Objectives The Glasgow‐Blatchford score (GBS) is a widely used risk assessment tool for patients with upper gastrointestinal bleeding. However, it only identifies a relatively low proportion of patients at low risk for adverse events and poor outcomes. We developed a simple diagnostic algorithm combining the GBS and nasogastric aspirate and evaluated its diagnostic performance. Methods A total of 115 consecutive patients with suspected nonvariceal upper gastrointestinal bleeding who underwent nasogastric tube placement and upper endoscopy at our emergency department were prospectively evaluated. We compared the diagnostic accuracy of the GBS and our algorithm for predicting high‐risk endoscopic lesions (HRELs) using receiver operating characteristic curve analysis. Results Thirty‐five patients had HRELs. Compared with the GBS, our algorithm showed superior performance with respect to the prediction of HRELs (area under the curve, 0.639 and 0.854, respectively; p < 0.001). With set optimal threshold values, the algorithm identified a significantly higher proportion of patients who did not have HRELs than the GBS (23.5% vs. 2.6%, p < 0.001). Conclusions The novel algorithm has improved the diagnostic performance of the GBS and predicted more patients who did not have HRELs than the GBS alone. After further validation, it may be a useful tool for making clinical management decisions for patients with nonvariceal upper gastrointestinal bleeding.

No abstract

No abstract