Fourteen RCTs (n=25,307) were included.
The authors stated that, because there was no significant difference between the results of the random-effects and fixed-effect models, they only reported the results of the fixed-effect model.
All studies included in the analysis, irrespective of the INR.
The combination of A+W had no significant effect on the risk of an MAE (OR 0.96, 95% CI: 0.90, 1.03, P=0.30), but increased the risk of MB (OR 1.77, 95% CI: 1.47, 2.13, P<0.00001; NNH 100). It was also associated with an increased risk of extracranial bleeding (OR 2.2, 95% CI: 1.64, 2.96, P<0.00001) but had no significant effect on intracranial bleeding (OR 1.37, 95% CI: 0.79, 2.37, P=0.27).
Combination therapy had no significant effect on all-cause death or nonfatal MI, although it significantly reduced the risk of nonfatal thromboembolic stroke (OR 0.81, 95% CI: 0.67, 0.97, P=0.02; NNT 100).
Analyses restricted to studies with INRs of between 2 and 3.
The combination of A+W significantly reduced the risk of an MAE (OR 0.73, 95% CI: 0.63, 0.84, P<0.0001; NNT 33), but increased the risk of MB (OR 2.32, 95% CI: 1.63, 3.29, P<0.00001; NNH 100). It was also associated with an increased risk of extracranial bleeding (OR 2.37, 95% CI: 1.37, 4.10, P=0.002) but had no statistically significant effect on intracranial bleeding (OR 3.02, 95% CI: 0.61, 15.02, P=0.18).
Combination therapy had no significant effect on all-cause death, but it was associated with a 57% reduction in the risk of nonfatal thromboembolic stroke (OR 0.43, 95% CI: 0.27, 0.70, P=0.0007; NNT 100) and a 30% reduction in the risk of nonfatal MI (OR 0.70, 95% CI: 0.52, 0.95, P=0.0003; NNT 50).
Thirteen of the 14 studies were of a good quality and had greater than 97% follow-up. No study was excluded because of a follow-up of less than 80%. In the sensitivity analysis, the exclusion of any trial did not significantly alter the overall results. Significant heterogeneity was found when data were pooled for MAEs irrespective of INR (P=0.001), but when the analysis was restricted to the studies with an INR of 2-3 there was no significant heterogeneity.
The funnel plot suggested the possibility of publication bias. However, Egger's test showed no publication bias for studies with an INR of 2-3 (P=0.141 for all studies and P=0.646 for studies with an INR of 2-3).