S692 Tracking Competence in Real Time: A Fellow-Reported Study of Colonoscopy Success Factors

Introduction: Assessing trainee performance in endoscopy remains challenging due to limited data on real-time involvement and procedural skill. Existing tools like ProVation document performed interventions but often fail to reflect a fellow's autonomy or actual contribution. Without standardized, fellow-specific metrics, programs struggle to deliver targeted feedback and improve training outcomes. To address this, we developed a novel, fellow-led data collection system that allowed trainees to record procedural details directly. This study analyzes outcomes from one institution utilizing this method to identify predictors of colonoscopy success over a single academic year (2023 to 2024). Methods: We analyzed 341 colonoscopies recorded by fellows across 3 levels of training (year 1 to 3). Fellows documented their training year, degree of autonomy (50% vs 100%), and perceived case difficulty (0 to 100 scale). The primary outcome was successful ileal intubation. Logistic regression was used to assess predictors of success. Results: Ninety-one percent (n = 309) of colonoscopies were successful. Most procedures were performed by year 2 (53%) and year 3 (40%) fellows; year 1 fellows performed 6.5%. Full autonomy was reported in 95% of cases. Perceived difficulty was overall low (mean = 1.3; SD = 5.1). Autonomy was significantly associated with success; fellows with 100% autonomy had over 5 times the odds of success compared to those with 50% autonomy (odds ratio = 5.54; 95% CI: 0.98 to 26.6; P = 0.036). Training level and perceived difficulty were not statistically significant predictors. Conclusion: Autonomy was the only significant predictor of successful ileal intubation. However, this may reflect supervision bias, as fellows likely received full autonomy only when deemed competent. While year 2 fellows appeared to have lower odds of success, the small number of year 1 cases limits meaningful comparison and suggests possible confounding. Despite these limitations, this fellow-led data tool provides valuable insight into procedural competence and supports the use of real-time trainee-reported metrics in gastroenterology education.