Predictors of Clinical Success in Patients Selected for Endoscopic Management of Gastrointestinal Transmural Defects

Background and study aim: To identify factors associated with clinical success after endoscopic treatment of gastrointestinal transmural defects and to develop an exploratory internally validated prediction model. Methods: We retrospectively analyzed consecutive patients treated endoscopically for gastrointestinal transmural defects between January 2014 and December 2024. Technical success was assessed at procedure level, whereas clinical success and prognostic modeling were analyzed at patient level to avoid overrepresentation of repeated procedures. Repeated interventions for the same defect were considered part of the same therapeutic episode. Clinical success was defined as complete defect resolution without need for additional surgical intervention. A ridge-penalized logistic regression model with bootstrap internal validation was developed. Results: Ninety-five endoscopic procedures were performed in 83 patients, predominantly for postoperative (79.5%) and upper gastrointestinal (95.2%) defects. Procedure-level technical success was achieved in 89/95 procedures (93.7%). Patient-level clinical success was achieved in 50/83 patients (60.2%). Larger defect size was independently associated with lower probability of clinical success (OR 0.90 per mm increase, 95% CI 0.83–0.99; p=0.026), whereas negative-pressure therapy showed a positive association trend (OR 9.27, 95% CI 0.88–97.81; p=0.064). Model discrimination after bootstrap internal validation was moderate, with reasonable agreement between predicted and observed outcomes. Conclusions: Endoscopic management of gastrointestinal transmural defects achieves high technical success but moderate clinical success in real-world practice. An exploratory internally validated prediction model may support risk stratification, although external validation is required before routine clinical use.