Graduate medical education and physician practice location. Implications for physician workforce policy

To determine the relationship between graduate medical education and physician practice location.

Cross-sectional analysis of physicians in active practice in 1993, classified by state of graduate medical education and stratified by specialty and professional activity. Logistic regression analysis was used to examine predictors of physicians remaining to practice in the same state in which they trained.

There were 82,871 allopathic physicians (national random sample) and 15,076 osteopathic physicians (universe) who completed graduate medical education between 1980 and 1992.

Practice location in the same state as graduate medical education.

Overall, 51% of physicians are practicing in the state in which they obtained their graduate medical education (range among states, 6% to 71%). Generalist physicians are more likely than specialists to remain in their state of graduate medical education (odds ratio [OR], 1.36; 95% confidence interval [CI], 1.33 to 1.40) There is a weak negative association between the number of physicians in training per capita in a state and the likelihood of a physician remaining in the state to practice (OR, 0.90; 95% CI, 0.90 to 0.91, for an increment in resident supply of 10 per 100,000 population). New York and Massachusetts, the states with the highest numbers of residents per capita, retained 51% and 49%, respectively, of their graduates, placing them near the median among states.

Most physician training and practice locations function as a national market, with physicians dispersing relatively widely after completing graduate medical education. States that produce high numbers of physicians per capita do not appear to play a unique role in training physicians to serve a national market. These findings pose challenges for states attempting to modify their physician supply and specialty mix.