A Comparison of statistical methods for hospital performance assessment

During hospital quality improvement activities, statistical approaches are critical to help assess performance for benchmarking. Current used primarily research and reimbursement purposes. In this multiinstitutional study, these established methods were evaluated applications. Leveraging a dataset of 42,199 patients who underwent coronary artery bypass grafting surgery from 2014 2016 across 90 hospitals, six applied. The non-shrinkage were: (1) indirect standardization without effect; (2) with fixed (3) direct effect. shrinkage (4) random (5) (6) Bayesian method. Hospital related operative mortality major morbidity or was compared based on variation in adjusted rates, rankings, outliers. Method procedure volume terciles: small (< 96 cases/year), medium (96-171), large (> 171). Shrinkage reduced inter-hospital (min-max) (observed: 0%-10%; adjusted: 1.5%-2.4%) 2.6%-35%; 6.9%-17.5%). shrunk rates toward the group mean. Direct effect, resulted 16.7%-38.9% hospitals changing quintile ranking. Indirect effect no outliers among mortality, while logistic identified one three outlier hospitals. choice method greatly impacts ranking outlier’ status. These findings should be considered when benchmarking activities.