Major Bleeding Events Among Cancer and Non-Cancer Patients Taking Rivaroxaban for Venous Thromboembolism Treatment in a Department of Defense Health System Cohort

Background: Cancer patients are at increased risk of both venous thromboembolism (VTE) and bleeding, but real world bleeding rates with contemporary anticoagulants are not known.

Purpose: To describe the incidence of major bleeding (MB) in VTE patients treated with rivaroxaban, and to compare differences in MB incidence and characteristics among patients with active cancer, history of cancer, and no current or past cancer.

Methods: We queried over 10 million electronic medical records (EMRs) from the United States Department of Defense healthcare system from November 2, 2012 to September 30, 2015 to identify MB events in VTE patients treated with rivaroxaban. The Cunningham algorithm was used for identifying MB, and VTE was determined by diagnosis codes.

Presence of cancer was assessed from 5 years prior to index rivaroxaban exposure through end of study, and categorized by active cancer, history of cancer, and no cancer.

Active cancer was defined by one of the following: 1) a metastatic diagnosis code within 6 months prior to or overlapping with rivaroxaban exposure, 2) chemotherapy and/or radiation codes within 6 months prior to rivaroxaban exposure, 3) cancer-related surgery overlapping with rivaroxaban exposure, or 4) leukemia and/or indolent lymphoma codes at any point within the entire assessment window. History of cancer was defined as the presence of any cancer diagnosis within the 5-year baseline period not meeting the definition of active cancer. Patients with active cancer or history of cancer were further categorized by cancer site/type.

Incidence, outcomes, demographics, and bleeding risk scores were evaluated by cancer status. A Cox proportional hazard model was used to assess the association between cancer status and rate of MB adjusting for baseline characteristics.

Results: The study population comprised 9,638 VTE patients on rivaroxaban, including 1,728 (17.9%) with active cancer, 1,548 (16.1%) with history of cancer and 6,362 (66.0%) with no cancer. Of these, 130 (1.3%) experienced MB. After stratifying by cancer status, MB occurred at a rate of 2.61 (95% CI 1.80-3.78) per 100 person-years in those with active cancer, 3.18 (95% CI 2.17-4.67) per 100 person-years in those with history of cancer, and 2.25 (95% CI 1.80-2.81) per 100 person-years in those with no cancer.

No significant difference in the incidence of MB was found between those with cancer (active or history) and those without cancer (HR 1.01; 95% CI 0.70-1.47, p-value 0.94) after adjusting for age, sex, and baseline comorbidities. Neither history of cancer nor active cancer when independently compared to no cancer was significantly associated with MB after multivariate adjustment. MB rates varied notably by cancer site. Additional key findings are presented in Table 1.

Conclusion: In this large United States Department of Defense cohort study of rivaroxaban users treated for VTE, incidence of MB is relatively low and not significantly different between cancer and non-cancer patients. Fatal outcomes associated with bleeding hospitalization were also uncommon across all the cancer groups. These data should provide assurance to oncology providers regarding the safety of rivaroxaban use in the real-world setting.

View largeDownload slide

View largeDownload slide

Close modal