Abstract 70: Risk of Hemorrhage in Veterans Who Switch to Dabigatran for Stroke Prevention in Atrial Fibrillation

Abstract

BACKGROUND: The REL-Y trial found similar rates of hemorrhage for dabigatran and warfarin in patients with atrial fibrillation (AF). These findings have not been evaluated in real world clinical settings. OBJECTIVE: We used national administrative Veterans Affairs (VA) data from 2011-2012 to evaluate the risks of any bleeding, intracranial bleeding, and gastrointestinal bleeding for patients who switched to dabigatran after at least 6 months on warfarin, compared to patients remained on warfarin. DATA: We identified 93,950 patients with AF who had been on warfarin for at least 180 days during the data period, including 1419 (1.5%) who subsequently received dabigatran. Baseline and time-dependent patient characteristics identified included comorbid conditions, CHADS2 stroke risk score, HAS-BLED bleeding risk score, and INR time in therapeutic range. For each type of bleed, the date of the first bleeding event after June 2011 was identified. The use of dabigatran or warfarin during the follow up period was identified using VA National Pharmacy Extracts. ANALYSIS: Marginal structural models (MSM) were used to determine the hazard of bleeding with dabigatran relative to warfarin. MSM use weights to reduce bias in observational data caused by nonrandom censoring or treatment assignment. The weighted logistic regression models were fit using generalized estimating equations, and reflected baseline and time dependent covariates (e.g., HAS-BLED, CHADS2), weekly indicators of anticoagulant type (warfarin or dabigatran), and censoring events (e.g., ceasing anticoagulation therapy, death). RESULTS: Compared to patients who never used dabigatran, patients who used dabigatran at least once were younger, more likely to be white, had lower INR time in therapeutic range on warfarin, lower CHADS2 scores, and similar HAS-BLED scores. A total of 21,083 patients experienced bleeding events, including 129 events after dabigatran use. Risk adjusted rates of any bleeding and intracranial bleeding were similar for warfarin and dabigatran. However, the likelihood of gastrointestinal bleeding was 70% higher with dabigatran use. CONCLUSIONS: Dabigatran may increase the likelihood of gastrointestinal bleeds, but is not associated with other bleeding.