Abstract Session D1: Lipkin Award Finalist Session Moderators: - - PDF document

Abstract Session D1: Lipkin Award Finalist Session Moderators: Eugene C. Rich, MD and William M. Tierney, MD

STATIN PRESCRIBING FOR PRIMARY PREVENTION: WHAT MIGHT THE 2013 ACC/AHA GUIDELINES ACHIEVE? Harb Harb1; Michael B. Rothberg1; Brian Wells2; Bo Hu2. 1Cleveland Clinic, Cleveland, OH; 2Cleveland Clinic, Cleveland, OH. (Tracking ID #1937737) BACKGROUND: The Adult Treatment Panel III guidelines for cholesterol testing and treatment in primary prevention were based

• n cardiovascular risk factors and LDL cholesterol targets. In contrast, the new American College of Cardiology/American Heart

Association (ACC/AHA) 2013 guideline focuses exclusively on cardiovascular risk, with statins recommended for all patients with a 10-year risk of >7.5% using a new population-based risk calculator. One concern is that the ACC/AHA guidelines may lead to initiation of statin therapy for more patients, including many who would have been considered to be lower risk by the earlier criteria. In contrast, the previous focus on LDL may have led to overprescribing for patients with high LDLs and low risk, or underprescribing for those with low LDLs but high risk. The objective of this study is to quantify the potential change in statin prescribing patterns using the ACC/AHA guidelines compared to current practice for patients without CHD equivalents. METHODS: We conducted a retrospective cross sectional study of patients aged between 30 to 75 years old who received a first

• utpatient primary care visit at the Cleveland Clinic between January 1, 2005 and December 31, 2012. Patients with incomplete data,

contraindications to lipid lowering medications (history of statin induced rhabdomyolysis, myositis or myopathy, transaminitis, statin allergy); lipid-lowering agent(s) prescribed at their first encounter (implying that they were already taking lipid lowering therapy); and patients with CHD or CHD equivalents (CVD, PVD, Diabetes) were excluded. For each patient we collected the following information from the time of their first lipid panel: age, sex, and race; as well as total, HDL and LDL cholesterol, smoking status, systolic blood pressure and whether they were treated with an antihypertensive medication. Patients were then stratified according to 10-year cardiovascular risk into 3 groups—low (<5%), intermediate (5-15%) and high (15%)—based on the Framingham equation. We then compared the number of patients in each stratum who actually received statins to the number in that same stratum that would be recommended to receive statins based on the ACC/AHA guidelines (i.e. those with a 10-year risk of >7.5% based on the new risk calculator). We then calculated the number of cardiac events prevented over 10 years under current practice to those that could have been prevented by following the ACC/AHA guidelines using the formula: E=N * R * RRR where E is the number of events prevented, N is the number of patients taking statins, R is the average Framingham risk of the patients taking statins, and RRR is the relative risk reduction related to statin use. We estimated the risk reduction to be 27%, based on a Cochrane Review of statins for primary prevention. RESULTS: Of 98,136 patients who had an initial visit (with LDL measured) during the study period, 85,079 (87%) patients met inclusion criteria. Of these, 41,376 (49%) were low risk, 30,135 (36%) intermediate risk, and 12,768 (15%) high risk based on their Framingham risk scores. Mean patient age was 48 years, 42% were male, 79% were white, 37% were smokers and 27% were taking medication for hypertension. The comparison of statin use appears in the Table. The total number of patients treated with statins under current practice was 13,896; 31% of these were high risk and 22% were low risk. Under the ACC/AHA guidelines, 17,294 patients would qualify for treatment; 70% would be high risk and 1.4% low risk. Assuming that statins reduce the risk of a major cardiac event by 27%, then the total number of events currently being prevented is 510 per 10 years, with NNT= 27. Under ACC/AHA criteria the number of events prevented would be 1034, with NNT=17. CONCLUSIONS: Compared to current practice, following the ACC/AHA guidelines would increase the total number of patients on

• statins. However, it would decrease use among low and intermediate risk patients, while markedly increasing use among high risk

patients, thereby decreasing the total number of major cardiac events as well as number needed to treat. Statin prescriptions by CHD risk

Framingham 10-year CHD risk All Patients Statins Under Current Practice Statins Under ACC/AHA Guidelines p-value N (%) N (%)* N (%)* Low (<5%) 41,376 (49%) 3000 (7%) 234 (0.6%) <0.001 Medium (5-15%) 30,135 (36%) 6441 (21%) 4973 (17%) <0.001 High (>15%) 12,768 (15%) 4317 (34%) 12087 (95%) <0.001

*Row percentage

INTERHOSPITAL TRANSFERS: PATIENT CHARACTERISTICS AND OUTCOMES Cecelia N. Theobald1,2; Stephan Russ3; Jesse Ehrenfeld4; Sunil Kripalani2. 1VA Tennessee Valley Healthcare System, Nashville, TN; 2Vanderbilt University, Nashville, TN; 3Vanderbilt University, Nashville, TN;

4Vanderbilt University, Nashville, TN. (Tracking ID #1939070)

BACKGROUND: The transfer of inpatients between hospitals is often necessitated by differential expertise and capacity among facilities. At some institutions, interhospital transfers constitute nearly 20% of all inpatient admissions and there is concern they may experience poorer outcomes. Much of the literature to date has examined small subpopulations of interhospital transfers (such as ICU, trauma, and burn patients) without evaluating the characteristics of transfer patients as a population. Our objective was to compare the arrival characteristics and outcomes of interhospital transfer patients with those of patients directly admitted to a large academic medical center. METHODS: We conducted a retrospective cohort study of patients transferred into an academic medical center and compared these with patients directly admitted during an eighteen month period. Patients were excluded if they were transferred into the Emergency Department, Labor and Delivery, or admitted to burn or trauma

• services. Admission characteristics studied included demographics, site of admission (ICU vs. non-ICU),

severity of illness (measured using the modified Elixhauser comorbidity index, range of possible scores -14 to 60), admitting service, and time of arrival. Outcome measures included length of stay, ICU length of stay, in- hospital mortality, and timeliness of initial inpatient care. Transfer patients were compared with non-transfer patients using simple univariate analysis. RESULTS: Transferred and non-transferred patients had similar rates of ICU admission (46.5% vs. 47.4%, p = 0.53) but transferred patients had higher severity of illness (mean modified Elixhauser score 12.0 vs. 9.3, p < 0.001). Nearly two thirds, or 65% of interhospital transfers arrived during overnight hours (6PM to 7AM), vs.

• nly 56.2% of non-transferred patients (p < 0.001). Transferred patients waited on average 18 minutes longer

for admission orders (p < 0.001) and 15 minutes longer for non-PRN medication orders (p < 0.001). Furthermore, transferred patients experienced significantly longer length of stay (6.0 vs. 3.1 days, p < 0.001) and ICU length of stay (4.4 days vs. 2.5 days, p < 0.001). Finally, transferred patients had over twice the in- hospital mortality of non-transferred patients (10.9% vs. 4.9%, p < 0.001). CONCLUSIONS: Interhospital transfer patients as a population have an increased severity of illness and experience longer length of stay, delays in initial inpatient care, and higher in-hospital mortality when compared with non-transferred patients. Hospitals may want to focus specific resources on this unique high-risk population. Table: Comparison of transfer and non-transfer population

Characteristic Transfers (n=1715) Non-transfers (n=6176) p value Age, mean (SD) 55.3 (16.6) 57.7 (16.9) <0.0011 Male % 48.6% 56.3% <0.0012 ICU admission (%) 46.5% 47.4% 0.532 Modified Elixhauser index, mean (SD) 12.0 (10.5) 9.3 (9.6) <0.0011 Service Medicine Surgery Neurology OB/Gyn 55.7% 34.1% 8.8% 1.5% 63.6% 29.7% 6.0% 0.6% <0.0012 Time of arrival Day shift (07:00 - 17:59) Night shift (18:00 - 06:59) 35.2% 64.8% 43.8% 56.2% <0.0012 Time to in minutes, median (IQR) Admission order entry Non PRN medication order Antibiotic order 46 (19 - 90) 28 (12 - 61) 199 (81 - 505) 28 (3.5 - 70) 13 (0 - 41) 190.5 (35 - 615) < 0.0013< 0.0013 0.00113 Length of stay, median (IQR) Total hospital ICU 6.0 (3.2 - 11.3) 4.4 (2.3 - 8.9) 3.1 (1.8 - 6.3) 2.5 (1.3 - 5.0) < 0.0013< 0.0013 In-hospital mortality % 10.9% 4.9% <0.0012 Statistical tests used: 1 T-test 2 Chi-square 3 Wilcoxon rank-sum

AGING WITH HIV: SYMPTOM BURDEN AND IMPACT ON QUALITY OF LIFE Meredith Greene1,2; Christine Ritchie1; Jeff Martin4; Joy A. Madamba3; Monica Mattes5; Yinghui Miao2; Steven Deeks3; Victor Valcour1,6. 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

• f 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

• f 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. 69% of participants reported 2 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
• r hands" (n=54, 38%) and insomnia (n=53, 37%) were the most commonly reported symptoms. 55% 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.

SEX, GENDER, SOCIOECONOMIC STATUS AND ACCESS TO CARDIAC CATHETERIZATION IN THE CONTEXT OF UNIVERSAL HEALTHCARE COVERAGE Gabriel E. Fabreau2,1; Alexander A. Leung5,3; Danielle Southern3; John Z. Ayanian4,1; William A. Ghali3,4.

1Harvard Medical School, Boston, MA; 2Brigham and Women's Hospital, Boston, MA; 3University of Calgary,

Calgary, AB, Canada; 4University of Michigan, Ann Arbor, MI; 5University of Calgary, Calgary, AB, Canada. (Tracking ID #1939475) BACKGROUND: Sex and gender disparities have been described in cardiac care and outcomes following acute coronary syndromes (ACS). Socioeconomic status (SES) may additionally affect medical care and health

• utcomes, partly through barriers in timely access to cardiac catheterization. In Canada, a universal healthcare

system may reduce barriers related to sex/gender and SES. We sought to determine whether sex/gender and SES interact to modify the receipt of cardiac catheterization and mortality following an ACS in a universal healthcare system. METHODS: Using a provincial multicenter cardiac registry, we assembled a cohort of 14,012 patients admitted with an ACS to any cardiology service in the southern health zones of Alberta, Canada between April 18, 2004 and December 31, 2011 by linking census, vital statistics and clinical registry data. SES was estimated using residential neighbourhood median household income from the 2006 Canadian census. We compared the

• dds of receiving a cardiac catheterization within 1 and 30 days after admission, and the odds of death within 30

days and 1 year after admission according to income quintiles and stratified by sex. Using multivariable logistic regression we controlled for age, geography, cardiac risk factors and clinical comorbidities to estimate the adjusted odds ratios (ORs) of receiving cardiac catheterization and of death. RESULTS: Unadjusted rates of cardiac catheterization were higher for men compared to women (41% vs. 31% at one day, and 68% vs. 59% at 30 days, p<0.001 for both time points), and in this context, men had lower mortality rates (2% vs. 2.8% at 30 days, and 5.2% vs. 7.4% at 1 year, p<0.001 for both time points). In models adjusting for SES, women were less likely to receive cardiac catheterization within 1 day (OR 0.79, 95% confidence interval [CI], 0.71 - 0.87) and 30 days (OR 0.73, 95% CI, 0.62-0.86) of admission with an ACS. When examined across quintiles of SES, adjusted models revealed differing relationships among variables for men vs. women: each incremental decrease in income quintile was associated with a 7% lower odds of receiving cardiac catheterization for women (p=0.005) vs. a smaller 3% decrement in odds for men (p=0.03). Additionally, among women, each decrease in income quintile was associated with a 13% higher odds of 30- day mortality (p=0.02) vs. a 4% higher odds of 30-day mortality for men (p=0.39). No SES differences were seen in the adjusted 1-year odds of death for either men or women, nor between them. CONCLUSIONS: The relationships between socioeconomic status and use of cardiac catheterization and mortality after ACS are different for men vs. women, with women seemingly more vulnerable to the detrimental associations of low income. These findings were present despite a universal healthcare system that does not have any patient user fees. This suggests that factors other than insurance status and/or ability to pay are at play, and that elements of sex and/or gender are effect modifiers. Care protocols designed to improve access to care and improve outcomes in women, especially low SES women, are required.

COMMUNITY COLLEGE PATHWAYS: IMPROVING THE U.S. PHYSICIAN WORKFORCE PIPELINE Efrain Talamantes1; Carol Mangione2,3; Karla Gonzalez2; Alejandro Jimenez4; Fabio R. Gonzalez5; Gerardo Moreno2,6. 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

• f 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.

INSTITUTIONAL PREDICTORS OF MEDICAL STUDENTS' INTERACTIONS WITH THE PHARMACEUTICAL AND MEDICAL DEVICE INDUSTRIES James Yeh1; Kirsten Austad1; Jessica Franklin1; Susan Chimonas2; Eric Campbell3; Jerry Avorn1; Aaron

• Kesselheim1. 1Brigham and Women's Hospital and Harvard Medical School, Boston, MA; 2Columbia

University College of Physicians and Surgeons, New York, NY; 3Massachusetts General Hospital and Harvard Medical School, Boston, MA. (Tracking ID #1921316) BACKGROUND: Medical students frequently interact with pharmaceutical and medical device industries. Students receive biased information from these interactions, which may lead to non-evidence-based prescribing. Thus, medical schools have designed industry interaction policies regulating students' contacts with the

• industry. Professional societies have developed metrics to evaluate the strength of these policies. We compared

metrics from two prominent professional societies and investigated which aspects of industry interaction policies were most closely associated with student outcomes. We also considered whether certain characteristics

• f medical schools were associated with their students' behavior and attitudes towards pharmaceutical

marketing. METHODS: A national sample of 1610 first- and fourth-year medical students answered survey questions on acceptance of gifts from the industry, interaction with sales representatives, and perceived adequacy of faculty and industry separation. We used hierarchical logistic regression models to determine the association between these reported outcomes with their schools' strength of industry interaction policies. The strength of the policies was measured using the American Medical Student Association (AMSA) PharmFree Scorecard and the Institute

• n Medicine as a Profession's (IMAP) Score. The model included adjustments for year in training, medical

school size, and the level of National of Institute of Health (NIH) funding received by the medical school. We also performed LASSO logistic regression models to determine which policy domains were best associated with students' reported outcomes. RESULTS: IMAP Score had a lower median value (1.75, interquartile range [IQR]:2-2.45) than the AMSA Score (2.18, IQR:1.5-2). The policy dimensions shared by AMSA and IMAP were not well-correlated (gift policies r=0.28, 95% confidence interval [CI]:0.11-0.44); sales representative access policies r=0.51, 95% CI:0.36-0.63). Students from schools with the highest-ranked industry interaction policies were less likely to accept gifts (AMSA Score odds ratio [OR]:2.65, 95% CI:1.38-5.07; IMAP Score OR:2.19, 95% CI:0.96-5.03) and less likely to interact with sales representatives (AMSA Score OR:2.99, 95% CI:1.43-6.27; IMAP Score OR:2.71, 95% CI:1.05-6.97) than students from schools with lowest-ranked policy scores. The association became non-significant when fully adjusted for NIH funding. Policies governing receipt of gifts (OR:1.01- 1.30), meals (OR:1.13), and speaking bureaus (OR:1.02-1.13) were best associated with students' reported behavior of gifts receipt and sales representatives interaction. By contrast, policies regulating industry's role in educational activities (e.g., CME, travel compensation, and scholarship) were best associated with outcome of perceived adequate separation between faculty and industry (OR:1.01-1.32). CONCLUSIONS: Policies restricting gifts had significant impact on students' behavior in reducing the receipt

• f industry gifts, sales representative interactions, and increased perception of adequate separation among the

faculty and industry. Medical schools' strength of industry interaction policy, as determined by AMSA or IMAP scores, was associated with student reported behavior. However, this relationship was confounded by the level

• f NIH funding. As medical schools revise their industry interactions policies, limitation on the receipt of gifts

should be emphasized and greater attention paid to less research-intensive schools.