Indirect costs of teaching in Canadian hospitals.

We sought to determine whether there are indirect costs of teaching in Canadian hospitals. To examine cost differences between teaching and nonteaching hospitals we estimated two cost functions: cost per case and cost per patient-day (dependent variables). The independent variables were number of beds, occupancy rate, teaching ratio (number of residents and interns per 100 beds), province, urbanicity (the population density of the county in which the hospital was situated) and wage index. Within each hospital we categorized a random sample of patient discharges according to case mix and severity of illness using age and standard diagnosis and procedure codes. Teaching ratio and case severity were each highly correlated positively with the dependent variables. The other variables that led to higher costs in teaching hospitals were wage rates and number of beds. Our regression model could serve as the basis of a reimbursement system, adjusted for severity and teaching status, particularly in provinces moving toward introducing case-weighting mechanisms into their payment model. Even if teaching hospitals were paid more than nonteaching hospitals because of the difference in the severity of illness there should be an additional allowance to cover the indirect costs of teaching.