BACKGROUND AND METHODS: In a nationally representative population 65 years of age or older, we have demonstrated that about one quarter of coronary angiographies and upper gastrointestinal endoscopies and two thirds of carotid endarterectomies were performed for reasons that were less than medically appropriate. In this paper we examine whether specific characteristics of patients (age, sex, and race), physicians (age, board-certification status, and experience with the procedure), or hospitals (teaching status, profit-making status, and size) predict whether a procedure will be performed appropriately. RESULTS: In general, we found that little of the variability in the appropriateness of care (4 percent or less) could be explained on the basis of standard, easily obtainable data about the patient, the physician, or the hospital. For all three procedures, however, performance in a teaching hospital increased the likelihood that the reasons would be medically appropriate (P = 0.09 for angiography, P = 0.30 for endoscopy, and P less than 0.01 for endarterectomy). In addition, angiographies were more often performed for appropriate reasons in older or more affluent patients (P less than 0.01 for both). Being treated by a surgeon who performed a high rather than a low number of procedures decreased the likelihood of an appropriate endarterectomy by one third, from 40 to 28 percent (P less than 0.01). CONCLUSIONS: Appropriateness of care cannot be closely predicted from many easily determined characteristics of patients, physicians, or hospitals. Thus, for the present, if appropriateness is to be improved it will have to be assessed directly at the level of each patient, hospital, and physician.
BACKGROUND AND METHODS: In a nationally representative population 65 years of age or older, we have demonstrated that about one quarter of coronary angiographies and upper gastrointestinal endoscopies and two thirds of carotid endarterectomies were performed for reasons that were less than medically appropriate. In this paper we examine whether specific characteristics of patients (age, sex, and race), physicians (age, board-certification status, and experience with the procedure), or hospitals (teaching status, profit-making status, and size) predict whether a procedure will be performed appropriately. RESULTS: In general, we found that little of the variability in the appropriateness of care (4 percent or less) could be explained on the basis of standard, easily obtainable data about the patient, the physician, or the hospital. For all three procedures, however, performance in a teaching hospital increased the likelihood that the reasons would be medically appropriate (P = 0.09 for angiography, P = 0.30 for endoscopy, and P less than 0.01 for endarterectomy). In addition, angiographies were more often performed for appropriate reasons in older or more affluent patients (P less than 0.01 for both). Being treated by a surgeon who performed a high rather than a low number of procedures decreased the likelihood of an appropriate endarterectomy by one third, from 40 to 28 percent (P less than 0.01). CONCLUSIONS: Appropriateness of care cannot be closely predicted from many easily determined characteristics of patients, physicians, or hospitals. Thus, for the present, if appropriateness is to be improved it will have to be assessed directly at the level of each patient, hospital, and physician.