Derivation and validation of a novel bleeding risk score for elderly patients with venous thromboembolism on extended anticoagulation

Summary Existing clinical scores do not perform well in predicting bleeding in elderly patients with acute venous thromboembolism (VTE). We sought to derive an easy-to-use clinical score to help physicians identify elderly patients with VTE who are at high-risk of bleeding during extended anticoagulation (>3 months). Our derivation sample included 743 patients aged ≥65 years with VTE who were enrolled in a prospective multicenter cohort study. All patients received extended anticoagulation with vitamin K antagonists. We derived our score using competing risk regression, with the time to a first major bleeding up to 36 months of extended anticoagulation as the outcome, and 17 candidate variables as predictors. We used bootstrapping methods for internal validation. Sixty-six (9%) patients suffered major bleeding. The clinical score is based on seven clinical factors (previous bleeding, active cancer, low physical activity, anemia, thrombocytopenia, antiplatelet drugs/NSAIDs, and poor INR control). Overall, 48% of patients were classified as low-risk, 37% as moderate-risk, and 15% as high-risk of bleeding. The rate of major bleeding was 1.4 events in low-risk, 5.0 events in moderate-risk, and 12.2 events per 100 patientyears in high-risk patients. The c-statistic was 0.78 at 3 months and 0.71 at 36 months of extended anticoagulation. Model calibration was excellent (p=0.93). Internal validation showed similar results. This simple clinical score accurately identified elderly patients with VTE who are at high risk of major bleeding and who may not benefit from extended anticoagulation. Further validation of the score is important before its implementation into practice. The study is registered to https://clinicaltrials.gov as NCT00973596. This work was carried out at the Department of General Internal Medicine in the Bern University Hospital, Switzerland.