Bayesian interpretation of non-inferiority in transcatheter versus surgical aortic valve replacement trials. A systematic review and meta-analysis

Objectives: The concept of non-inferiority is widely adopted in randomized trials comparing transcatheter aortic valve replacement (TAVR) and surgical aortic valve replacement (SAVR). However, uncertainty exists regarding the long-term outcomes of TAVR, and non-inferiority may be difficult to assess. We performed a systematic review and meta-analysis of randomized trials comparing TAVR and SAVR, with a specific emphasis on the non-inferiority margin for 5-year all-cause mortality. Methods: A systematic search was applied to 3 electronic databases. Randomized trials comparing TAVR and SAVR were included. Bayesian methods were implemented to evaluate the posterior probability of non-inferiority at different trial non-inferiority margins under either a vague, Cauchy, or a literature-based prior. Primary outcomes were 5-year actuarial all-cause mortality, and the probability of non-inferiority at various transformed trial non-inferiority margins. Secondary outcomes were long-term survival and 1- and 2-year actuarial survival. Results: Eight trials (n = 8698 patients) were included. Kaplan-Meier-derived 5-year survival was 61.6% (95% CI 59.8-63.5%) for TAVR, and 63.7% (95% CI 61.9-65.6%) for SAVR. Six trials (n = 6370 patients) reported all-cause mortality at 5-year follow-up. Under a vague prior, the posterior median relative risk for all-cause mortality of TAVR was 1.14, compared to SAVR (95% credible interval 1.06-1.22, probability of relative risk <1.00 = 0.01%, I2 = 0%). Similar results in terms of point estimate and uncertainty measures were obtained using frequentist methods. Based on the various trial non-inferiority margins, the results of the analysis suggest that non-inferiority at 5 years is no longer likely. Conclusions: It is unlikely that TAVR is still non-inferior to SAVR at 5 years in terms of all-cause mortality.