Surgical Left Atrial Appendage Occlusion During Cardiac Surgery in Atrial Fibrillation: A Contemporary Systematic Review and Meta-Analysis

Central Illustration. LAAO vs no LAAO during cardiac surgery in patients with AF. AF, atrial fibrillation; LAAO, left atrial appendage occlusion; MI, myocardial infarction; OR, odds ratio; RCT, randomized controlled trial.

Figure 1. Flow diagram of included patient population.

Figure 2. Association between LAAO and all-cause mortality. Forest plots comparing all-cause mortality in patients who underwent LAAO with that in those who did not undergo LAAO, stratified by statistical model. ORs with 95% CIs are displayed for each individual study and the overall pooled effect. (A) Fixed-effects Mantel-Haenszel model; (B) random-effects REML model. LAAO, left atrial appendage occlusion; REML, restricted maximum likelihood.

Figure 3. Association between LAAO and risk of ischemic stroke. Forest plots comparing ischemic stroke in patients who underwent LAAO with that in those who did not undergo LAAO, stratified by statistical model. ORs with 95% CIs are shown for each individual study and the overall pooled effect. (A) Fixed-effects Mantel-Haenszel model; (B) random-effects REML model. LAAO, left atrial appendage occlusion; OR, odds ratio; REML, restricted maximum likelihood.

Figure 4. Contour-enhanced funnel plot for ischemic stroke. Contour-enhanced funnel plot assessing publication bias across studies reporting ischemic stroke outcomes in patients who underwent LAAO with those in patients who did not undergo LAAO. Each dot represents an individual study plotted by its standard error and log odds ratio. Shaded regions represent different significance thresholds. No significant publication bias was detected based on Egger test (P = .966) and Begg test (P = .974). LAAO, left atrial appendage occlusion; REML, restricted maximum likelihood.

Figure 5. Leave-1-out sensitivity analysis for ischemic stroke. Sensitivity analysis evaluating the influence of each individual study on the pooled effect estimate for ischemic stroke using a leave-1-out approach under a random-effects REML model. Each point represents the recalculated OR and 95% CI after sequential omission of 1 study. Consistent results across iterations suggest robustness of the overall findings. OR, odds ratio; REML, restricted maximum likelihood.

Figure 6. Leave-1-out sensitivity analysis for all-cause mortality. Sensitivity analysis assessing the impact of individual studies on the pooled effect estimate for all-cause mortality using a leave-1-out method under a random-effects REML model. The ORs and 95% CIs were recalculated after omitting 1 study at a time. The result remained stable except when Hawkins et al30 was excluded, indicating its influence on the overall findings. OR, odds ratio; REML, restricted maximum likelihood.

Figure 7. Subgroup analysis of all-cause mortality based on year of publication. Forest plot comparing all-cause mortality in patients with LAAO vs that in patients without LAAO, stratified by year of publication (before 2020 vs 2020 and after), using a random-effects REML model. ORs with 95% CIs are presented for each subgroup and overall. No significant mortality benefit was observed in either subgroup, and the test for subgroup differences was not statistically significant (P = .72). LAAO, left atrial appendage occlusion; OR, odds ratio; REML, restricted maximum likelihood.

Figure 8. Subgroup analysis of all-cause mortality based on study design. Forest plot comparing all-cause mortality in patients with LAAO vs that in patients without LAAO, stratified by study type (RCTs vs observational studies), using a random-effects REML model. ORs with 95% CIs are shown for each subgroup and the overall population. No significant mortality benefit was observed in either subgroup, and the test for subgroup differences was not statistically significant (P = .84). LAAO, left atrial appendage occlusion; OR, odds ratio; REML, restricted maximum likelihood.

Figure 9. Cumulative meta-analyses for ischemic stroke and all-cause mortality. Cumulative meta-analyses assessing the impact of LAAO on ischemic stroke and all-cause mortality using a random-effects REML model. Studies were sequentially added by publication year, with pooled ORs and 95% CIs recalculated at each step. (A) Ischemic stroke: a consistent and statistically significant reduction in stroke risk with LAAO was observed in recent years. (B) All-cause mortality: no consistent or statistically significant mortality benefit was observed across the study period. LAAO, left atrial appendage occlusion; OR, odds ratio; REML, restricted maximum likelihood.