The study aim was to identify predictive factors for major bleeding in patients receiving the novel oral factor Xa inhibitor rivaroxaban or enoxaparin-vitamin K antagonists (VKAs) for the treatment of acute symptomatic venous thromboembolism. We analysed data from patients included in the phase III EINSTEIN DVT and EINSTEIN PE studies. Factors associated with major bleeding events were assessed with best subset variable selection using Cox proportional hazards regression model. Three time windows were considered, i.e. the initial three weeks, after the third week onwards, and the entire duration of the anticoagulant treatment. Model discrimination was estimated using the C-statistic and validated internally by bootstrap techniques. Major bleeding occurred in 40 (1.0%) of 4130 patients receiving rivaroxaban and in 72 (1.7%) of 4116 receiving enoxaparin/VKAs, with 44% of the major bleeding events occurring in the first three weeks of treatment. Significant risk factors for major bleeding were older age, black race, low haemoglobin concentrations, active cancer, and antiplatelet or non-steroidal anti-inflammatory drug therapy. The discrimination of the model for major bleeding was high for the first three weeks (C-statistic 0.73), from the fourth week onwards (C-statistic 0.68), and the entire period of anticoagulant treatment (C-statistic 0.74). This analysis identified risk factors for major bleeding in patients receiving the novel oral anticoagulant rivaroxaban or enoxaparin/VKAs for the treatment of acute venous thromboembolism. The prognostic model based on the combination of identified risk factors may be informative to estimate the risk of major bleeding both during the initial and later phases of anticoagulation.
Keywords: Haemorrhage; anticoagulants; randomised controlled trial; risk factors; thrombosis.