National Trends of Hospital Performance in Acute Myocardial Infarction Care: Department of Veterans Affairs, 2011-2014.

Abstract

Rates of mortality and readmission, both within 30 days of an admission with an acute myocardial infarction (AMI), are key measures of hospital performance. A recent report demonstrated that Veterans Administration (VA) hospitals had lower AMI mortality but higher readmission rates at 30 days compared with Medicare fee for service hospitals from 2010 to 2013.1 Although this information provides an overall assessment of the quality of care provided, it is also important to understand facility-level and temporal variations of these performance metrics to inform ongoing quality improvement efforts. Accordingly, we sought to characterize contemporary facility-level variations (ie, variation across hospitals not explained by variation in patient risk) in 30-day risk-standardized mortality rates (RSMRs) and 30-day risk-standardized readmission rates (RSRRs) across the VA, to identify underperforming or overperforming sites, and to assess temporal variations in these metrics. Data from the VA External Peer Review Program, in which trained staff abstracted clinical data from comprehensive chart reviews, were linked with the VA Corporate Data Warehouse—a source of inpatient and outpatient administrative data inclusive of vital status. Patients who were admitted to a VA hospital from January 1, 2011, to February 28, 2014, and assigned an International Classification of Diseases, Ninth Revision: Clinical Modification diagnosis code of 410.xx irrespective of age were included. Those transferred into the VA system were excluded. If a patient had multiple AMIs in the study time period, 1 AMI was randomly chosen for inclusion in the study. Outcome measures of interest were 30-day all-cause mortality and 30-day unplanned, all-cause readmission. The latter included VA and non-VA hospital readmissions for which the VA was billed. Hospitals with <20 AMI admissions during the study period were excluded. To assess outcome variation across hospitals, hospital-specific RSMRs and RSRRs were estimated using hierarchical log-binomial regression models to account for the clustering …