Construction and validation of a predictive model for the risk of rebleeding in patients with esophageal and gastric varices hemorrhage.

Background and purpose: Esophageal and gastric varices hemorrhage (EGVH) is a life-threatening condition with the 6-week mortality rate of 15-25%. Up to 60% of patients with EGVH may experience rebleeding with a mortality rate of 33%. The existing scoring systems, such as RS scoring system (Rockall score, RS) and GBS scoring system (Glasgow-Blatchford score, GBS), have limitations in predicting the risk of rebleeding. Our study was to construct and validate a novel predictive model for the risk of rebleeding in patients with EGVH and to compare the predictive power of the predictive model with GBS and pRS.

Methods: Data of patients with EGVH was collected in the First Affiliated Hospital of Dalian Medical University from January 2016 to June 2020. Binary logistic and stepwise regression was performed to construct a predictive model. We compared the predictive power of the new predictive model to the GBS and pRS scoring systems.

Results: Clinical data from a total of 265 patients with EGVH was collected. Six factors including systolic blood pressure, transfusion requirement, CA199, platelet count, upper esophageal varices and severity of esophageal varices were included in our new predictive model. The AUCs of the specificity of the predictive model, GBS and pRS are 0.82, 0.60 and 0.56.

Conclusion: This study successfully constructed a predictive model for the risk of rebleeding in patients with EGVH. This predictive model demonstrated higher predictive ability than pRS and GBS scoring systems for assessing rebleeding risk in EGVH patients.