ABSTRACT
BACKGROUND AND AIM: Risk stratification is recommended in all patients with acute variceal bleeding (AVB). It remains unclear whether liver disease severity or upper gastrointestinal bleeding (UGIB) scoring algorithms offer superior predictive ability. We aimed to validate the AIMS65 score as a predictor of mortality in AVB, and to compare AIMS65 with established UGIB and liver disease severity risk stratification scores. METHODS: International Classification of Diseases, Tenth Revision codes identified patients presenting with AVB to three tertiary centers over a 48-month period. Patients were risk-stratified using AIMS65, Rockall, pre-endoscopy Rockall, Child-Pugh, Model for End-stage Liver Disease (MELD) and United Kingdom MELD (UKELD) scores. Primary outcomes were inpatient and 6-week mortality and inpatient rebleeding. RESULTS: Two hundred and twenty-three patients were included. Inpatient and 6-week mortality were 13.9% and 15.5% respectively. Prediction of inpatient mortality by AIMS65 (area under the receiver-operating characteristic curve [AUROC: 0.84]) was equivalent to UGIB (Rockall: 0.79, pre-Rockall: 0.78) and liver risk scores (MELD: 0.81, UKELD: 0.79, Child-Pugh: 0.78). AIMS65 score ≥3 best defined high- and low-risk groups for inpatient mortality (mortality 37.7% vs 4.9%). AIMS65 (AUROC: 0.62) was equivalent to UGIB risk scores (pre-Rockall: 0.64, Rockall: 0.70) in predicting inpatient rebleeding and superior to liver risk scores (MELD: 0.56, UKELD: 0.57, Child-Pugh: 0.60). CONCLUSIONS: AIMS65 is equivalent to established UGIB and liver disease severity risk stratification scores in predicting mortality, and superior to liver scores in predicting rebleeding.
