610 The Comparative Effectiveness of Endoscopy versus Angiography in Patients With Recurrent Non-Variceal Upper Gastrointestinal Bleeding

Abstract

INTRODUCTION: Patients with recurrent non-variceal upper gastrointestinal bleed (NVUGIB) undergo repeat EGD, and for any further bleeding episode, consideration of angiography or interventional radiology (IR) with arterial embolization is employed. However, there is minimal evidence to support these recommendations promulgated by the national societies (refer to guideline). We sought to compare the effectiveness of repeat EGD with IR in achieving hemostasis in patients with recurrent NVUGIB after initial endoscopy. METHODS: This is a single-center retrospective cohort study. Between 2008-2018 consecutive ICU patients who underwent an EGD for diagnosis of NVUGIB (melena, hematochezia, hematemesis, coffee-ground emesis) were included. Continuous bleeding was defined by no hemostasis achieved, while recurrent bleeding was defined as re-bleeding within 30 days of hemostasis after index EGD. When hemostasis was not achieved, patients were stratified into treatment modalities of repeat EGD, IR without embolization, empiric embolization or lesion embolization. Generalized linear mixed models were used to evaluate if treatment effects re-bleeding when all bleeding episodes were analyzed. RESULTS: A total of 1067 patients were eligible for analysis. Mean age was 63 years and 44% were female. Following initial EGD, 90% (n = 960) of patients achieved hemostasis, 16% (n = 157) had recurrent bleeding and 10% (n = 107) had continuous bleeding without initial hemostasis. When compared to patients who underwent EGD and IR, patients with repeat EGD only had significantly higher Charlson Comorbidity Index (4.8 ± 3.2 vs. 3.4 ± 2.4; P < 0.001) and fewer re-bleeding episodes (P < 0.001). When all episodes of recurrent and continuous bleeding were analyzed, EGD had a three times higher likelihood of hemostasis when compared to IR without embolization (OR 2.9, 95%CI 1.05-7.7; P = 0.039). There is no significant difference between EGD and IR with empiric or lesion embolization or when all forms of IR were combined. CONCLUSION: Repeat EGD is three times more likely than IR without embolization in achieving hemostasis for all episodes of recurrent or continued NVUGIB after an initial endoscopy. However, EGD and IR with embolization appear equally effective modalities in achieving hemostasis after recurrent or continuous NVUGIB. The study is the first to directly compare EGD with IR and validates consensus recommendations of repeating upper endoscopy when hemostasis is not initially achieved as well as utilizing IR when necessary.