Oral anticoagulation versus antiplatelet or placebo for stroke prevention in patients with heart failure and sinus rhythm: Systematic review and meta-analysis of randomized controlled trials

Background Previous meta-analyses of randomized controlled trials of oral anticoagulation in patients with heart failure and sinus rhythm reported reduced stroke risk and increased bleeding risk compared to antiplatelets or placebo. However, the effect estimates may be subject to imprecision, as all included trials were prematurely terminated; stroke was not the primary outcome and overall results were primarily driven by a single trial. Recently, new trial data became available. Aim To provide more accurate estimates of the effect of oral anticoagulation on stroke risk in heart failure patients with sinus rhythm by systematic review and meta-analysis of available randomized controlled trials including recently published evidence. Methods We searched PubMed and Scopus for full-text articles of randomized controlled trials of oral anticoagulation versus antiplatelet or placebo in heart failure patients with sinus rhythm published between inception and 28 August 2018. The outcomes assessed were any stroke, major bleeding, and death. Results In five trials (9490 patients; 21,067 patient-years), oral anticoagulation-treated patients had lower stroke risk (odds ratio (OR) 0.60, 95%CI: 0.46-0.78, absolute-risk-reduction: 1.3%, number-needed-to-treat: 77), higher major bleeding risk (OR: 1.92, 95%CI: 1.51-2.45, absolute-risk-increase: 2.0%, number-needed-to-harm: 50), and no significant difference in death rates (OR: 0.90, 95%CI: 0.73-1.11) compared to antiplatelets or placebo. Conclusions In the largest meta-analysis to date, oral anticoagulation is associated with a considerable reduction of stroke risk, which is offset by a significant increase in major bleeding risk. For every 1000 patients treated with oral anticoagulation rather than antiplatelet or no antithrombotic treatment for 2.21 years, 13 strokes are prevented but 20 additional major hemorrhages occur, without significant difference in death rates.