A comparative analysis of risk scoring systems in predicting clinical outcomes in upper gastrointestinal bleed

Upper gastrointestinal bleed (UGIB) is a life threatening condition which presents as hematemesis, coffee-ground vomiting or melena. There are multiple scoring systems which have been developed for the same, which predict different outcomes, are important in the management of UGIB and are essential to determine high and low risk patients.

To compare the sensitivity and specificity for various risk scoring systems for UGIB and to determine the optimum cut-off values of various risk scoring systems in assessment of patients with UGIB.

81 patients presenting with acute UGIB to M. S. Ramaiah hospitals between October 2018 to July 2020 were evaluated and Glasgow Blatchford score (GBS), AIMS65, pre-rockall and full- rockall scores were calculated for risk assessment in predicting clinical outcomes such as need for ICU care, blood transfusion, in-hospital mortality or discharge and endoscopic intervention.

The GBS was the best at predicting mortality with the cut off of >14 (p 0.01, sensitivity 80%, NPV 98.6%) and blood transfusion at a cut-off >6 (sensitivity 97.6%, NPV 94.6%). The GBS and AIMS65 scores of more than 8 (p 0.000) and 1 (p 0.032) respectively were good predictors of need for ICU care. AIMS65 was the best at predicting duration of hospitalisation (p 0.032).

The GBS was superior in predicting clinical outcomes, categorisation into high risk and low risk and need for blood transfusion. GBS and AIMS65 scores help in predicting the need for ICU care; AIMS65 being a simple score will reduce the cost burden of unnecessary ICU admissions.