Is narrow band imaging superior to high-definition white light endoscopy in the assessment of diminutive colorectal polyps?

Background and Aim: Diminutive polyps measuring <= 5 mm in size constitute 80% of polyps in the colon. We prospectively assessed the performance of high-definition white light endoscopy (hWLE) and narrow band imaging (NBI) in differentiating diminutive colorectal polyps. Methods: In this prospective, multicenter study, videos of 50 diminutive polyps (31 hyperplastic, 19 adenomatous) in hWLE followed by NBI (total 100 videos) were initially obtained and placed in random order into five separate folders (each folder 20 videos). Eight endoscopists were then invited to predict the histology (each endoscopist 100 videos, 800 video assessments in all). Polyps were classified into types 1-3 (hyperplastic) and type 4 (adenoma). Feedback on individual performance was given after each folder (20 videos) was assessed. Results: The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy in differentiating hyperplastic from adenomatous polyps by hWLE (400 videos) and NBI (400 videos) were 67.8%, 90.7%, 81.7%, 82.1%, and 82.0%; and 82.2%, 81.5%, 73.1%, 88.2%, and 81.8%, respectively. In the pretest and post-test analysis, the accuracy with NBI improved markedly from 68.8% to 91.3% (P = 0.001) compared with hWLE, 76.3-78.8% (P = 0.850). Overall, the interobserver agreement was 0.46 for hWLE (moderate) and 0.64 for NBI (good). Conclusions: NBI was as accurate as hWLE in differentiating diminutive colorectal polyps. Once a learning curve was reached, NBI achieved significantly higher accuracies with good interobserver agreement. Using a simplified classification, a didactic learning session and feedback on performance, diminutive colorectal polyps could be predicted with high accuracies with NBI.