The Diagnostic Yield of Repeated Endoscopic Evaluation in Patients with Gastrointestinal Bleeding and Left Ventricular Assist Devices

Hassan K. Dakik, Alyson A. McGhan, Shih-Ting Chiu, Chetan B. Patel, Carmelo A. Milano, Joseph G. Rogers, Shein-Chung Chow, Daniel M. Wild

Left ventricular assist devices (LVADs) are increasingly common in patients with advanced heart failure. GI bleeding (GIB) occurs in 20-30 % of these patients and can arise anywhere in the GI tract. Given the high rates of GIB in this population, our aim was to determine the diagnostic yield of repeated endoscopic evaluation in these patients. We performed a retrospective review of all 257 patients who had LVADs placed between 2008 and 2013 at Duke University Hospital and identified all patients who underwent any endoscopic evaluation for GIB. Of the 257 patients with LVADs placed, 78 (30 %) underwent at least one endoscopy for GIB. A source was identified in 36 % of cases, most commonly angioectasias (53.6 %). Treatment was performed in 67.9 % of patients and hemostasis was achieved in all. 64.1 % of the cohort underwent a second endoscopy for GIB. 42.9 % of these exams revealed a bleeding source. Endoscopic treatment was employed in 76.2 %. 38.5 % of the cohort underwent a third endoscopic exam for bleeding and a source was identified in 53.3 % with angioectasias remaining most common (56.3 %). By Fisher's exact and Chi-square testing, only the presence of a bleeding source (p = 0.0034) and use of hemostatic therapy (p = 0.0127) on the index examination were significantly associated with re-bleeding. GIB is common in patients with LVADs. The diagnostic and therapeutic yield of endoscopy is remains high with repeated interventions. Despite these high yields, a large portion of the cohort requires repeated interventions for recurrent bleeding.