Robert E Burke, Thomas Glorioso, Anna K Barón, Peter J Kaboli, P Michael Ho

Novel measures of hospital quality are needed. Because quality improvement efforts seek to reduce variability in processes and outcomes, hospitals with higher variability in adverse events may be delivering poorer quality care. We sought to evaluate whether within-hospital variability in adverse events after a procedure might function as a quality metric that is correlated with facility-level mortality rates. We analyzed all percutaneous coronary interventions (PCIs) performed in the Veterans Health Administration (VHA) system from 2007 to 2013 to evaluate the correlation between within-hospital variability in 30-day postdischarge adverse events (readmission, emergency department visit, and repeat revascularization), and facility-level mortality rates, after adjustment for patient demographics, comorbidities, PCI indication, and PCI urgency. The study cohort included 47,567 patients at 48 VHA hospitals. The overall 30-day adverse event rate was 22.0% and 1-year mortality rate was 4.9%. The most variable sites had relative changes of 20% in 30-day rates of adverse events period-to-period. However, within-hospital variability in 30-day events was not correlated with 1-year mortality rates (correlation coefficient = .06; p = .66). Thus, measuring within-hospital variability in postdischarge adverse events may not improve identification of low-performing hospitals. Evaluation in other conditions, populations, and in relationship with other quality metrics may reveal stronger correlations with care quality.