CLASSIFICATION MODEL THAT PREDICTS MEDICAL-STUDENTS CHOICES OF PRIMARY CARE OR NONPRIMARY CARE SPECIALTIES

To identify factors that influence students to choose primary care or non-primary care specialties, the authors surveyed the 509 graduating students at the Medical College of Georgia School of Medicine in 1988, 1989, and 1990. Using a Likert-type scale, the 404 responding students rated potential influences on their speciality choices from 1, very negative, to 7, very positive. The students choosing primary care specialities were positively influenced significantly more often by their desire to keep options open (85% versus 58%, p < .001) and their desire for longitudinal patient care opportunities (95% versus 54%, p < .001). Those choosing non-primary care specialties were more often influenced by their desire for monetary rewards (69% versus 35%, p < .001) and by their perceptions of lifestyle following residency (74% versus 60%, p < .01) and prestige of the specialty (57% versus 36%, p < .001). The authors used multiple discriminant analysis to derive a discriminant function that would permit classification of students into primary care and non-primary care groups. The potential influences of desire for longitudinal care opportunities and desire for monetary rewards were statistically and clinically significant for all three years. Using the discriminant function, the authors correctly classified 81%, 79%, and 78% of the students' specialty choices for 1988, 1989, and 1990, respectively. The authors suggest that addressing the issue of monetary rewards will be necessary before the primary care fields again become attractive to students.