Prognostic value of scoring systems in geriatric patients receiving warfarin presenting with upper gastrointestinal bleeding

This study compared the performance of the Glasgow-Blatchford Score (GBS), AIMS65, and the pre-endoscopic Rockall score in predicting clinical outcomes in patients aged ≥ 65 years receiving warfarin who presented with upper gastrointestinal bleeding (UGIB). In this multicenter retrospective study, geriatric patients presenting to the emergency departments of tertiary care hospitals over a 3-year period were evaluated. The primary outcome was in-hospital mortality, while secondary outcomes included RBC transfusion requirement and intensive care unit (ICU) admission. Predictive performance was assessed using receiver operating characteristic (ROC) curve analysis, and pairwise comparisons of correlated ROC curves were performed using the DeLong test with Bonferroni adjustment. A total of 126 patients were included. AIMS65 demonstrated the highest AUC for predicting in-hospital mortality (AUC = 0.815, 95% CI: 0.723–0.887), with significantly better discrimination than the pre-endoscopic Rockall score (DeLong p = 0.0005). GBS showed the highest AUC for predicting transfusion requirement (AUC = 0.655, 95% CI: 0.564–0.739, p = 0.002) and ICU admission (AUC = 0.739, 95% CI: 0.652–0.815, p < 0.001). GBS significantly outperformed AIMS65 for predicting transfusion requirement ( p = 0.0023), whereas no other pairwise differences were statistically significant. These findings indicate that the three scoring systems demonstrated different patterns of predictive performance across the evaluated clinical outcomes, with AIMS65 showing the strongest discrimination for in-hospital mortality and GBS showing the strongest numerical discrimination for RBC transfusion requirement and ICU admission.