Background Assessment of the emergent endoscopy for upper gastrointestinal bleeding (UGIB) patients has important clinical implications. There is no validated criterion to triage. Aims To develop a simple score predicting an endoscopic intervention. Methods A prospective cohort study was conducted at a tertiary care centre. Primary outcome was the high-risk stigmata which were well-established endoscopic findings to determine the need for an endoscopic intervention. We created a simple score by multivariable logistic regression and compared with the Glasgow Blatchford Score (GBS). External validation was performed in a second cohort. Results 284 of consecutive 568 patients with suspected UGIB had the high-risk stigmata. Three variables were selected: “no daily use of proton pump inhibitors during one week before examination (+1 point)”, “shock index (heart rate/systolic blood pressure) ≥ 1 (+1 point)” and “urea/creatinine ≥ 140 (blood urea nitrogen/creatinine ≥ 30) (+1 point)”. The accumulating score (range 0–3) achieved an area under the receiver–operating characteristic curve (AUC) of 0.74 (95% confidence interval [CI], 0.70–0.78), which was superior to the GBS (AUC, 0.63; 95% CI, 0.59–0.68; p < 0.001). Validation in an external cohort demonstrated superiority to the GBS (AUC, 0.78 vs. 0.59; p < 0.001). Conclusions The simple score has greater accuracy than the GBS for assessing the need for an endoscopic intervention in cases of suspected UGIB. Further external validation should be performed to verify generalizability.
ASJC Scopus subject areas