Percutaneous Left Atrial Appendage Occlusion Versus Oral Anticoagulation in Atrial Fibrillation: A Systematic Review and Bayesian Meta-analysis

BACKGROUND: Conflicting randomized trial results have generated uncertainty regarding percutaneous left atrial appendage occlusion (LAAO) vs oral anticoagulation (OAC) for stroke prevention in atrial fibrillation (AF). Thus, the objective of this study is to estimate the posterior probability that LAAO confers clinically meaningful excess ischemic stroke risk relative to OAC. METHODS: A Bayesian hierarchical meta-analysis was performed of RCT comparing LAAO with OAC in nonvalvular AF. The primary outcome was ischemic stroke, analyzed using Bayesian random-effects Poisson regression under a vague (primary), skeptical, and enthusiastic prior specification. RESULTS: Six RCT (7004 patients, +18,000 patient-years) were included. Ischemic stroke incidence rate ratio (IRR) was 1.34 (95% credible interval [CrI], 0.91-1.97; P [IRR > 1.0] = 94%). The probability of exceeding 0.5 and 1.0 excess events/100 patient-years was 12% and 0.5%, respectively. The effect on any-stroke was smaller (IRR, 1.09; 95% CrI, 0.76-1.56; P [IRR > 1.0] = 70%; P [absolute rate difference >0.5%/y] = 4.7%), because of a reduction in hemorrhagic stroke (IRR, 0.55; 95% CrI, 0.23-1.18; P [IRR < 1.0] = 94.5%). The decrease in hemorrhagic stroke was attenuated among trials using direct oral anticoagulants for medical management (IRR, 0.80; 95% CrI, 0.33-1.75). CONCLUSIONS: In AF, there is a high posterior probability that LAAO is associated with a higher rate of any-stroke, driven by an increase in ischemic stroke relative to OAC. However, the 95% CrI for ischemic stroke crossed unity, and the estimated absolute excess is small. This possible excess must be weighed against a substantial reduction in the risk of hemorrhagic stroke.