Can we identify medical students at admission who will enter primary care residencies?

We sought to estimate the predictive value of admissions interviewers' impressions about student interest in primary care at the University of Kentucky College of Medicine (UKCOM). We counted the number of UKCOM students considered by the admissions interviewers to be interested in primary care careers prior to matriculation. We then compared the interviewers' impressions with the same students' choices of residency programs at graduation. Analyses were restricted to students who entered UKCOM in 1995, 1996, and 1997, and graduated on time in 1999, 2000, and 2001. Each student completed two admissions interviews. Over the three-year study period, both admissions interviewers agreed that 107 (45%) of 237 graduating students demonstrated a clear preference at matriculation for entering a primary care career. Of these 107 graduates, 69 (64%) entered a primary care field. Of the 69 students labeled as not inclined toward primary care, 26 (39%) did enter a primary care field. Of the 61 students about whom the interviewers had split opinions, 30 (49%) entered a primary care field. All total, 125 (53%) of the 237 graduates entered a primary care field. Agreement by both interviewers on students' predilections toward (or away) from primary care were correct 64% of the time. A major limitation in our analysis was lack of information on how many primary carebound students would ultimately pursue subspecialty fellowships. We conclude that interviewers can help select medical students interested in primary care, that their performance might be improved by more meticulous application of known predictive factors, and that their performance could be more accurately measured if data on ultimate practice type at the end of all training were available.