Outcomes and Predictors of Readmissions with GI Bleeding in Patients with Left Ventricular Assist Devices.

OBJECTIVES: Gastrointestinal (GI) bleeding is a major comorbidity in patients with left ventricular assist devices (LVADs). The study aim was to estimate the rate of hospital readmissions for GI bleeding in patients with LVADs using a nationally representative database. Additionally, we evaluated the etiologies, costs, endoscopy utilization, mortality, and predictors of GI bleeding readmissions in these patients.

METHODS: We analyzed data from the National Readmissions Database (NRD) from 2010 through 2014. We compared hospitalized adult patients with congestive heart failure (CHF) who underwent LVAD implantation (cases) with CHF patients without LVAD or heart transplant (controls). Three age- and sex-matched controls were randomly selected per single case. A multivariate Cox regression model was used to compare the hazards of 60-day all-cause and GI bleeding readmission between the groups, controlling for significant confounders.

RESULTS: A total of 3293 hospitalized patients with CHF who had LVAD placement (cases) and 9879 who did not have LVADs (controls) were included in the study. At 60 days, patients with LVAD had a significantly higher readmission rate with GI bleeding (8.7% vs 2.3%, adjusted hazard ratio [aHR] 4.45, 95% confidence interval 3.71-5.33, P < 0.0001). The all-cause readmission rate also was higher (43.3% vs 35.7%, aHR 1.23, 95% confidence interval 1.12-1.34, P < 0.0001). The most common etiologies of bleeding in patients with LVADs were gastroduodenal and small intestinal arteriovenous malformations (28.6%). During bleeding readmissions, patients with LVAD were more likely to undergo endoscopy (72.1% vs 33.5%, P < 0.0001) and receive packed red blood cell transfusions (62% vs 36.6%, P < 0.0001) compared with controls. GI bleeding readmissions were more costly ($40,936 vs $35,313, P < 0.0001), and longer (12 vs 10.9 days, P < 0.0001) in patients with LVADs compared with controls. Independent risk factors for 60-day GI bleeding readmission were increasing age (aHR 1.04, P < 0.0001) and GI bleeding during index admission (aHR 2.68, P < 0.0001). In those without bleeding during index admission, increasing age and chronic anemia were associated with 60-day GI bleeding readmission. Mortality during bleeding readmission was similarly low in patients with LVADs compared with CHF controls (0.2% vs 0.3%, P = 0.14).

CONCLUSIONS: After LVAD implantation, there is a fivefold increased risk of readmission with GI bleeding within 60 days. Gastroduodenal and small intestinal arteriovenous malformations are the most common culprit lesions. These findings suggest that small bowel enteroscopy should be considered as the initial test of choice in patients with suspected upper gastroduodenal bleeding. Readmissions with bleeding in patients with LVADs increase morbidity and cost of care but not mortality. Older patients and those with a history of bleeding during LVAD implantation are at higher risk of bleeding readmission and may benefit from close monitoring and cautious anticoagulation to prevent rebleeding.