Tu1157 Gastroenterology Trainees Can Perform Real Time Optical Diagnosis of Diminutive Colorectal Polyps Using Narrow Band Imaging

BACKGROUND Proposed clinical thresholds for optical diagnosis of colorectal polyps recommend ≥90% agreement between opticaland histopathology-based surveillance intervals, and ≥90% negative predictive value in the rectosigmoid colon. Experienced gastroenterologists have exceeded these thresholds in studies, but the performance of trainees is unknown. We aimed to prospectively assess real-time trainee performance of optical diagnosis of diminutive colorectal polyps. METHODS We studied the optical diagnoses of colorectal polyps using narrow band imaging (NBI) in gastroenterology trainees over a 9-month period during a prospective, randomized polyp detection tandem colonoscopy trial. Trainees viewed a computer-based learning module on NBI International Colorectal Endoscopic (NICE) classification and took a test that required minimum 90% accuracy prior to the study. For each polyp found at colonoscopy, trainees stated the diagnosis (neoplastic vs. non-neoplastic) and confidence level in the diagnosis (high vs. low), then removed and submitted it for histopathologic diagnosis. Central, blinded pathology was the reference standard. An optical diagnosis was high confidence if the polyp had at least one of the features of neoplasm or non-neoplasm and none of the features of the opposing diagnosis. Otherwise, low confidence