Despite concerted efforts by major stakeholders over several decades, the quality of care in US hospitals is still known to be variable and often suboptimal. CMS, in conjunction with JCAHO and AHA, developed a “starter set” of 10 hospital-quality indicators to begin to assess the extent to which acute-care hospitals provide optimal care; the set of indicators collected has since expanded to 20. Approaches are being tested for using indicators of this type to develop pay-for-performance (P4P) methods that will reward hospitals for high quality.

There is considerable interest in analysis of these indicators, yet to date all information has been based on hospital-level aggregate results. In this project, we are using the national, patient-level data collected through this initiative—known as the Hospital Quality Alliance (HQA)—to describe the quality of care in the United States, to develop alternative composite measures, and to simulate the impact of alternative P4P scoring methods on hospital rankings. We are addressing the following questions:

• How do select groups of hospitals (safety net, teaching, rural, and high minority service) fare under the current scoring method in terms of quality rankings?

• What are quality-scoring alternatives and how would they affect the rankings for select groups of hospitals? Of particular interest are methods that measure “perfect” or “complete” scores, computed as the fraction of patients receiving optimal (that is, all indicated) care; “point-of-care” process scores (for example, care at admission, discharge care and planning), rather than measures that are condition specific; performance-improvement scores (rather than scores based on meeting performance cutoffs in a given year); and disparity scores, computed as differences in quality scores between different types of patients.

• What are variations in quality by patient characteristics, including race, gender, insurance status, region, and socioeconomic status of patient zip code?