Identifying Top-Performing Hospitals by Algorithm: Results from a Demonstration Project

Background: Because of the move toward performance-based reimbursement, identification of top-performing hospitals has acquired new importance. Methods: The High Performance Algorithm (HPA) for hospitals was developed on the basis of the following principles: (1) the approach must be data driven and transparent, (2) all hospitals providing the same service are held to the same standard, (3) top-performing hospitals must perform well on easily achieved and difficult quality measures, and (4) high performance demands sustained excellence over time. The HPA algorithm was applied to 16 quality measures from the national Hospital Quality Alliance (July 2003-June 2004) for acute myocardial infarction (AMI), heart failure (HF), and pneumonia. Top-performing hospitals were defined as those with the top 1% of HPA scores (n = 45). Results: From all 3,867 hospitals, median HPA scores (interquartile range) were 17.0 (16.0-19.0) for top-performing hospitals and 3.0 (1.0-6.0) for others (p <.001). Mean performance on quality measures was higher for top hospitals on all 16 measures. For example, on administration of angiotensin-converting enzyme inhibitors to patients with HF, the mean score for top-performing hospitals was 93.3%, compared with 76.5% for others (p <.001). Although many hospitals achieved excellence on individual measures, sustained top performance across multiple conditions and time periods was uncommon, with < 1% of hospitals scoring > 16/36 points on the HPA scale. Discussion: Using national, publicly reported data, the HPA provided good discrimination between top-performing and other hospitals. This project sets the stage for future comparisons of organizational, leadership, and policy differences between top-performing and other hospitals.