Prediction of Bleeding Risk in Patients on Extended Oral Anticoagulation for Venous Thromboembolism

Background: While tools exist to predict the risk of major bleeding in patients on oral anticoagulation therapy (OAT) for venous thromboembolism (VTE), these have focused on the higher risk period during the first 3 months of OAT therapy (risk of major bleeding 2.4% in first 3 months vs. 2% per year thereafter), and have not been developed or evaluated for bleeding risk after the first 3 months of therapy (extended OAT). It is widely considered that a bleeding risk tool that is able to identify patients with an annual rate of major bleeding over 3% is needed, as this is the cut-point at which the risk of continued anticoagulant therapy exceeds the benefit.

Aims: We sought to evaluate if the rates of major bleeding were over 3% when four existing, previously published tools (RIETE,outpatient bleeding risk index (OBRI), ACCP and HAS-BLED) classified VTE patients during extended OAT as being high risk for major bleeding, and to assess if the differences in major bleeding risk between high risk and not high risk patients were statistically significantly different.

Methods: The Bleeding Risk study was a multicentre, multinational prospective cohort study of patients on extended OAT for unprovoked VTE, or provoked VTE with prior VTE, designed to generate a new prediction tool for major bleeding. Patients were enrolled after at least 3 months of OAT. All major bleeding events during long term OAT were captured and adjudicated. We applied each of the 4 tools above to determine the risk for major bleeding according to patient scores. Not all variables in these tools were collected, including drug and alcohol history, and INR control.

Results: 2514 patients enrolled at 12 sites have contributed over 7000 years of observation. The mean patient age was 60.2±14.7 years, 64% were male, 92% Caucasian, average BMI was 31.3, and 9% of patients were on antiplatelet agents. Patients were followed for a mean of 2.8 years (range, 0.1 to 6.8 years. 121 patients (4.8%) experienced at least one episode of major bleeding. The annual rate of bleeding was 1.7 per 100 patient years of observation.

The proportion of patients classified as high risk of major bleeding by the RIETE (score of 3 or 4), OBRI (score of 3), ACCP (score ≥ 2) and HAS-BLED (score ≥ 3) scores were 12.3%, 13.3%, 28.0% and 23.1%, with major bleeding rates of 3.9%, 3.2%, 3.4% and 3.4% per year, respectively. The major bleeding rates of patients who were classified as not high risk of bleeding were 1.4%%, 1.5%, 1.1% and 1.2% per year, respectively. All differences were statistically significant with p values < 0.0001.

Conclusion: Despite the potential to underestimate risk due to missing variables, all currently available prediction tools are able to identify patients with a 3% or higher risk of major bleeding per year. , The HAS-BLED and ACCP scores were able to identify the highest proportions of patients as high risk for major bleeding.