Abstract
INTRODUCTION: Endoscopic submucosal dissection (ESD) has been popular worldwide to treat laterally spreading tumors and large polyps. Post-ESD coagulation syndrome (PECS) is more common than the two major ESD-related complications, perforation, and bleeding. The aim of this study was to assess the prevalence of PECS, identify the risk factors for PECS, and create a risk prediction model for PECS. METHODS: Retrospective cross-sectional study analyzed a total of 986 patients who underwent colorectal ESD. Logistic regression models were used to assess risk factors with PECS. Each risk factor was scored, and the 3-step risk stratification index of prediction model was assessed. RESULTS: The prevalence of PECS was 21.4% (95% confidence interval [CI] = 18.9-24.1%). The risk factors of PECS in the multivariate logistic regression were tumor size (+1 cm: odds ratio [OR], 1.29; 95% CI, 1.16-7.09), cecal lesion (OR, 1.96; 95% CI, 1.09-1.53), procedure time (+30 min: OR, 1.19; 95% CI, 1.02-1.39), and ESD with snaring (OR, 0.64; 95% CI, 0.43-0.95). Applying a simplified weighted scoring system based on adjusted OR increments of 1, the risk of PECS was 12.3% (95% CI, 0.3-16.0%) for the low-risk group (score ≤4) and was 36.0% (95% CI = 29.4-43.2%) for the high-risk group (score ≥8). Overall discrimination (C-statistic = 0.629; 95% CI = 0.585-0.672) and calibration (p = 0.993) of the model were moderate to good. CONCLUSION: PECS occurs frequently, and the prediction model can be helpful for effective treatment and prevention of PECS.