Comparative evaluation of left atrial appendage occlusion and oral anticoagulation: a Bayesian meta-analysis of randomized controlled trials

Randomized trials comparing percutaneous left atrial appendage occlusion (LAAO) with oral anticoagulation (OAC) for stroke prevention in patients with atrial fibrillation have yielded conflicting results across device generations and anticoagulant eras. This Bayesian meta-analysis of randomized trials aimed to compare LAAO vs. OAC for stroke or systemic embolism (SSE) and non-procedural clinically relevant bleeding (CRB).

We conducted a systematic search across five databases (PubMed, Embase, Scopus, Web of Science, and Cochrane CENTRAL) to identify randomized trials of LAAO vs. OAC. Primary outcomes of stroke or systemic embolism, non-procedural CRB and clinical benefit composite were estimated using a Bayesian hierarchical random effects model and findings confirmed across robust sensitivity and subgroup analyses. Posterior probabilities for superiority, non-inferiority and equivalence were estimated using clinically meaningful margins. Six randomized trials including 7004 participants in the intention-to-treat groups, were analysed. LAAO was not associated with lower SSE vs. OAC (pooled RR, 1.11; 95% CrI, 0.80–1.49), with a 74.6% posterior probability of favouring OAC. The posterior probability of crossing non-inferiority was P(HR < 1.4) of 89%. In contrast, LAAO reduced non-procedural CRB (pooled RR, 0.59; 95% CrI, 0.45–0.78). The net clinical benefit composite, pooling contemporary trials, demonstrated a pooled RR of 0.90 (95% CrI 0.51–1.50), with P(RR < 1) of 76%, with substantial heterogeneity for this outcome.

This Bayesian meta-analysis found no clear evidence supporting clinically meaningful non-inferiority of LAAO over OAC for stroke prevention. Despite reduced non-procedural bleeding, LAAO did not demonstrate a similar reduction in major bleeding.