A novel scoring system to predict therapeutic intervention for non-variceal upper gastrointestinal bleeding

Various scoring systems have been developed to predict the need for endoscopic treatment in patients with non-variceal upper gastrointestinal bleeding (NVUGIB). However, they have rarely been applied in clinical practice because the processes are complicated. The aim of this study was to establish a simple scoring system that predicts the need for endoscopic intervention in patients with NVUGIB. We retrospectively enrolled 509 consecutive patients with suspected NVUGIB who underwent emergency endoscopy. In the development cohort (from January 2016 to December 2018), risk factors that predict the need for endoscopic intervention were determined from 349 patients' data by multivariate logistic regression analysis. This led to the development of a novel scoring system named the Nagoya University score (N score). In the validation cohort (from January 2019 to September 2020), we evaluated the diagnostic value of the N score, the Hirosaki score, and the Glasgow-Blatchford scores (GBS) by receiver operating characteristic (ROC) curves using another 160 patients' data. Multivariate logistic regression analysis revealed syncope, hematemesis, blood urea nitrogen (BUN), and BUN/Cr as significant predictive factors for endoscopic intervention. In the validation study, the N score was superior to the GBS and equal to the Hirosaki score in predicting the endoscopic intervention (AUC, N score 0.776 [95% CI 0.702-0.851] vs. GBS 0.615 [0.523-0.708], Hirosaki 0.719 [0.636-0.803]). The N score revealed a sensitivity of 84.5% and a specificity of 61.8%. Our N score, which is consisted of only four factors, would select patients who require endoscopic intervention with high probability.