Effect of pharmacological therapies for stroke prevention on major gastrointestinal bleeding in patients with atrial fibrillation.

Various antiplatelet and anticoagulation options are available for stroke prevention in patients with atrial fibrillation (AF). Currently, it is unclear whether these agents differ in their propensity to cause major gastrointestinal bleeding (MGIB). To our knowledge, no systematic evaluation of MGIB rates from randomised controlled trials (RCTs) of pharmacological stroke prevention in patients with AF has been conducted. Two independent investigators conducted systematic literature searches in MEDLINE and CENTRAL from the earliest possible date through November 2010. To be included, RCTs had to evaluate an adult population with AF or flutter and report data on the incidence of MGIB. Peto's odds ratios (ORs) with associated 95% confidence intervals (CIs) were calculated for all possible pair-wise comparisons of pharmacological stroke prevention alternatives. A total of 16 unique trials (n = 42,983) met inclusion criteria. The reported incidence of MGIB in placebo or control arms of identified trials was as high as 1.5%. Upon pair-wise meta-analysis of different pharmacological strategies, adjusted-dose vitamin K antagonists (VKAs) were found to be associated with a higher odds of MGIB compared with placebo/control (OR 3.21, 95% CI 1.32-7.82) and aspirin (or triflusal or indobufen) (OR 1.92, 95% CI 1.08-3.41). The addition of aspirin (or triflusal) to an adjusted-dose VKA resulted in greater odds of MGIB compared with aspirin alone (OR 4.72, 95% CI 1.35-16.49) and adjusted-dose VKA alone (OR 2.66, 95% CI 1.05-6.74). While aspirin increased the odds of MBIG by 3.23-fold compared with placebo/control, this finding did not reach statistical significance. The combination of aspirin and clopidogrel increased patients' odds of MGIB compared with aspirin alone (OR 1.93, 95% CI 1.46-2.56). Dabigatran was associated with a 30% increased odds of MGIB compared with adjusted-dose VKA (OR 1.30, 95% CI 1.06-1.59); however, ximelagatran was not. Low-intensity VKA therapy, alone or in combination with aspirin, was not associated with increased odds of MGIB compared with any (active-) comparator. The MGIB is a concern for patients with AF receiving pharmacological stroke prevention. Current RCT data suggest that dabigatran and adjusted-dose VKA therapy are associated with the highest odds of MGIB. Aspirin was not found to increase patients' odds of MGIB; however, this finding may be the result of type 2 error. Dual therapy resulting from the addition of an antiplatelet agent was typically associated with further increased odds of MGIB compared with monotherapy.