Using knowledge of multiple levels of variation in care to target performance incentives to providers.

Background: In developing "pay-for-performance" and capitation systems that provide incentives for improving the quality and efficiency of care, policymakers need to determine which healthcare providers to evaluate and reward.

Objectives: This study demonstrates methods for determining and understanding the relative contributions of facilities and physicians to the quality and cost of care. Specifically, this study distinguishes levels of variation in resource utilization (RU), based on research to support the development of an expanded Medicare dialysis prospective payment system.

Research Design: Mixed models were used to estimate the variation in RU across institutional providers, physicians, patients, and months (within patients), after adjusting for case-mix.

Subjects: The study includes 10,367 Medicare hemodialysis patients treated in a 4.2% stratified random sample of dialysis facilities in 2003.

Measures: Monthly RU was measured by the average Medicare allowable charge per dialysis session for separately billable dialysis-related services (mainly injectable medications and laboratory tests) from Medicare claims.

Results: There was financially significant variation in RU across institutional providers and to a lesser degree across physicians, after adjusting for differences in case-mix. The remaining variation in RU reflects unexplained differences across patients that persist over time and transitory fluctuations for individual patients.

Conclusions: The greater variation in RU occurring across dialysis facilities than across physicians is consistent with targeting payments to facilities, but alignment of incentives between facilities and physicians remains an important goal. Similar analytic methods may be useful in designing payment policies that reward providers for improving the quality of care.