Bayesian Interpretation of Non-Inferiority in TAVI Versus SAVR Trials: A Systematic Review and Meta-Analysis

Non-inferiority study designs are frequently used in randomized controlled trials (RCT) comparing transcatheter aortic valve replacement (TAVR) and surgical aortic valve replacement (SAVR). However, the long-term outcomes of TAVR are unknown, and the concept of non-inferiority is challenging to define and assess. This systematic review and meta-analysis published in Interdisciplinary Cardiovascular and Thoracic Surgery (ICVTS) compares TAVR and SAVR, specifically emphasizing the non-inferiority margin for five-year all-cause mortality. To this end, the authors performed a systematic search in three electronic databases. A total of eight RCTs (n = 8,698 patients) comparing TAVR and SAVR were included. Bayesian methods were implemented to evaluate the posterior probability of non-inferiority at different trial non-inferiority margins. The primary study outcomes were five-year actuarial all-cause mortality and the probability of non-inferiority at various transformed trial non-inferiority margins. Kaplan-Meier-derived five-year survival was 61.6 percent (95 percent CI 59.8–63.5 percent) for TAVR and 63.7 percent (95 percent CI 61.9–65.6 percent) for SAVR. The posterior median relative risk for all-cause mortality of TAVR was 1.14 compared to SAVR. The posterior probability of a mortality difference in favor of SAVR was 99.9 percent. The authors concluded that the non-inferiority of TAVR compared to SAVR is unlikely after five years in terms of all-cause mortality.