Systematic review and meta-analysis of randomized controlled trials on safety and effectiveness of oral anticoagulants for atrial fibrillation in older people.

A meta-analysis of prospective, randomized controlled trials on novel oral anticoagulants (NOACs) versus warfarin, as most commonly used vitamin K antagonists (VKAs), was done to evaluate their effect on stroke risk and bleeding complications in nonvalvular atrial fibrillation (AF) patients. The study aims to evaluate efficacy and safety of NOACs versus warfarin between patients < 75 years and ≥ 75 years old. Prospective, randomized controlled trials (RCTs) comparing NOACs with warfarin with at least 1-year follow-up in nonvalvular AF patients were included. Search was done at MEDLINE, without time and language restriction. "Cochrane risk of bias 2.0 tool" was used to assess risk of bias. In meta-analysis, random effect model was used. Q statistics was used to assess heterogeneity where it was indicated and I2 was used to assess inconsistency across studies. Five studies as RELY, ARISTOTLE, ENGAGE-AF, ROCKET-AF, and J-ROCKET-AF were included. Overall risk of bias was low for trials. Total patient number in meta-analysis was 73,122. Mean age was between 70 and 73 as a whole. Follow-up period was between 499 days and 2.8 years. Analysis of major stroke and systemic embolism was favoring NOACs than warfarin in older patients group period (RR (95% CI) = 0.57 (0.42, 0.76)). Similar trend was seen in younger patient group without statistical significant (RR (95% CI) = 0.74 (0.43, 1.27)). Analysis of major bleeding showed a general statistically significant trend of occurrence of major bleeding in whole group (RR (95% CI) = 0.70 (0.58, 0.86)) with NOACs. In subgroup analysis, it was consistent for both younger and older patients. Older patients have higher stroke risk that leads to more benefit with anticoagulation than younger patients. NOACs have better efficacy profile than warfarin as they decrease stroke more than warfarin, without increasing bleeding risk. Need for real-world data in old population is necessary.