Different oral anticoagulants may be associated with gastrointestinal bleeding (GIB) from different locations or mucosal lesions. We aimed to test this hypothesis. Two blinded gastroenterologists independently analyzed source documents from the randomized evaluation of long-term anticoagulant therapy (RE-LY) trial of dabigatran 150 mg BID (D150), dabigatran 110 mg BID (D110) versus warfarin in non-valvular atrial fibrillation (NVAF). Major GIB events (total n = 546) and life-threatening GIB events (n = 258) were more common with D150 versus warfarin (RR 1.57 [1.28-1.92] and RR 1.62 [1.20-2.18], respectively) and similar for D110 compared to warfarin (RR 1.11 [0.89-1.38] and RR 1.16 [0.84-1.61], respectively). Fatal bleeding was similarly rare across treatment groups. Lower GI major bleeding and life-threatening bleeding were more common with D150 compared to warfarin (RR 2.23 [1.47, 3.38] and RR 2.64 [1.36, 5.13], respectively) and with D110 compared to warfarin (RR 1.78 [1.16, 2.75] and RR 2.00 [1.00, 4.00], respectively). MGIB from colonic angiodysplasia was increased with dabigatran versus warfarin (P < 0.01 for both dose comparisons). Subacute and chronic MGIB events were more common with D150 than with warfarin (RR 1.72 [1.06, 2.78] and RR 1.66 [1.12, 2.45], respectively), as were hematochezia or melena (RR 1.67 [1.18, 2.36] and RR 1.72 [1.20, 2.47], respectively). In a chronic NVAF population, D150 but not D110 is associated with increased major and life-threatening GI bleeding in comparison with warfarin. At both dabigatran doses, increased bleeding from the colorectum, in particular from angiodysplasia, is seen.
