Performance measurement: problems and solutions.

Abstract

PROLOGUE: The issues around the quality of medical care have become a subject of increasing complexity, confusion, and public interest. As managed care has rapidly emerged as the preferred delivery and financing mode of purchasers, if not all providers, it has thrust upon physicians new demands for accountability. However, the available tools to accurately measure performance of providers are quite limited. In this lead paper David Eddy discusses the challenges facing individuals and organizations that are striving to create better ways to measure performance and—importantly—offers some constructive ways to address the shortcomings. As a physician (with a degree from the University of Virginia) who operates at the intersection of medicine and applied mathematics (doctorate, Stanford University), Eddy is certainly well equipped for the task. Beyond his academic credits, Eddy also has devoted considerable energy to helping the National Committee for Quality Assurance (NCQA) pursue the complex tasks around improved measurement, including service on the Committee on Performance Measurement, which produces the Health Plan Employer Data and Information Set (HEDIS) measures. Eddy also sees performance measurement from the viewpoint of providers and plans. He has consulted widely with many medical organizations and currently serves as senior adviser for health policy and management at Kaiser Permanente Southern California. Health Affairs invited several responses to Eddy's paper, which was based on the Richard and Hinda Rosenthal Lecture that he presented recently at the Institute of Medicine, National Academy of Sciences. Recent efforts to measure performance have established its feasibility and value. However, its full potential is currently limited by several problems. They include the probabilistic nature, rarity, and confounding of many health outcomes; the inadequacies of information systems; the multiplicity of measurers and measures; the complexity of health plans; and the availability of funding. Solutions are to rely more on process measures; to justify every measure with a formal, evidence-based rationale; to improve information systems; to supplement population-based measures with case-based measures; to develop a single, nationally standardized set of measures; and to provide nonpolitical, public funding for the design and administration of measures.