Nature and nurture in the family physician's choice of practice location

An understanding of the contextual, professional, and personal factors that affect choice of practice location for physicians is needed to support successful strategies in addressing geographic maldistribution of physicians. This study compared two categories of predictors of family practice location in non-metropolitan areas among undergraduate medical students: individual characteristics (nature), and the rural program component of their training program (nurture). The study aimed to identify factors that predict the location of practice 2 years post-residency training and determine the predictive value of combining nature and nurture variables using administrative data from two undergraduate medical education programs.

Databases were developed from available administrative sources for a retrospective analysis of two undergraduate medical education programs in Canada: Université de Sherbrooke (UdeS) and University of British Columbia (UBC). Both schools have a strong mandate to evaluate the impact of their programs on physician distribution. The dependent variable was location of practice 2 years after completing postgraduate training in family medicine. Independent variables included individual and program characteristics. Separate analyses were conducted for each program using multiple logistic regression.

The nature and nurture variables considered in the models explained only 21% to 27% of the variance in the eventual location of practice of family physician graduates. For UdeS, having an address in a rural/small-town environment at application to medical school (OR=2.61, 95% CI: 1.24-6.06) and for UBC, location of high school in a rural/small town (OR=4.03, 95% CI: 1.05-15.41), both increased the chances of practicing in a non-metropolitan area. For UdeS the nurture variable (ie length of clerkship in a non-metropolitan area) was the most significant predictor (OR=1.14, 95% CI: 1.067-1.22). For both medical schools, adding a single nurture variable to the model using only nature variables significantly increased the amount of variation accounted for in predicting location of practice in non-metropolitan areas.

Aspects of graduates' rural background increase the chances of practicing in a non-metropolitan area. A third-year clerkship experience in a rural area may increase the chances of non-metropolitan practice. Although the total variation predicted by both nature and nurture variables in this study was small, adding a nurture variable significantly improves the prediction of individuals who will practice in a non-metropolitan area. The fact that total variation predicted was small is likely to be due to the limitations of the administrative databases used. Different strategies are being implemented in each university to improve the quality of existing administrative databases, as well as to collect relevant data about intent-to-practice, training characteristics, and the attitudes, beliefs and backgrounds of students.