potential sources of data, and to produce data and reports that are understandable to a broad audience. Involving the community and responding to its concerns may increase the community's interest in and acceptance of the findings, particularly negative ones.

The Center for the Advancement of Health in Washington, D.C., in connection with the California Wellness Foundation's Health Improvement Initiative, has considered how performance indicators developed for health plans might become a tool for accountability to stakeholders in communities served by health plans (see Sofaer, 1995). These stakeholders include consumers, employers, and public agencies, including regulators. The Center's expanded view of health emphasizes psychosocial and behavioral aspects of the delivery of health services and a public health perspective for the assessment of services to improve health.

The project identified several functions of performance indicators: specifying criteria for evaluation and values regarding health and health services; making explicit the expectations for some aspects of health care delivery; providing information for decisions on health services; supporting quality assessment and improvement; and, potentially, guiding the development of information systems. Further consideration focused on the normative, technical, strategic, and operational aspects of performance indicators. The normative element reflects value judgments made in selecting areas of performance (i.e., health outcomes) for which health plans or other organizations or individuals will be held accountable. Technical aspects of performance indicators include measurement issues such as the quality of the data being used and the validity and reliability of the indicators. Indicators must also permit meaningful comparisons across entities. Strategic concerns relate to the purposes indicators are expected to serve. The appropriate number, focus, and mix of indicators (e.g., outcomes versus structure or process) require consideration as does setting targets for desired performance at levels that will lead to meaningful improvements. Operational issues include the feasibility of obtaining data and approaches to disseminating results.