Bleeding incidence and risk factors among cancer patients treated with anticoagulation

Abstract

Compared to age-matched controls, cancer patients have increased risk of bleeding when treated with anticoagulation. However, there are little data regarding bleeding as it relates to anticoagulant choice and other risk factors. We evaluated the six-month incidence of bleeding among patients treated with anticoagulation who had bleeding risk factors. Data were obtained from Explorys (IBM Watson, Inc.), which pools data from multiple US healthcare organizations. Cohorts of patients were created to compare bleeding events between cancer and non-cancer patients treated with anticoagulation within six months of starting anticoagulation. Potential bleeding risk factors such as cancer type, metastatic disease, obesity, chronic kidney disease stage III or higher, and platelet count were evaluated. We compared ratios of numbers of patients in specific cohorts using chi-squared tests with continuity correction. The cohort comprised 3 283 140 cancer patients, of whom 435 140 (13.3%) received anticoagulation within six months of their cancer diagnosis. Bleeding incidence was higher in cancer vs non-cancer patients across all anticoagulants studied: warfarin 20.2% vs 12.6%, rivaroxaban 16.7% vs 12.1%, LMWH 13.2% vs 9.7%, and apixaban 14.5% vs 9.3%, P < .001 for all comparisons. Among all anticoagulants except warfarin, we found increased bleeding incidence in cancer patients with metastatic disease, gastrointestinal primary, CKD ≥ stage III, and platelets <100,000 × 10⁹ /L. Bleeding incidence was higher in cancer patients regardless of the anticoagulant used. Patients with gastrointestinal malignancies had a higher incidence of bleeding compared to other tumors across all anticoagulants. Other factors associated with increased risk of bleeding included metastatic disease, chronic kidney disease, and thrombocytopenia.