Document resource
Objective Accurate real-time assessment of cancer risk in large (≥20 mm) non-pedunculated colorectal polyps (LNPCPs) guides endoscopic versus surgical management. We evaluated whether a structured, interactive e-learning intervention improves cancer detection in LNPCPs among general endoscopists.Design/method Early-career gastroenterologists (≤3 years qualified) and trainees in Flanders (Belgium) completed two online assessments of 20 standardised adenomatous LNPCP videos, separated by a 90-minute interactive e-learning session covering overt (demarcated area) and covert (size, granularity, location, Paris) cancer features. Videos were shown in random order at each assessment without histopathology feedback. Diagnostic outcomes (accuracy, sensitivity, specificity) were estimated using mixed-effects logistic models with a random intercept for reader; 95% CIs and two-sided p values are reported.Results 17 participants (13 trainees and four early-career gastroenterologists) completed all assessments (680 total assessments). Following the e-learning intervention, accuracy improved significantly from 50.4% (95% CI 42.6% to 58.2%) to 70.3% (95% CI 63.1% to 76.7%, p<0.001), sensitivity from 69.1% (95% CI 57.2% to 78.9%) to 88.2% (95% CI 78.2% to 94.0%, p=0.008) and specificity from 45.5% (95% CI 37.4% to 53.8%) to 65.6% (95% CI 57.5% to 72.9%, p<0.001). Gains were concentrated among less experienced endoscopists (≤200 colonoscopies ever), with performance rising to that of more experienced peers. Non-confident assessments decreased from 14.1% to 0.9%.Conclusion A single 90-minute structured e-learning intervention significantly enhanced accuracy, sensitivity and specificity of cancer prediction in LNPCPs among general endoscopists, particularly those with limited experience. These findings highlight the potential of online training programmes to optimise endoscopic decision-making, reduce unnecessary surgeries for benign lesions and improve patient outcomes.Trial registration number NCT05699954.