Endoscopists without prior training achieve ASGE PIVI thresholds with NBI

WASHINGTON — Academic endoscopists without prior training in narrow band imaging were able to achieve the American Society for Gastrointestinal Endoscopy PIVI thresholds for optical diagnosis of diminutive colorectal polyps, according to data presented at Digestive Disease Week.

“Experts have demonstrated a high level of performance in using narrow band imaging (NBI) to determine diminutive polyp histology,” Swati G. Patel, MD, from University of Michigan, said during her presentation. “However, it is unclear if endoscopists without prior training in NBI can achieve a similar level of performance. The primary aim of our study was to determine whether academic endoscopists with no prior training in NBI can achieve the ASGE PIVI thresholds recently presented for the NBI optical diagnosis of diminutive colorectal polyps.”

The ASGE PIVI thresholds recommend endoscopists achieve 90% or greater negative predictive value for adenomas in the rectosigmoid and 90% or greater agreement in predicted and actual surveillance intervals for high-confidence predictions.

In the first phase of the study, 26 academic gastroenterologists at two centers were trained on NBI interpretation and achieved greater than 90% accuracy for high-confidence predictions. In the second phase of the study, they made real-time predictions of diminutive polyp histology during routine colonoscopy and predicted surveillance intervals. Their performance was evaluated by comparing predictions with actual histology and surveillance intervals.

A total of 1,403 colonoscopies were performed and surveillance intervals were predicted for 62%. Overall 2,770 diminutive polyps were detected, 76% were predicted with high-confidence and 56% were adenomas. Negative predictive value for the 76% of 1,055 rectosigmoid polyps that were predicted with high confidence was 94.7% (95% CI 92.2-96.6%) and agreement in predicted vs. actual surveillance interval was 88.8% (95% CI 86.1-90.4%).

The negative predictive value goal was achieved by 77% of participants, the surveillance interval agreement goal was achieved by 50% and both goals were achieved by 35%. Significant predictors of accurate diagnoses included high-confidence (OR = 3.07; 2.51-3.74) and years of experience (OR = 1.18; 1.05-1.32).

“We can conclude with: attending endoscopists with no prior training with NBI can achieve a 95% negative predictive value in the rectosigmoid and an 89% agreement in surveillance interval with the vast majority of patients being screened on time or early,” Patel said. “There is a critical importance of distinguishing between high- and low-confidence diagnoses, given that high confidence was a strong predictor of performance in our study.”

Streamlined auditing strategies should be developed in the event that resect-and-discard becomes widely implemented “to insure adequate individual performance over time,” she added. “Furthermore, the cost-effectiveness of the strategy may need to be reevaluated, taking into account the higher proportion of low-confidence diagnoses in our study as well as the risk and cost of potentially performing procedures earlier. And finally, … advanced imaging training should be incorporated into fellowship educational curricula as well as continuing medical education.” – by Adam Leitenberger

For more information:

Patel SG, et al. Abstract 448. Presented at: Digestive Disease Week, May 16-19, 2015; Washington, D.C.

Disclosure: Patel reports no relevant financial disclosures. Please see the DDW faculty disclosure index for all other researchers’ relevant financial disclosures.