Project Summary Among patients with cancer, the risk of anticoagulant-related major bleeding is more than double that of patients without cancer. These bleeds have a fatality rate of 7%. Current guidelines recommend indefinite anticoagulant therapy in many patients with cancer and venous thromboembolism (VTE) with periodic assessments of bleeding risk. However, no evidence-based models exist to quantify bleeding risk in patients with cancer taking anticoagulants. In addition, recent trials have demonstrated effectiveness of direct oral anticoagulants as primary prophylaxis for prevention of initial VTE in patients with cancer. Thus, there is a strong need to balance the competing risks of VTE and anticoagulant-related bleeding in patients with cancer. While there are models that quantify bleeding risk in non-cancer populations (i.e. atrial fibrillation, non- cancer associated VTE) taking anticoagulants, they have not been validated in patients with cancer. Quantification of bleeding risk in patients with cancer and VTE would allow for an evidence-based approach to preventing and treating VTE by minimizing anticoagulant therapy in patients at high-risk of bleeding and allowing for judicial use in patients at low-risk of bleeding. The overall objective of this study is to develop a prediction model for anticoagulant-related bleeding in patients with cancer and VTE on anticoagulant therapy and to calibrate that model in cancer patients at high-risk of VTE being considered for primary thromboprophylaxis with anticoagulant therapy. We will accomplish our overall objective through the following specific aims: (1) validate eight existing bleeding prediction models in a cohort of 7489 patients with cancer and VTE receiving anticoagulant therapy, (2) refine and validate a cancer-specific prediction model to quantify risk of anticoagulant-associated bleeding in patients with cancer on anticoagulant therapy for VTE, and (3) calibrate the refined bleeding prediction model in a cohort of patients with cancer receiving primary anticoagulant thromboprophylaxis enrolled in two randomized trials (AVERT and CASSINI). These aims will be addressed with an expert research team in the areas of bleeding, VTE (especially in patients with cancer) and development and implementation of risk prediction models. This proposal is innovative as it addresses a current gap in the management of patients with cancer with or at high-risk of VTE by developing a prediction model to quantify the risk of the intervention (anticoagulant therapy) to inform and guide patient care. This research will be significant because it has the potential to decrease anticoagulant-related bleeding and thus bleeding related morbidity and mortality in patients with cancer.