We collected information through four different modalities: a literature review, a written survey of medical school Deans and students, focus groups of preclinical and clinical medical students and dean, and site visits at and interviews with medical schools with reportedly outstanding student health promotion programs.
Medical student and dean surveys
There were 17 respondents to the Dean Survey (DS), representing 12 of the 16 schools in the nationally representative Healthy Doc (HD) project [2
]. Two deans responded from Mercer, RWJ/UMDNJ, Tulane, UCLA, and University of Pennsylvania, while Colorado, Creighton, Emory, Georgetown, Loma Linda, Medical College of Georgia, and University of Rochester each had one dean respond. It was not always clear whether the Dean of Curriculum or the Dean of Student Affairs was the respondent, therefore we did not differentiate in the analyses by dean type. We also compared Deans' responses with responses (83% response rate) from the 1336 medical students in the Class of 2003 in these Deans' schools, as they were about to begin on wards. All medical students in that class were eligible to complete a self-administered questionnaire covering personal and professional health promotion topics. Our sample of schools was designed to be representative of all U.S. medical schools in our geographic distribution, age (our freshman average was 24 vs. 24 nationally), school size (our schools averaged 563 medical students/school vs. 527 nationally), NIH research ranking (our average was 64 vs. 62 nationally), private/public school balance (51% in private schools vs. 41% nationally), under-represented minorities (13% Blacks, Hispanics, and Native Americans, vs. 11% nationally), and gender (45% women vs. 43% nationally)5–7
Methodology for gathering medical student data in HD has been more fully described elsewhere [2
]. DS data were collected between February 2002 and April 2003.
In analyses comparing DS and HD data, DS schools with two respondents were first averaged so that each school is represented by one value (since repeated measures analysis was not available for the desired analyses). Variation between deans representing a school was quite low for all but one pair. By averaging for the five dean pairs (and consequently having a sample size of 12 rather than 17), the tests are conservative. Student opinion scores were also averaged for each of the twelve schools from which we received Dean responses; these averages were then correlated with the Dean's scores using Spearman's correlation method. For questions with fairly uniform responses by either Deans or students, Wilcoxon's Signed Rank Test was used to test if there were consistent differences between student and Dean opinion. The two variables to be correlated were ordinal variables, each with 5 levels. The type of correlation method was therefore limited to a non-parametric method. Additionally, the raw student data was clustered by school, requiring methods suitable for correlated data. Since the non-parametric method needed is not available for correlated data, we determined that the best method was to take the student mean values at each school to correlate with the dean values. While this ignored the student variability within school, this deficit was balanced by the fact that the much smaller n would require much stronger evidence of a relationship to evince a significant result.
Deans were also asked to rate their school relative to other schools. To compare these ratings to students' opinions, schools were ranked using their mean student scores on each question related to prevention and healthy activities encouraged by the school. All 16 schools in the HD cohort were used in the ranking process (1 = highest, 16 = lowest), not just the 12 schools represented by the responding deans, as the 16 were the intended sample, and are representative of US medical schools [2
]. Therefore, the twelve schools for which we have Dean data could have rank values between 1 and 16. For Deans' survey questions without comparative HD data, only simple descriptive statistics are presented.
Medical student and dean focus groups
For our focus groups (conducted in 2002), we identified opportunities where there would be a wide and nationally representative range of medical schools. The first focus group was convened at the AMSA Chapter Officers' Training Conference (COC) attended by student leaders (primarily rising second years) from every U.S. osteopathic and allopathic medical school. AMSA invited a random sampling of those attending the COC to participate in the focus group. Since the first focus group of students attracted 10 first and second year students, the second focus group was a random sample of 12 clinical students; both student focus groups had an even gender mix. Because Philadelphia has so many medical schools (five), we sampled for the second focus group from those Philadelphia students who were listed in AMSA's membership database. Deans of Primary Care were invited to the third focus group convened at the annual conference of the Association of American Medical Colleges. AMSA used the list of Primary Care Deans and invited a random sample of them to attend the focus group; four attended. An outside contractor (Bennett, Petts & Blumenthal) assisted AMSA in developing the focus group guide, conducted all three focus groups, transcribed the conversations and analyzed the notes for trends in responses.
Site visits and interviews
In 2002–2003, we identified medical school campuses with intensive programs in medical student well-being through literature and web searches, recommendations from project advisory panel members, results from the Association of Academic Health Centers' American Network of Health Promoting Universities assessment, and participants in the HRSA-funded UME-21 project. Site visits and in-depth interviews were conducted using a protocol which sought information and recommendations on the following topics:
• Student well-being programming, including the policies, activities, and evaluation for such efforts as stress reduction, exercise, diet, and mentoring.
• Prevention in the curriculum using the Healthy People 2010 objectives and how the various topics are integrated, taught, and evaluated.
• Deans' office support (including financial) for prevention in the curriculum and student wellness activities.
• Student assessments and recommendations regarding their schools' efforts.