Predicting hospital charge and stay variation. The role of patient teaching status, controlling for diagnosis-related group, demographic characteristics, and severity of illness.

Teaching hospitals have higher charges than nonteaching hospitals, and teaching patients within a single institution incur higher charges than do private patients. Teaching hospitals claim that these higher charges are due to caring for patients with more complex illnesses. Critics claim that the increased charges are primarily related to the hospitals' medical education function. Hospital charges and stay were compared across three physician groups: private patients with a community attending physician, teaching patients with a community attending physician, and teaching patients with a faculty attending physician. Severity of illness differences were controlled for by selecting patients from four Diagnosis Related Groups; refining these groups by surgical codes and age; identifying prehospital and inhospital severity of illness control variables; and omitting death and transfer cases in a second analysis. Least squares regression with logarithmic transformations of the dependent variables revealed that placement of patients on the teaching service with a faculty attending physician led to significantly higher charges as compared with private, nonteaching patients. The patients on the teaching service with a community attending physician did not incur significantly higher charges. It was also revealed that many severity of illness variables predicted a significant amount of variation in hospital charges, even after controlling for diagnostic group and teaching status of the patient.