Community Gastroenterologists Can Easily Learn Characterization of Diminutive Colorectal Polyps With Narrow Band Imaging (NBI) Amit Rastogi*, Neil Gupta, Deepthi S. Rao, Sachin Wani, Prateek Sharma, Ajay Bansal, Scott R. Ketover, John I. Allen Gastroenterology, University of Kansas School of Medicine, Kansas City, KS; Gastroenterology, Veterans Affairs Medical Center, Kansas City, MO; Gastroenterology, Washington University in St. Louis, St. Louis, MO; Gastroenterology, University of Colorado, Denver, CO; Gastroenterology, Minnesota Gastroenterology, Bloomington, MN Background: Real time endoscopic assessment of histology of diminutive polyps ( 5mm) has the potential for enormous cost saving. For the ultimate goal of community gastroenterologists (GI) implementing this in practice, training tools have to be developed and the technology used should be feasible, and easy to learn. Aim: To evaluate the impact of a computer based teaching module on the performance of community GI in the characterization of diminutive polyps using NBI video clips. Methods: A 20 minute audiovisual Power Point presentation was created detailing the previously validated endoscopic features differentiating adenomas from hyperplastic polyps using NBI. To simulate clinical practice, 80 high quality NBI video clips of diminutive colorectal polyp were selected and distributed into two sets (pre test and post test) of 40 videos each (26 adenomas and 14 hyperplastic) by a computer generated random sequence stratified by polyp histology. 15 gastroenterologists in a large community practice group participated in the study. Each participant was provided with the video clips in a Pre test and Post test DVD along with the teaching module. They first reviewed the pre test video clips and entered their responses for polyp histology and the confidence in diagnosis: high 90% or low 90%. Following this, they reviewed the teaching module and then assessed the post test videos. Performance characteristics were calculated for both pre and post test by comparing predicted histology with actual histology. Fisher’s exact test was used for analysis and kappa statistic was calculated for interobserver agreement. Results: 15 gastroenterologists in community practice with varying levels of experience (1-28 years) and performing 300 1200 colonoscopies annually completed the study. Sensitivity, specificity and accuracy in characterization of polyp histology improved significantly in post test compared to pre test (Table 1). Furthermore, performance when diagnosis was made with high confidence was superior compared to low confidence in both sets (Table 2). In post test, participants had an accuracy of 92% for high confidence diagnoses and the proportion of these increased with training from 46% (pre test) to 64% (post test); p 0.001. Interobserver agreement for diagnosis improved from fair (kappa 0.23) in pre test to moderate (kappa 0.56) in the post test. Conclusions: Community GI with varying levels of experience can learn polyp histology characterization by NBI with the help of a short computer based teaching module. Their accuracy for high confidence predictions in post test ( 90%) was comparable to that reported for experts in NBI. These results suggest that real time endoscopic assessment of diminutive polyps using NBI could be incorporated into community clinical practice with similar expected results compared to expert centers.
Read full abstract