Evaluating individual hospital quality through outcome statistics

Case abstract data are routinely collected by hospital abstracting services, peer review organizations, and some state agencies. These data have proved invaluable in the analysis of patterns of performance across large numbers of hospitals and have shown, for example, the inverse relation between diagnosis- or procedure-specific volume and outcome. Routinely collected data also appear to be an attractive means for identifying hospitals, and perhaps physicians, with particularly good or poor outcomes for their patients. Unfortunately, problems of small numbers of patients and relatively low rates of poor outcomes make it difficult to be confident in the identification of individual performers. Recent data for cardiac catheterization patients are used to illustrate this problem.