Introduction The non-vitamin k antagonist (VKA) oral anticoagulants (NOAC) comprise inhibitors of thrombin and factor Xa. They are being increasingly used in patients requiring long-term anticoagulation due to non-inferiority and relative ease of use when compared to VKA and low molecular weight heparin (LMWH). Prospective randomised studies testing the efficacy of NOACs have shown a possible increased risk of gastrointestinal (GI) bleeding. To clarify the risk of GI bleeding we performed a systematic review and Bayesian network meta-analysis.
Methods We followed a pre-specified and peer reviewed PRISMA extension guidelines and checklist for reporting systematic reviews and network meta-analyses. A comprehensive search of the literature was performed to identify any study, prospective or retrospective, which compared NOAC with VKA or LMWH and reported on GI bleeding. Studies on all indications for these medications were included. We did not include studies which compared against placebo or antiplatelet medication. We performed a Bayesian random effects regression model. To estimate the effect of NOAC on GI bleeding we calculated incidence rate ratios (IRR) based on the number of patient years exposed to the medication. We grouped medications together by mechanism of action; VKA, LMWH, thrombin inhibitor or factor Xa inhibitor. Pre-planned subgroup analyses were performed on; indication, population, study design, GI bleed definition and NOAC dose.
Results We identified 35 studies reporting on 481, 534 patients exposed to 405, 026 patient years of anticoagulant medication. The overall GI bleeding incidence was 1.3 events per 100 patient years. We found no difference in the incidence rate of major GI bleeding when comparing NOAC medications with VKA and LMWH. This results was sustained on sensitivity analyses comparing observational and prospective trial data, the indication for anticoagulation and prophylactic versus therapeutic doses. When analysing all severities of GI bleeding, we found a statistically significant reduction when comparing Xa inhibitors with VKA (IRR 0.20, 95% CrI 0.05–0.65), and thrombin inhibitors (IRR 0.20, 95% CrI 0.05–0.67) respectively.
Conclusion We have shown that there is no difference in the risk of major GI bleeding when comparing NOAC, VKA and LMWH anticoagulation medications. There is a significant reduction in all GI bleeding events with the use of factor Xa inhibitors when compared to VKA and direct thrombin inhibitors.
Disclosure of Interest None Declared