Objective: Risk assessment in nonvariceal upper gastrointestinal bleeding (UGIB) is not well validated and remains unclear in hemodynamically stable patients at emergency department admission. We compared the prognostic value of risk-scoring systems for predicting adverse outcomes in patients with nonvariceal UGIB and normotension.
Participants and methods: A single-center prospective observational study was carried out. Patients with consecutive nonvariceal UGIB, presenting with normotension (systolic blood pressure ≥90 mmHg) to the emergency department, were included. We compared the areas under the curves (AUC) of Glasgow Blatchford score (GBS), the pre-endoscopy Rockall score, AIMS65, the shock index, and the modified shock index with respect to adverse outcomes defined as embolization, surgery, ICU admission, rebleeding, and in-hospital mortality.
Results: In total, 1233 patients were included. Adverse outcomes occurred in 165 (13.4%) patients; in-hospital mortality was 1.2%. AUC of the GBS for adverse outcome was higher than that of the shock index, but not significantly different (0.647 vs. 0.569, P=0.23). AUC values of the modified shock index, AIMS65, and the pre-endoscopy Rockall score were 0.565, 0.593, and 0.533, respectively. The cut-off value of the GBS (≥8) was associated with 85% sensitivity and 35% specificity for predicting adverse outcome.
Conclusion: Pre-existing risk scores have shown suboptimal predictive ability for adverse events in normotensive patients with nonvariceal UGIB. The GBS (≥8) might help to identify patients prone to adverse events; however, further studies with risk scores or new scores are needed because of the low accuracy of these scores.