Hemorrhage risk of direct oral anticoagulants in real-world venous thromboembolism patients.

INTRODUCTION:Venous thromboembolism (VTE) management increasingly involves anticoagulation with direct oral anticoagulants (DOACs). Few studies have used competing-risks analyses to ascertain the mortality-adjusted hemorrhage and recurrent VTE (rVTE) risk of individual DOACs. Furthermore, hemorrhage risk factors in patients treated with apixaban remain underexplored. MATERIALS AND METHODS:Patients diagnosed with VTE receiving anticoagulation were identified from the Optum Clinformatics Data Mart (2003-2019). Study endpoints included readmissions for intracranial hemorrhage (ICH), non-intracranial hemorrhage (non-ICH hemorrhage), and rVTE. Coarsened exact matching was used to balance baseline clinical characteristics. Complication incidence was evaluated using a competing-risks framework. We additionally modeled hemorrhage risk in apixaban-treated patients. RESULTS:Overall, 225,559 patients were included, of whom 34,201 received apixaban and 46,007 received rivaroxaban. Compared to rivaroxaban, apixaban was associated with decreased non-ICH hemorrhage (sHR = 0.560, 95%CI = 0.423-0.741), but not ICH, and rVTE (sHR = 0.802, 95%CI = 0.651-0.988) risk. This was primarily in emergent readmissions (sHR[emergent hemorrhage] = 0.515, 95%CI = 0.372-0.711; sHR[emergent rVTE] = 0.636, 95%CI = 0.488-0.830). Contributors to emergent hemorrhage in apixaban-treated patients include older age (sHR = 1.025, 95%CI = 1.011-1.039), female sex (sHR = 1.662, 95%CI = 1.252-2.207), prior prescription antiplatelet therapy (sHR = 1.591, 95%CI = 1.130-2.241), and complicated hypertension (sHR = 1.936, 95%CI = 1.134-3.307). Patients anticipated to be "high-risk" experienced elevated ICH (sHR = 3.396, 95%CI = 1.375-8.388) and non-ICH hemorrhage (sHR = 3.683, 95%CI = 2.957-4.588) incidence. CONCLUSIONS:In patients with VTE receiving anticoagulation, apixaban was associated with reduced non-ICH hemorrhage and rVTE risk, compared to rivaroxaban. Risk reduction was restricted to emergent readmissions. We present a risk-stratification approach to predict hemorrhage in patients receiving apixaban, potentially guiding future clinical decision-making.