Comparative risk of major bleeding with rivaroxaban and warfarin: Population‐based cohort study of unprovoked venous thromboembolism

OBJECTIVES: To assess the association between rivaroxaban and warfarin and major bleeding risk in unprovoked venous thromboembolism (VTE) patients. METHODS: Using US MarketScan claims from 1/2012-12/2016, we identified patients who had ≥1 primary hospitalization/emergency department visit diagnosis code for an unprovoked VTE, newly initiated on rivaroxaban or warfarin within 30 days after the VTE and ≥12 months of insurance coverage prior to the VTE. Differences in baseline covariates were adjusted using inverse-probability-of-treatment weights based on propensity scores (residual absolute standardized differences <0.1 achieved for all covariates). Endpoints included any major, gastrointestinal, genitourinary, intracranial, and other bleeds. Patients were followed for up to 12 months or until endpoint occurrence, index oral anticoagulant discontinuation/switch, insurance disenrollment or end of follow-up. RESULTS: We identified 10 489 rivaroxaban and 26 364 warfarin patients with an unprovoked VTE. Upon Cox regression, rivaroxaban reduced patients' hazard of major bleeding by 27% (95% confidence interval [CI] = 8%-42%), gastrointestinal bleeding by 38% (95% CI = 14%-55%), and intracranial hemorrhage by 81% (95% CI = 41%-99%) vs warfarin. No subtype of major bleeding occurred statistically more often in rivaroxaban vs warfarin-treated patients. CONCLUSIONS: Rivaroxaban was associated with a reduced risk of overall, gastrointestinal, and intracranial major bleeding vs warfarin in unprovoked VTE. No bleeding subtype was significantly more frequent in rivaroxaban patients.