Mixed Treatment Comparison Meta-Analysis of Aspirin, Warfarin, and New Anticoagulants for Stroke Prevention in Patients With Nonvalvular Atrial Fibrillation

Background: Warfarin and aspirin are used to prevent stroke in patients with atrial fibrillation (AF). There are inherent challenges with both treatments, including variable and inconsistent benefit and increased bleeding risks. The availability of new anticoagulants offers some alternatives. Objective: A mixed treatment comparison meta-analysis to evaluate direct and indirect treatment data including aspirin, warfarin meta-analysis, dabigatran, edoxaban, and rivaroxaban for the prevention of primary or secondary stroke in patients with AF. Methods: A comprehensive, systematic literature search was conducted to identify randomized trials comparing aspirin, warfarin, apixaban, dabigatran, edoxaban, and rivaroxaban in patients with AF requiring treatment for stroke prevention. Open-label and blinded designs were included if they evaluated any stroke or any bleeding event. Data on stroke and bleeding events were abstracted, verified, evaluated, scored, and entered into Aggregate Data Drug Information System version 1.16 to generate a mixed treatment comparison meta-analysis. Direct and indirect comparisons were evaluated, and we looked for inconsistency in closed loop structures. Data are reported as rate ratios with 95% credible intervals. In addition, we reviewed variance statistics and explored variance with node-splitting models. Results: Our literature search yielded 30 articles, 21 of which were included. All treatments except aspirin reduced the risk of any stroke compared with placebo. Warfarin (0.43 [0.33-0.57]), apixaban (0.37 [0.27-0.54]), dabigatran (0.34 [0.21-0.57]), rivaroxaban (0.36 [0.22-0.60]), and aspirin with clopidogrel (0.73 [0.53-0.99]) were more protective than aspirin alone. Warfarin and the new anticoagulants were similar in the reduction of stroke, vascular death, and mortality. There was no difference in major bleeding between any treatment group. There were more nonmajor bleeding events when comparing warfarin and apixaban (1.83 [1.05-4.03]); no other differences between warfarin and the other new anticoagulants were found. Conclusions: This mixed treatment comparison meta-analysis found similarity between warfarin and the new anticoagulants with the exception of one comparison, in which warfarin was associated with more non-major bleeding than apixaban. Thus, the new anticoagulants are therapeutically comparable when warfarin is inappropriate. (C) 2013 Elsevier HS Journals, Inc. All rights reserved.