Factors Influencing Internal Medicine Program Directorsʼ Decisions about Applicants

Objective: With the introduction of the Electronic Residency Application Service (ERAS) for internal medicine programs in 1999, residency directors needed to use its categories of information. We examined which factors in the applicant's file were most useful to residency program directors in deciding about interview invitations and ranking candidates. Description: In conjunction with the Association of Program Directors in Internal Medicine (APDIM), we undertook a Web-based survey of all 407 program directors in internal medicine. An e-mail message containing a direct Internet link to the survey was distributed to all APDIM listserve members. The response rate was increased by personalized e-mails, letters, and faxed requests to the directors. We received 332 responses (81.5%). All data were down-loaded into SAS for analysis. On the survey, programs categorized themselves by type of program (university, community), size, and geographic location. They rated the usefulness of dean's letters, personal statements, transcripts, application forms, USMLE scores, and chairman's letters for making decisions about interview invitations and ranking applicants. Applicant interview was an additional variable for ranking applicants. In rating usefulness, respondents could choose highly, moderately, mildly, and not. Optional text boxes for comments were provided. The items rated highly or moderately useful for interview decisions were USMLE scores (94%), dean's letters (87%), transcripts and application forms (85%), chairman's letters (79%), and personal statements (61%). The items rated highly or moderately useful for ranking decisions were applicant interviews (96%), USMLE scores (93%), transcripts (83%), dean's letters (82%), chairman's letters (77%), application forms (74%), and personal statements (57%). Chi-square analysis by program type was done with the interview and ranking criteria. A logistic regression model was developed with program type as outcome, and three applicant-ranking criteria (transcripts, chairman's letters, and USMLE scores) using 322 observations. The overall p value (.0001) indicated that the predictors reliably distinguished between university and community programs. The university programs were 0.70 times as likely to rate transcripts more useful than community programs (−0.68, −0.057, confidence limits, p =.02); they also were 0.70 times as likely to rank the chairman's letter as useful (−.68, −.085, p =.01). However, community programs were 1.7 times as likely as university programs to rank USMLE scores as useful (0.22, 0.93, p =.001). Discussion: USMLE scores are consistently important since they allow programs to make objective comparisons between applicants. Community programs found these scores more useful than did university programs, which found transcripts and chairman's letters more useful. With ERAS, programs can filter pass/fail on USMLE, but filtering by score is not possible. While personal statements often reveal unique individual characteristics, they are seen as the least useful information and are harder to evaluate. The interview is an opportunity to assess personal characteristics and is critical for determining candidates' suitability. Comments revealed that while dean's letters are seen as useful, many programs start inviting candidates for interviews before the letters are released. In addition, the considerable variability in quality of these letters was noted, as was the redundancy between dean's and chairman's letters. This information about the way ERAS information is being used will be of interest to residency directors and applicants.