Comparison of the Rockall Score and Revised Rockall Score for predicting outcomes in nonvariceal upper gastrointestinal bleeding

The Rockall score (RS) is used to assess the prognosis of patients with nonvariceal upper gastrointestinal bleeding. Aims: To establish the Revised Rockall Score (RRS) that incorporates an assessment of endoscopic treatment results, and analyze the discriminative ability of the RRS for 30-day rebleeding and mortality in patients with nonvariceal upper gastrointestinal bleeding. We retrospectively analyzed data from 432 patients between January 2016 and December 2019 (derivation cohort), and further evaluated in an independent population of 290 patients between January 2020 and December 2022 (validation cohort). Clinical records and biological data were collected. The outcome variables were rebleeding and mortality, whereas the explanatory variables were the RS and RRS. The predictive accuracy of the two scoring systems was evaluated using the receiver operating characteristic curve (AUROC). In the derivation cohort, the RRS achieved a higher area under the AUROC for predicting rebleeding (AUROC: RS, 0.75; RRS, 0.88; p < 0.0001), mortality (AUROC: RS, 0.87; RRS, 0.94; p = 0.001), and rebleeding and/or mortality (AUROC: RS, 0.78; RRS, 0.90; p < 0.0001) than the RS. In the validation cohort, the RRS also achieved a higher AUROC for predicting rebleeding (AUROC: RS, 0.80; RRS, 0.89; p < 0.001), mortality (AUROC: RS, 0.79; RRS, 0.89; p = 0.004), and rebleeding and/or mortality (AUROC: RS, 0.80; RRS, 0.91; p < 0.001) than the RS. Compared to the RS, the RRS had higher discriminative ability in predicting the risk of rebleeding and mortality.