Evaluation of the prognostic performance of the Rockall and Glasgow-Blatchford scoring systems in non-variceal upper gastrointestinal bleeding

Introduction/Objective. Upper gastrointestinal bleeding is a significant medical emergency requiring prompt assessment and intervention. Various risk stratification tools, including the Rockall Score and Glasgow-Blatchford Score (GBS), are used to predict clinical outcomes such as mortality, intensive care unit admission, and the need for blood transfusion. Methods. This study analyzed a cohort of 199 patients admitted to our hospital for non-variceal upper gastrointestinal bleeding between October 1, 2020, and October 1, 2024. Demographic data, vital signs (pulse rate, systolic and diastolic blood pressure), length of hospital and ICU stay, comorbidities, and medication use were recorded. The Rockall Score and GBS were calculated for each patient, and their predictive accuracy was assessed using sensitivity and specificity analyses. Results. The GBS (AUC = 0.887) demonstrated superior predictive performance for blood transfusion compared to the Rockall Score (AUC = 0.786, p < 0.001). However, both scores exhibited poor predictive ability for ICU admission (AUC = 0.624 vs. 0.605, respectively, p < 0.05), with Rockall outperforming GBS. For mortality prediction, both scores performed similarly (Rockall: AUC = 0.847, GBS: AUC = 0.837, p = 0.239), indicating no significant difference. Conclusion. GBS outperforms the Rockall Score in predicting blood transfusion need, while both scores show poor ICU admission prediction, with Rockall performing slightly better. For mortality prediction, both scores are comparable. GBS is preferable for transfusion assessment, but additional factors may improve ICU and mortality predictions.