Comparison of prediction of outcomes in upper GI bleed using non-endoscopic scoring systems

ABSTRACT

Background: Several scoring systems have been designed for risk stratification and prediction of outcomes in upper GI bleed. Endoscopy plays a major role in the diagnostic and therapeutic management of UGIB patients. However not all patients with UGIB need endoscopy. The objective of the present study was compared the prediction of mortality using different scoring systems in patients with upper GI bleed. A decision tool with a high sensitivity would be able to identify high and low risk patients and for judicious utilization of available resources.Methods: 100 patients were assessed with respect to their clinical parameters, organ dysfunction, pertinent laboratory parameters and five risk assessment scores i.e. clinical Rockall, Glasgow Blatchford, ALBI, PALBI and AIMS65 were calculated.Results: For prediction of outcomes, AIMS65 was superior to the others (AUROC of 0.889), followed by the GBS (AUROC of 0.869), followed by clinical Rockall score (AUROC 0.815), followed by ALBI score (AUROC of 0.765), followed by PALBI score (AUROC of 0.714) all values being statistically significant.Conclusions: The AIMS65 score is best in predicting the mortality in patients with upper GI bleed. The optimum cut off being >2. Though GBS may be better in predicting the need for intervention, it is inferior in predicting the mortality. The newer scores like ALBI and PALBI are inferior to AIMS65 and GBS in predicting mortality.