Economic profiling of physician specialists: use of outlier treatment and episode attribution rules.

This paper examines the influence of episode attribution methodology and cost outlier methodology on the accuracy of physicians' economic profiles. Four years of claims data from a mixed model HMO were processed using the leading episode grouper software. Episode grouped results then were applied to construct input distributions for a simulation model. For each of four specialties (cardiology, family practice, general surgery, and neurology), we employed sets of 18 simulations to investigate the effects of three alternative episode attribution methodologies and six alternative cost outlier methodologies on sensitivity, specificity, and positive predictive error in classifying cost-efficient and cost-inefficient physicians. For identification of cost-efficient physicians, the most accurate profiling results were obtained when Winsorizing outliers at 2% and 98% of episode-type cost distributions, and attributing responsibility for episode costs to physicians who accounted for at least 30% of associated professional and prescribing fees. No consistent combination of outlier methodology and episode attribution rule was found to be superior for identifying cost-inefficient physicians.