Risk Prediction Models for Delayed Bleeding After Endoscopic Submucosal Dissection in Patients with Early Gastric Cancer: A Systematic Review and Meta-Analysis

BACKGROUND/OBJECTIVES: Delayed bleeding is a common complication after endoscopic submucosal dissection (ESD) for early gastric cancer (EGC), but the quality and clinical applicability of existing risk prediction models remain uncertain. This meta-analysis was conducted to systematically evaluate and quantitatively synthesize published studies on risk prediction models for delayed bleeding after ESD in patients with EGC. METHODS: We performed a systematic review and meta-analysis. All studies were searched in electronic databases (PubMed, Web of Science, Cochrane Library, Embase, CINAHL, CNKI, VIP, and Wanfang databases) from inception to June 8, 2025. Two independent researchers screened the literature, extracted information, and assessed the risk of bias and applicability of the included studies. Meta-analysis was performed using Stata 18.0 Software. RESULTS: Of 1245 studies, 12 studies were included in this review. The incidence of delayed bleeding after ESD in patients with EGC ranged from 4.7% to 18.33%. The most common predictors were lesion location and tumor size (maximum diameter of the lesion). The reported AUC values ranged from 0.53 to 0.99; the meta-analysis results showed that the pooled AUC for internal validation of 8 delayed bleeding risk prediction models was 0.74 (95%CI 0.67–0.80), while the AUC for external validation of 4 models was 0.70 (95%CI 0.61–0.79). The overall risk of bias was high. CONCLUSIONS: Risk prediction models for delayed bleeding after ESD in EGC patients demonstrated some degree of discriminatory power; however, all these studies were at high risk of bias in terms of quality. PROSPERO Registration Number: CRD420251074300.