Association of Bleeding Severity With Mortality in Extended Thromboprophylaxis of Medically Ill Patients in the MAGELLAN and MARINER Trials

Background: Extended thromboprophylaxis has not been widely implemented in acutely ill medical patients due to bleeding concerns. The MAGELLAN and MARINER trials evaluated whether rivaroxaban compared with enoxaparin or placebo could prevent venous thromboembolism (VTE) without increased bleeding. We hypothesized that patients with major bleeding (MB) but not those with non-major clinically relevant bleeding (NMCRB), would be at an increased risk of all-cause mortality (ACM).Methods: We evaluated all bleeding events in patients taking at least one dose of study drug and their association with ACM in 4 mutually exclusive groups: (1) no bleeding, or first event was (2) NMCRB, (3) MB, or (4) trivial bleeding. Using a Cox proportional hazards model adjusted for differences in baseline characteristics associated with ACM, we assessed the risk of ACM after such events.Results: Compared to patients with no bleeding, the risk of ACM for patients with NMCRB was not increased in MARINER (HR 0.43, p=0.235) but was increased in MAGELLAN (HR 1.74 p=0.021). MB was associated with a higher incidence of ACM in both studies, while trivial bleeding was not associated with ACM in either study.Conclusions: Patients with MB had an increased risk of ACM, while NMCRB was not consistently associated with an increased risk of death. These results inform the risk-benefit calculus of extended thromboprophylaxis in medically ill patients.