Limited Benefit of Additional Inpatient Diagnostic Evaluation in Patients With Upper Gastrointestinal Bleeding and a Negative Initial EGD

Abstract

Introduction: Upper gastrointestinal bleeding (UGIB) is a common reason for hospitalization and is associated with significant costs and resource utilization. For patients with a negative initial esophagogastroduodenoscopy (EGD) it is unclear if further inpatient testing affects clinical outcomes. Aims: To evaluate whether additional in-hospital diagnostic testing affect outcomes in patients with UGIB and a negative initial EGD. Methods: We conducted a prospective observational cohort study of consecutive adult patients admitted to an academic medical center from April 2016 to April 2017 with melena, hematemesis or coffee ground emesis and who underwent EGD. The cohort was divided between those with diagnostic and non-diagnostic EGDs and between those who underwent further inpatient testing or did not. Further testing included repeat EGD, colonoscopy, flexible sigmoidoscopy, capsule endoscopy, enteroscopy, and CT or IR angiogram. In-hospital outcomes included endoscopic intervention, re-bleeding, ICU admission, transfusion requirement, number of red cells transfused, length of stay (LOS) and in-hospital mortality. Two-tailed Students T test and Fisher's exact test were used to compare the two groups. Results: One hundred and ninety four patients underwent EGD for symptoms of UGIB, 113 (58%) had source of bleeding identified on initial EGD, and 58 (30%) underwent endoscopic intervention. Patients with a positive EGD had lower initial hemoglobin, higher BUN, higher heart rate, and higher Glasgow Blatchford scores (see Table 1). They were more likely to require ICU stay, red cell and plasma transfusion (P values 0.007, 0.01, 0.05; percent difference 20, 19, 12). Of those with negative initial EGD, 50 (62%) patients underwent additional inpatient testing. A source of bleeding was identified in 17 (34%) patients (11 with colonic source including 3 diverticular bleeds, 4 with small bowel source including 2 AVMs, and 2 gastric ulcers missed on initial EGD). Two patients with small bowel AVMs and 1 with a colonic ulcer underwent endoscopic intervention. There were no significant differences in re-bleeding, ICU stay, red cell transfusion requirement, or mortality whether patients underwent additional inpatient testing or not (P values 0.15, 0.82, 0.07, 1.00; percent difference 17, 4, 21, 2).Table: Table. Baseline characteristics and outcomes of cohort and those with source identified or not identified on index EGDConclusion: Among patients with UGIB and non-diagnostic initial EGD further inpatient diagnostic testing did not impact outcomes.Table: Table. Comparison of baseline characteristics and outcomes of those who underwent EGD only (with no source identified) and those who underwent additional inpatient diagnostic studies after negative EGDFigure: Intervention and source identification flowchart for patients presenting with UGIB and who underwent EGD.