Five RCTs (23,063 patients, range 246 to 14,264). Overall the RCTs were considered to be at a low risk for bias; four of the RCTs were reported as open-label but had blinded outcome assessment. Median treatment duration ranged from 84 to 590 days. Follow-up ranged from 84 to 707 days.
Across all five RCTs the proportion of patients with the composite of major and clinically relevant non-major bleeding was not significantly different between rivaroxaban (5.9% to 20.7%) and vitamin K antagonists (6.3% to 20.3%); RR 0.99 (95% CI 0.93 to 1.06; Ι²=2.9%). Rivaroxaban was associated with a significant decrease in fatal bleeding (RR 0.48, 95% CI 0.31 to 0.74; NNT=500; Ι² 0%; four RCTs) but not non-major bleeding (four RCTs) or all-cause mortality (four RCTs).
Further results for individual trials and sensitivity analyses were presented.