Predictive model for bleeding after gastric submucosal dissection before and after guidelines: A single-center retrospective study.

Objectives: In July 2017, supplementary guidelines on anticoagulants, including direct oral anticoagulants, were published in Japan. We investigated the changes in endoscopic submucosal dissection (ESD) of gastric mucosal lesions after the publication of the supplement, examined the risk factors, and developed a predictive model for post-ESD bleeding. Methods: We included 2272 gastric ESD cases from our hospital between May 2003 and June 2021 and classified them into two groups: 1789 cases before and 483 after the publication of the supplementary guidelines. A predictive model for post-ESD bleeding was developed using the pre-publication cohort data. Results: The proportion of patients receiving warfarin decreased (5.0% vs. 1.4%) and those receiving direct oral anticoagulants increased (1.2% vs. 6.8%) after the publication of the supplementary guidelines. Post-ESD bleeding occurred in 61 patients, but there was no significant difference in the bleeding rate between the groups (50 [2.8%] vs. 11 [2.3%] patients, respectively). Five risk factors (number of antithrombotic agents, dialysis, heparin replacement, resection specimen size, and procedure time) were identified for model development. The C-statistic for the model and post-publication cohorts were 0.83 and 0.72, respectively. In the model, each risk factor for postoperative bleeding was scored, and the risk was classified into three levels according to the total score. Bleeding rates at low, intermediate, and high risks were 1.6%, 10.3%, and 38.9%, respectively. Conclusion: Despite changes in patient characteristics and clinical practice regarding ESD before and after the publication of the supplementary guidelines, we could still develop a simple and useful predictive model.