Composite Metric for Benchmarking Site Performance in Transcatheter Aortic Valve Replacement: Results From the STS/ACC TVT Registry

Background: Transcatheter aortic valve replacement (TAVR) is a transformative therapy for aortic stenosis. Despite rapid improvements in technology and techniques, serious complications remain relatively common and are not well described by single outcome measures. The purpose of this study was to determine whether there is site-level variation in TAVR outcomes in the United States using a novel 30-day composite measure. Methods: We performed a retrospective cohort study using data from the Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapies Registry to develop a novel ranked composite performance measure that incorporates mortality and serious complications. The selection and rank order of the complications for the composite was determined by their adjusted association with 1-year outcomes. Sites with risk-adjusted outcomes significantly more or less frequent than the national average based on a 95% probability interval were classified as performing worse or better than expected. Results: The development cohort consisted of 52 561 patients who underwent TAVR between January 1, 2015, and December 31, 2017. Based on associations with 1-year risk-adjusted mortality and health status, we identified 4 periprocedural complications to include in the composite risk model in addition to mortality. Ranked empirically according to severity, these included stroke, major, life-threatening or disabling bleeding, stage III acute kidney injury, and moderate or severe perivalvular regurgitation. Based on these ranked outcomes, we found that there was significant site-level variation in quality of care in TAVR in the United States. Overall, better than expected site performance was observed in 25/301 (8%) sites, performance as expected was observed in 242/301 sites (80%), and worse than expected performance was observed in 34/301 (11%) sites. Thirty-day mortality; stroke; major, life-threatening, or disabling bleeding; and moderate or severe perivalvular leak were each substantially more common in sites with worse than expected performance as compared with other sites. There was good aggregate reliability of the model. Conclusions: There are substantial variations in the quality of TAVR care received in the United States and 11% of sites were identified as providing care below the average level of performance. Further study is necessary to determine structural, process-related, and technical factors associated with high- and low-performing sites.