(554) - Risk of Thromboembolic Events After Short Term Discontinuation of Anticoagulation Following a Gastrointestinal Bleed

Abstract

Gastrointestinal bleeding (GIB) is a frequent complication of left ventricular assist devices (LVAD). Short term discontinuation of anticoagulation is sometimes necessary in response to a GIB, but this may increase the risk of thromboembolic (TE) events. We sought to determine whether patients who suffer a GIB are at increased for a TE event when anticoagulation is discontinued for a period of time. Single center retrospective review of all patients implanted with a LVAD between January, 2011 and October, 2014 and who survived to hospital discharge. GIB was defined as melena, hematochezia, hematemesis, or anemia requiring either a transfusion or an endoscopy/colonoscopy. TE was defined as ischemic stroke, transient ischemic attack, hemolysis, suspected or confirmed pump thrombosis. The analysis was restricted to GIB that occurred after discharge from the hospitalization for LVAD implantation. Subsequent TE events were those that occurred within 365 days of the GIB. The study cohort included 141 patients, 55 with HeartMate II (HMII) and 86 with HVAD, who were followed for a median of 1.28 years and 0.9 years respectively. 29 GIBs occurred in 16 patients with HMII (0.24 event-per-person-year or eppy) and 35 GIBs occurred in 12 patients with HVAD (0.16 eppy). Of those, 5 GIB events with HMII and 6 GIB events with HVAD were not on warfarin at the time of GIB. Warfarin was stopped due to GIB in 24 GIB events with HMII and 28 GIB events with HVADs. The mean time off warfarin after the GIB was 15±10 days (range 2-40) and 22±35 days (range 2-181) for patients with HMII and HVAD respectively. Subsequent to the GIB, there were 5 TE events, 4 in patients with HMII and 1 in a patient with HVAD, yielding event rates of 0.43 eppy and 0.07 eppy respectively. The TE events occurred a median of 33 days (range 23-327) after the GIB in patients with HMII, and 38 days in the patient with HVAD. At the time of the TE event, warfarin had been resumed in in all 4 patients with HMII (median INR of 1.5), but not in the HVAD patient. In our cohort, risk of TE events appears acceptable with short term discontinuation of anticoagulation after a GIB in HVAD patients, but not in HMII patients. Further studies evaluating the optimal anticoagulation management strategies after GIB are warranted.