LOW-INCOME PATIENTS WITH DIABETESVictoria L. Mayer¹; Kevin H. McDonough¹; Hilary K. Seligman²; Judith A. Long^1,3. ¹Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA; ²University of California San Francisco, San Francisco, CA; ³Philadelphia Veterans Affairs Medical Center, Philadelphia, PA. (Tracking ID #1933037)

BACKGROUND: In the U.S., people who are food insecure (lack sufficient food for an active and healthy life) have a higher prevalence of diabetes and worse diabetes control. However, the mechanisms underlying these associations are not well understood. In this study we evaluate the relationship between food insecurity and coping strategies hypothesized to worsen glucose control in a population of low-income diabetics.

METHODS: We performed a cross-sectional telephone survey among low-income adults with diabetes seen at a large, urban medical center. We interviewed people within 8weeks of having a hemoglobin A1c (HbA1c) evaluation and characterized individuals as having poor glycemic control (HbA1c ≥ 8.0) or adequate glycemic control (HbA1c < 8.0). The main independent variable was a dichotomous measure of food insecurity (18-item U.S. Department of Agriculture Household Food Security Survey Module). We assessed the use of coping strategies including a diet low in fruits and vegetables and high in added sugars (using the California Health Interview Survey Dietary Screener), foregone medications, foregone medical care, use of food assistance programs (including the Supplemental Nutrition Assistance Program - SNAP) and emergency food programs, and overeating in times of adequacy. Additional independent variables included socio-demographic and clinical covariates. In bivariate analyses, we compared glucose control and use of coping strategies in the food insecure and food secure groups. We developed logistic regression models to evaluate the association between food insecurity and glucose control, with adjustment for covariates and coping strategies. Finally, we assessed the role of interactions between food insecurity and coping strategies.

RESULTS: Of 413 respondents (response rate 51%), 40.4% were food insecure. The food insecure group had a higher proportion of patients with poor glycemic control (68.3% vs. 53.3%, p = 0.002). There were no significant differences between the food insecure and food secure groups in daily intake of fruits, vegetables, or added sugars. A significantly higher percentage of patients in the food insecure group reported foregone medications, foregone medical care, participation in food assistance programs and use of emergency food programs. Food insecure patients were also more likely to report overeating at times when food is adequate. However, these coping strategies were not significantly associated with glucose control. In the adjusted model without interaction terms, when compared to food secure individuals, food insecure individuals were more likely to have poor glucose control (OR 2.83, 95% CI 1.47–5.44). However, there was a significant interaction between food insecurity and receipt of SNAP: food insecure individuals receiving SNAP benefits were less likely to have poor glucose control (OR 0.28, 95% CI 0.09–0.90), while food insecure individuals not receiving SNAP were more likely to have poor glucose control (OR 7.41, 95% CI 2.41–22.77).

CONCLUSIONS: Food insecurity was associated with higher risk of poor glucose control. While more prevalent among patients who were food insecure, many coping strategies previously thought to underlie this relationship were not significantly associated with poor glucose control. However, receipt of food assistance among food insecure individuals was associated with better diabetes control, suggesting that in addition to helping individuals afford food, such programs may also play a role in improving health.

FORGET ME NOT: A RANDOMIZED TRIAL OF THE DURABILITY OF HOSPITAL-BASED EDUCATION ON INHALERS FOR PATIENTS WITH COPD OR ASTHMAValerie G. Press¹; Vineet Arora²; Kristin L. Constantine³; Edward T. Naureckas⁴; Steven R. White⁴; Jerry Krishnan⁵. ¹University of Chicago, Chicago, IL; ²University of Chicago, Chicago, IL; ³University of Michigan, Ann Arbor, MI; ⁴University of Chicago, Chicago, IL; ⁵University of Illinois at Chicago, Chicago, IL. (Tracking ID #1934140)

BACKGROUND: As part of a nationwide strategy to reduce readmissions for chronic diseases, hospitals are increasingly using the inpatient setting as a “teachable moment” to train patients in important self-management techniques to improve both patient-centered outcomes as well as to avoid preventable re-hospitalizations. Inpatient education is particularly salient for patients with COPD, which is the 3rd leading cause of readmission among Medicare patients, who are often not aware how to use their preventive and rescue inhaler medications correctly. While hospital-based education is not new, it is unclear how durable it is and when reinforcement may be needed. The objective of this two-center randomized clinical trial was to test the short-term effectiveness and longer-term retention of a guideline-based, patient-centered approach, called teach-to-goal (TTG) compared to a brief intervention (BI) of simple verbal instructions to teach inhaler technique to inpatients with COPD or asthma.

METHODS: Patients were randomized to TTG or BI, stratified by level of health literacy (high/low/unknown [due to insufficient vision]). Use of metered-dose inhaler (MDI) and Diskus® devices was assessed using step-by-step checklists. Misuse was defined as <75% of steps correct. TTG participants received cycles of inhaler demonstration followed by re-assessment of technique (“teach-back”). BI participants received verbal instructions without demonstration. Participants returned at 30-days and 3months post-hospital discharge. Significance was tested using Chi square and Fishers exact tests.

RESULTS: Between September 2011–October 2012, 120 participants were enrolled from two urban hospitals and were randomized to either the TTG (n = 62) or BI (n = 58) interventions. The majority were female (73%) and African-American (90%), with a mean age of 47years. Two-thirds (66%) had been hospitalized at least once in the last year and nearly half (49%) had a near-fatal event (ICU admission and/or intubation) in their lifetime. Fifteen percent had insufficient vision, 19% had low health literacy. Eighty-nine percent of participants returned at 30-days and 86% returned at 3months. There were no baseline differences between groups (TTG vs. BI) in the baseline prevalence of misuse for either MDI or Diskus. Immediately after education, misuse was less common in TTG vs. BI for MDI (11 vs. 60%, p < 0.001) and Diskus devices (5 vs. 61%, p = 0.001). At 30-days there were no significant differences in misuse for either MDI (53 vs. 70%, p = 0.09) or Diskus (53 vs. 59%, p = 0.73). Despite the lack of significant 30-day findings, MDI misuse was significantly less likely at 3months in the TTG vs. BI group (48 vs. 76%, p = 0.003). A similar result at 3months for the Diskus device was not seen (p = 0.2). Stratifying the results by level of health literacy did not produce any clear patterns [Table]. There were no differences between TTG and BI for any health literacy group at 30days; those with adequate health literacy were less-likely to misuse MDI at 3months in TTG vs. BI group (p = 0.02). Participants were less-likely to misuse Diskus if they received TTG vs. BI for those with adequate health literacy (p = 0.006) or less-than-adequate health literacy (p = 0.048), but not for those with insufficient vision (p > 0.99). There were no differences between TTG and BI for any health literacy group at 30-days or 3months for the Diskus device.

CONCLUSIONS: Although patient-centered TTG education can improve self-management skills among inpatients with asthma or COPD, the effects of hospital-based education wane over time. This suggests that in order to improve patient-reported outcomes and potentially decrease re-admissions, post-discharge educational reinforcement is likely necessary for long-term mastery of chronic disease self-management. Further, data are needed about the effectiveness of hospital-based education on subsequent clinical outcomes for hospitalized patients with COPD or asthma in the short-term and long-term after hospital discharge.

Proportion of Misuse

TTG: Teach To Goal; BI: Brief Intervention

FROM FLIPPING THE CLASSROOM TO FLIPPING THE HOMEWORK: A LEARNER-CENTERED APPROACH TO ADVANCE PREPARATION ASSIGNMENTSPaul Haidet; Klara Papp; Daniel R. Wolpaw. Penn State College of Medicine, Hershey, PA. (Tracking ID #1939412)

BACKGROUND: The commonly used convention for preparing learners to participate in active classroom learning is to assign them pre-readings or similar materials. We have found that such published resources do not always align well with educational objectives. Further, pre-reading assignments often present information according to the teacher’s organizational schema, rather than allowing students to identify learning needs and incorporate material according to their own emerging conceptual frameworks. We sought to develop a method for advance preparation that addresses these issues.

METHODS: We designed a 2-hour session on “introduction to educational concepts” for junior faculty enrolled in a 12-month faculty development fellowship. The in-class portion used the Team-Based Learning method to foster active learner participation. However, rather than assign specific readings for the pre-session preparation, we gave learners a list of 27 educational terms (e.g., ‘Bloom’s Taxonomy’, etc.) and instructed them to research unfamiliar terms and reflect on their own assumptions and beliefs regarding teaching. We performed internet searches using the terms to review information that our learners might access. We distributed a post-session survey containing three open-ended questions that addressed how the assignment: 1) affected learning about educational concepts, 2) differed from typical reading assignments, and 3) influenced levels of enjoyment during completion. We performed a thematic analysis of learner responses to the open-ended questions.

RESULTS: Nineteen learners (33% women) participated in the session and returned surveys (100% response). Forty-eight percent of learners were clinically focused in their scholarly work, 28% educationally focused, and 24% research focused. For our session, several themes emerged from the responses about the advance preparation assignment. First, some learners struggled with uncertainty and a need to find the “right” answer, particularly when they found conflicting information. A few learners viewed uncertainty as an opportunity to deepen understanding, because it forced them to critically appraise information, identifying knowledge gaps during the process. Second, learners generally perceived more active engagement as compared to typical reading assignments; however, this effect was tempered for learners who focused on collecting definitions at the expense of understanding concepts. Finally, the majority of learners expressed enjoyment with the assignment and appreciated the ability to allocate preparation time according to their own needs.

CONCLUSIONS: Using a learner-directed approach for advance preparation merits further exploration. Learner experiences suggested more active, critical thinking, and may represent more efficient and deeper learning of course concepts, and better preparation for subsequent learning in the classroom. Future work should more rigorously test these outcomes, and explore additional methods of fostering learner-centeredness in advance preparation assignments.

FULFILLING UNMET NEEDS: INTEGRATING WOMEN’S HEALTH INTO THE INTERNAL MEDICINE RESIDENCY PROGRAM CURRICULUMSneha Shrivastava; Abby Spencer. Allegheny General Hospital, Pittsburgh, PA. (Tracking ID #1941879)

BACKGROUND: National guidelines and the American Board of Internal Medicine (ABIM) have highlighted key areas of women’s health that are critical to the training of an internal medicine resident. The federated council for Internal Medicine (FCIM), who sets the core knowledge curriculum for IM residency programs, argues that all general internists should be competent in conditions that are “unique to, more common in or more serious in women or that have manifestations, risk factors, or interventions that are different in women”. Our objective is to improve the women’s health curriculum in a large academic community Internal Medicine residency program.

METHODS: A 147-item anonymous online survey was sent to PGY1&2 residents to assess their comfort, perceived knowledge, perceived importance, and exposure to 28 Women’s Health topics identified as critical to the training of an internist. Topics were selected based on a thorough literature review, curricular recommendations by ABIM, FCIM, and the NIH Women’s Health Congress. We used descriptive statistics to determine and rank comfort, exposure, perceived knowledge, and perceived importance of the topics. We then analyzed the In-Training exam results from the last 3years and compared the performance of our residents to national data on topics related to women’s health. Based on our survey results and ITE review, we designed a new longitudinal interactive women’s health curriculum to meet critical learning needs of our residents. Curricular methods included problem-based learning (PBL) workshops on osteoporosis, multi-disciplinary case conferences on depression and anxiety during pregnancy traditional didactics, and STD jeopardy. Our interprofessional panel experts during case conferences included faculty from internal medicine, psychiatry, and obstetrics-gynecology Pre-post surveys were conducted after problem-based learning (PBL) and multi-disciplinarycase conferences. The Wilcoxon signed-rank test was used to compare resident knowledge and comfort before and after case conferences.

RESULTS: Response rate on the initial survey was 63% (38/60). Non-responders were similar to responders in gender and pgy-year. 13/28 topics were identified as having the highest perceived importance to our IM residents. Of these, residents felt they had insufficient clinical exposure in prescribing contraception, evaluating breast symptoms, managing menopause, performing pap smears and pelvic exams, and diagnosing or managing polycystic ovarian syndrome (PCOS). Lower clinical exposure in these areas correlated significantly with lower comfort and lower perceived knowledge. Osteoporosis was identified by residents as their single greatest learning need; this topic was rated as highly important and residents had low knowledge and comfort with it. While residents reported a low perceived knowledge on several women’s health topics, their performance on the In-Service exam was on par with that of the national average. . After an interactive PBL conference on osteoporosis, 95.1% of the residents felt knowledgeable and comfortable in this topic as compared to 37,9%% who were knowledgeable and comfortable before the case discussion (p = 0.0003). After the multi-disciplinary case discussion on anxiety and depression during pregnancy, 80.1% of the residents felt more comfortable in this topic as compared to only 36.4% who were comfortable before the case discussion (p = 0.0009). Additionally, 94.7% of the residents preferred PBL conference and 88.9% of the residents preferred multi-disciplinary conference to a standard noon conference lecture. All the residents (100%) agreed that PBL cases and multi-disciplinary conferences are extremely effective methods of learning.

CONCLUSIONS: Our study highlighted significant gaps in our women’s health curriculum. Additionally, there is a concerning perception among residents that many topics critical to the health of women and the education of IM residents are not very important. Based on our findings, we are redesigning our didactic and experiential women’s health curriculum to augment knowledge, comfort, importance of, and exposure to areas in need. Problem based learning and multi-disciplinary case discussions are effective educational tools to achieve this. Enhancing education and increasing clinical exposure to fundamental women’s health issues will produce better residents and promote higher quality care for women.

GENDER AND RACIAL DIFFERENCES IN CONTROL OF CARDIOVASCULAR DISEASE RISK FACTORS AMONG VETERANSKaren M. Goldstein^1,3; S. Dee Melnyk^1,4; Leah L. Zullig^1,5; Karen Stechuchak¹; Eugene Z. Oddone^1,3; Lori A. Bastian^6,7; Susan Rakley^2,3; Maren Olsen^1,8; Hayden Bosworth^1,9. ¹Durham Veterans Affairs Medical Center, Durham, NC; ²Durham Veterans Affairs Medical Center, Durham, NC; ³Duke University, Durham, NC; ⁴University of North Carolina, Chapel Hill, NC; ⁵University of North Carolina, Chapel Hill, NC; ⁶University of Connecticut Health Center, Farmington, CT; ⁷VA Connecticut Healthcare System, West Haven, CT; ⁸Duke University, Durham, NC; ⁹Duke University, Durham, NC. (Tracking ID #1934589)

BACKGROUND: Female Veterans are the fastest growing Veterans Administration (VA) patient population, nearly doubling in the past decade. Compared with non-Veteran females, Veteran women have higher rates of cardiovascular disease (CVD) risk factors, such as smoking and obesity, which put them at significant risk for long-term cardiovascular morbidity and mortality. Recent findings suggest that black female Veterans are more likely to have a diagnosis of hypertension, diabetes, and obesity compared with white female Veterans. However, there is little information about CVD risk factor control between male and female Veterans, particularly by race. We sought to address the following: 1) Is there a significant difference in CVD risk factor control between men and women Veterans at high risk for CVD, defined as poor control of blood pressure (mmHg), hemoglobin A1c, and LDL cholesterol (mg/dL)?; and 2) Is there a significant difference in CVD risk factor control between African-American and White Veterans by gender, defined as blood pressure (mmHg), hemoglobin A1c, and LDL cholesterol (mg/dL)?

METHODS: We used analysis of variance, adjusting for age, to compare gender and racial differences in three risk factors that predispose to CVD (i.e. diabetes, hypertension, and hyperlipidemia) in a cohort of Veterans who were eligible for enrollment in a clinical trial. To be eligible for the trial, individuals had to be enrolled at the Durham VA Medical Center and have a diagnosis of hypertension, hyperlipidemia, or both. Means provided were estimated from our ANOVA model that adjusted for age. Our sample included 23,955 men (33.9% black and 54.0% white) and 1,010 women (54.8% black and 39.3% white).

RESULTS: LDL values were higher in female than male Veterans with hyperlipidemia with age-adjusted estimated mean values of 111.7mg/dl vs 97.6mg/dl (p < 0.01). There was no difference in LDL values between black and white women. There was a small but significant difference between black and white male Veterans with hyperlipidemia, specifically 99.6mg/dl vs 95.8mg/dl respectively (p < 0.01). Systolic blood pressures were similar between male and female Veterans with hypertension, though there was a significant difference in diastolic blood pressures of 79.0mmHg vs 76.8mmHg (p < 0.01) respectively. Blood pressures were higher among black than white female Veterans with hypertension with age-adjusted estimated mean systolic blood pressures of 136.3mmHg versus 133.5mmHg, respectively (p < 0.01), and diastolic blood pressures of 82.4mmHg versus 78.9mmHg (p < 0.01); a similar pattern between black and white male Veterans was also seen. There was no difference in hemoglobin A1c between male and female Veterans with diabetes. Age-adjusted estimated mean hemoglobin A1cs for African-American and White men were 7.5 and 7.3 (p < 0.01), respectively. Age-adjusted estimated mean hemoglobin A1cs for African-American and White women were 7.6 and 7.1 respectively, though this was not significant.

CONCLUSIONS: Among this Veteran population at high-risk for CVD, we found that female Veterans had worse control of hyperlipidemia than male Veterans. This difference was observed despite the fact that female Veterans in this cohort on average were almost 10years younger than male Veterans. In addition, we found that African-American women and men had worse blood pressure control than white women. Further examination of CVD gender and racial disparities in this population may help to develop targeted treatments and strategies applicable to the general population.

GENERAL MEDI