Composite Measures for Predicting Hospital Mortality with Surgery

Executive Summary: Individual quality measures have significant limitations for assessing surgical performance. Despite growing interest in composite measures, methods for empirically combining multiple domains of surgical quality are not well established. This paper outlines the methods for creating a simple composite score based on a combination of surgical mortality and hospital volume. The methods used are based on empirical Bayes techniques, where the mortality rate is weighted to the extent it is reliable and the remaining weight is placed on hospital volume. We validated this composite measure by assessing how well it predicts subsequent performance. This perspective is particularly relevant for selective-referral or public reporting contexts. Our results demonstrate that a simple composite of mortality and volume is a better predictor of subsequent performance than either measure alone for most operations. These measures will help patients and payers identify hospitals likely to have superior outcomes.