Objectives  Using a novel longitudinal tracking project, this study develops and evaluates the performance of a predictive model and index of rural medical practice intention based on the characteristics of incoming medical students.

Methods  Medical school entry survey data were obtained from the Medical Schools Outcome Database (MSOD) project implemented in all Australian and New Zealand medical schools and coordinated through Medical Deans Australia and New Zealand, the representative body for the Deans of 18 Australian and two New Zealand medical schools and faculties. The medical school commencement survey collects data on students’ education and family background, including rural upbringing, personal circumstances and scholarships, and on their practice intentions in terms of location and specialty. The MSOD will also allow tracking of medical graduates after graduation. Logistic regression modelling was used to develop a predictive model of rural practice intention. Split-sample validation was used to gain some insight into the stability of performance of the model.

Results  Response rates to the MSOD survey exceeded 90% on average. The model findings confirm and extend previous research examining the association of medical student characteristics with intention to take up rural medical practice. The statistically significant independent factors in the model included students’ rural backgrounds, financial arrangements and intentions regarding specialist versus generalist practice upon graduation. Model performance was good, with an area under the receiver-operator characteristics curve of 0.86, and reproducible, with an area in a validation sample of 0.83.

Conclusions  The model and related index provide important insights into individual factors associated with rural practice intention among students commencing medical studies. The model can also provide a means for optimising the use of scarce medical programme resources, thereby helping to improve the supply of rural medical practitioners. This study illustrates the power and potential of a robust, consistent, systematic longitudinal tracking project.