Addressing physician concerns about performance profiling: experience with a local Veterans Affairs quality evaluation program

Abstract

The Authors investigated the addition of novel quality indicators, patient risk adjustment, and simple statistics in an ongoing clinician feedback initiative that profiles diabetes care for 13 Veterans Affairs (VA) clinics. Data were extracted from a computerized database for calendar years 2004 to 2005. Performance was assessed with 4 monitoring measures, 3 intermediate outcomes, and 3 appropriate treatment measures. Attainment rates for each indicator were calculated by clinic. The effect of risk adjustment and the significance of clinic performance variation were determined with multivariate logistic models. Analysis of the 10 quality measures revealed lower attainment and greater clinic-level variation for the less familiar indicators. Statistically significant performance variations were detected among clinics, with several being of a clinically important magnitude. Risk adjustment did not substantially change performance. The addition of clinically relevant quality measures and simple statistics appeared to enhance the characterization of performance by this profiling program.