Abstract: Objective Systematic evaluation of predictors of esophageal and gastric variceal bleeding risk prediction models for clinical construction of relevant models and clinical practice. Methods A comprehensive search for studies related to predictive models of esophageal and gastric variceal bleeding (EGVB) risk was conducted in Chinese and English databases, including PubMed, EMbase, Web of Science, the Cochrane Library, SinoMed, China Knowledge, Wanfang, Wipro, and Chinese Medical Journal Full Text Databases, with a timeframe of the construction of the library to January 6, 2025, for the search. The information of the included literature was extracted by two researchers, and the risk of literature bias and applicability were evaluated using the PROBAST tool. The meta-analysis of the model was performed using the R program (version 4.4.2), MedCalc software. Results A total of 12 studies were included, including 4 English literature and 8 Chinese literature. The area under the curve (AUC) of the subjects in the predictive model was 0.802~0.994. Meta-analysis was conducted on the AUC of the model using MedCalc software, and the combined AUC was 0.888 (95%CI: 0.857~0.919). The results of the meta-analysis showed that platelet (OR=0.98, 95%CI: 0.97~0.99), spleen thickness (OR=1.11, 95%CI:1.09~1.14), liver rough surface (OR=6.54, 95%CI:4.65~9.24), ascites (OR=3.83, 95%CI:1.55~9.50), and red color sign (OR=4.48, 95%CI: 2.81~7.13) were effective predictors in the predictive model for the risk of esophageal and gastric variceal bleeding (P<0.05). Conclusion Predictive models for the risk of EGVB are still in the exploratory stage. In the future, prospective clinical studies with multicenter and expanded sample sizes can be carried out to improve the methodology of model construction and to develop a risk prediction model for EGVB with a higher degree of identification accuracy as early as possible.

Key words: Esophageal and gastric varices, Bleeding, Risk prediction model, Systematic evaluation, Meta-analysis