Performance of the Glascow-Blatchford and AIMS65 scores for various clinical outcomes in non-variceal upper gastrointestinal bleeding

Aims Risk stratification is crucial for proper management in patients with non-variceal upper gastrointestinal bleeding (NVUGIB). The Glascow-Blatchford score (GBS) and AIMS65 are some of the most used scoring systems. This study aimed to validate these scores’ ability to predict various clinical outcomes in patients with NVUGIB.

Methods Single-center retrospective cohort study including consecutive hospitalized patients managed for NVUGIB between January 2022 and July 2024. Receiver operating characteristic (ROC) curves and corresponding areas under the curve (AUC) were analyzed, for the following outcomes: need for blood transfusion, need for endoscopic therapy, re-bleeding, in-hospital death, composite endpoint (at least one of: need for radiological intervention, surgical intervention, or in-hospital death). The predictive performances of the scores were considered acceptable for AUC³0.70.

Results A total of 196 patients were included (140 males, 75.5±16.7 years, 54.1% peptic ulcers) of whom 116 (59.2%) received blood transfusion, 40 (20.4%) rebled, 144 (73.5%) underwent endoscopic therapy, 6 (3.1%) radiological intervention, 4 (2%) surgery and 14 (7.1%) deceased during their hospitalization. The composite endpoint was observed in 20 (10.2%) patients. The GBS (AUC=0.89), but not the AIMS65 (AUC=0.68), was predictive of in-hospital mortality. Both scoring systems predicted the need for blood transfusion (GBS: AUC=0.86, AIMS65: AUC=0.75) and the composite endpoint (GBS: AUC=0.84, AIMS65: AUC=0.77), whereas none of the scores predicted rebleeding (GBS: AUC=0.63, AIMS65: AUC=0.66) and need for endoscopic therapy (GBS: AUC=0.54, AIMS65: AUC=0.55). Optimal cut-points were GBS=12.5 (sensitivity/specificity: 1.00/0.75) and AIMS65=1.5 (sensitivity/specificity: 0.75/0.69) for predicting the composite endpoint, and GBS=12.5 (sensitivity/specificity: 1.00/0.62) AIMS65=2.5 (sensitivity/specificity: 0.50/0.89) for in-hospital mortality.

Conclusions Both GBS and AIMS65 are clinically useful scores in NVUGIB, although with certain limitations. AIMS65 does not perform acceptably in predicting mortality, whereas no scoring system can satisfactorily predict rebleeding and need for endoscopic therapy.

Publication History

Article published online:
27 March 2025