Efficacy and safety of left atrial appendage closure versus oral anticoagulation in atrial fibrillation: a two-part randomised-trial synthesis and a pooled analysis

Background/Introduction Percutaneous left atrial appendage closure (LAAC) is a non-pharmacological alternative to oral anticoagulation (OAC) for stroke prevention in atrial fibrillation (AF). However, randomised trials span different therapeutic eras-comparing LAAC to warfarin or direct oral anticoagulants (DOACs) and distinct clinical contexts, making a single composite interpretation of "net clinical benefit" challenging. Purpose To synthesise randomised evidence using a domain-specific framework that separately evaluates efficacy and safety across different therapeutic eras and clinical settings. Methods We conducted a systematic review and two-part meta-analysis of four randomised controlled trials (PROTECT AF, PREVAIL, PRAGUE-17, and OPTION) totalling 3,116 patients. Part 1 utilised conventional random-effects pooling of relative risks (RRs). Part 2 applied a hierarchical Bayesian synthesis to separate the efficacy domain (thromboembolism/death) from the safety domain (clinically significant bleeding). Results Pooled efficacy outcomes showed no significant difference between LAAC and OAC (RR 0.79; 95% CI 0.58–1.08; I2=6%). Directional heterogeneity was noted across trials based on comparator era. In the safety domain, the primary pooled estimate was imprecise (RR 0.74; 95% CI 0.37–1.47; I2=86%), reflecting non-exchangeable bleeding definitions. However, in the contemporary post-ablation/DOAC-era setting (OPTION trial), LAAC significantly reduced non-procedural major or clinically relevant nonmajor bleeding at 36 months (RR 0.47; 95% CI 0.36–0.62). Early device-related events occurred in 2.2% of patients within 7 days in the PREVAIL trial. Conclusion LAAC provides stroke prevention comparable to OAC. Safety advantages are context-dependent, with significant bleeding reductions observed in contemporary DOAC-era and post-ablation populations, though these are offset by early procedural risks. A domain-specific interpretation better reflects net clinical benefit than a single pooled composite across heterogeneous trials.Forest plot 2.1: Efficacy endpointsForest plot 3.1: Safety endpoints