Prediction of major and clinically relevant bleeding in patients with VTE treated with edoxaban or vitamin K antagonists.

Better understanding of risk factors for major bleeding events during anticoagulant treatment for venous thromboembolism (VTE) may help physicians when deciding on intensity and duration of treatment. The primary aim of this study was to identify risk factors for major and clinically relevant bleeding in patients receiving the oral factor Xa inhibitor edoxaban or warfarin for the treatment of acute VTE. We analysed data from 8240 patients who received ≥1 dose of study drug in the Hokusai-VTE study. Bleeding risk factors were evaluated in 4118 patients who received edoxaban and significant variables were combined in a prediction model. We used the C-statistic to estimate model discrimination and bootstrap techniques for internal validation. Major bleeding occurred in 56/4118 (1.4 %) patients given edoxaban and in 66/4122 (1.6 %) patients given warfarin. Clinically relevant bleeding occurred in 349 (8.5 %) and 423 (10.3 %), respectively. Significant risk factors for major bleeding during edoxaban treatment were female sex, concomitant antiplatelet therapy, haemoglobin ≤10 g/dl, history of arterial hypertension, and systolic blood pressure >160 mmHg. The discrimination of the model was high (C-statistic: 0.71) for major bleeding, lower for clinically relevant bleeding (C-statistic: 0.62) and when the model was applied to patients receiving warfarin (C-statistic 0.60). In conclusion, we identified five main predictors of major bleeding in patients receiving edoxaban for the treatment of acute VTE. A risk model based on these factors predicted an increased risk of bleeding with good discrimination.