ALBERT

Description: Introduction: Thrombosis is a major cause of morbidity and mortality in patients with classical myeloproliferative neoplasms (MPN), which include polycythemia vera, essential thrombocythemia and primary myelofibrosis. One third of patients with MPN suffer a thrombotic event, either arterial or venous. Despite this, there are no strong data to guide either the selection or duration of anticoagulants for MPN. Warfarin has been primarily used for long-term anticoagulation and is associated with a clear reduction in overall thrombotic events, however 20% of patients will have a recurrent thrombotic event despite ongoing warfarin anticoagulation, corresponding to a failure rate of 4-8% pt-yrs (AnnHematol 2015;94:911-918, Haematologica2008;93:372-380). The overall objective of this study was to evaluate recurrent thrombotic and major bleeding events in MPN-associated thrombosis treated with DOAC versus warfarin. Methods: The primary outcome was to compare the rate of thrombotic events in patients treated with DOAC versus warfarin for secondary thromboprophylaxis (treatment of first clot) among MPN patients. The secondary outcome compared the rate of bleeding events between the two groups. Electronic medical records for patients with a classical MPN diagnosis by PSVG criteria at the University of North Carolina Hospitals were queried to evaluate DOAC or warfarin medication orders between January 1, 2010 and May 31, 2017. Demographic laboratory data, concomitant medications, and incidence and severity of thrombotic or bleeding events were recorded for each MPN patient treated with a DOAC, warfarin, or aspirin. Descriptive statistics were used to characterize the population. Categorical variables were summarized as counts and percentages, while continuous variables were summarized as medians with first to third quartiles. The annual incidence of secondary thrombosis (ie: recurrent event) was calculated by dividing the number of events by the total number of patient-years (pt-yrs). 95% confidence intervals and comparisons were made using a Mid-P exact test. For all analyses, results were deemed significant if P