Anticoagulation Increases Bleeding Rates in Portal Vein Thrombosis: A National Database Study

Introduction: Liver cirrhosis is being increasingly recognized as a hypercoagulable state, mainly due to disproportionate reduction in antithrombotic factors (protein C, S and anti-thrombin III). Portal vein thrombosis (PVT) is a frequent sequala of liver cirrhosis that can lead to numerous complications and potential exclusion from transplant lists, at least until the resolution of the thrombus. The current evidence for anticoagulation (AC) for PVT in liver cirrhosis is limited to small retrospective studies. Major liver societies recommend limited anticoagulation with enoxaparin based on expert opinion (Category C). We sought to examine the incidence of bleeding after AC in cirrhotic patients with PVT. Methods: Data were obtained from a commercial de-identified database (Explorys, IBM, Inc.) that integrates electronic health records from 27 major integrated U.S. healthcare systems from 1999 to July 2019. Cases were defined as adult patients aged >=18 years having a new Systematized Nomenclature of Medicine Clinical Terms (SNOMED) diagnosis of PVT, had been anticoagulated for the first time following PVT and had experienced bleeding for the first time following AC. Controls were adults who had a diagnosis of PVT and did not get AC. We included older anticoagulants (warfarin and enoxaparin) as well as newer anticoagulants (apixaban, fondaparinux and rivaroxaban). We compared the incidence of bleeding (defined as bleeding from any site) in those with PVT who received AC to those with PVT who did not receive AC. In addition, we also compared the incidence of bleeding between older and newer anticoagulants, and also examined whether there were differences in gender, race, and insurance status for those who bled while on AC. Analysis comprised of calculating odds ratios (OR) and confidence intervals (CI) for the OR. Results: A total of 213,810 patients had liver cirrhosis and out of those, 7,570 (3.5%) patients had PVT. Four hundred and ten cases out of 1,430 patients who received AC bled for the first time whereas 980 cases out of 3,880 patients who did not receive AC bled for the first time. Cases on AC had 1.18 times higher odds of bleeding when compared to controls who did not receive AC (CI = 1.04 - 1.36). Newer AC were less likely to increase bleeding when compared to older AC (OR = 0.6, CI = 0.4 - 0.9). Females were significantly more likely to be first time bleeders on AC when compared to male first-time bleeders on AC (Table 1). However, race and insu