Identification of visible lesions during surveillance endoscopy for Barrett's esophagus: a video-based survey study.

Background And Aims: Visible lesion (VL) detection is essential in patients with Barrett's esophagus (BE). We sought to assess the rate of VL detection by academic and community endoscopists using high-definition white-light endoscopy (HD-WLE) and narrow-band imaging (NBI) during surveillance endoscopy.

Methods: Fifty endoscopists were invited to participate in a prospective video survey study. Participants viewed 25 standardized clips of patients referred for endoscopic therapy. Participants noted identification of anatomic landmarks and VLs using HD-WLE and NBI and reported practice-level data. The criterion standard of VL identification was established by consensus of 5 BE experts. Our primary outcome was the rate of VL identification using HD-WLE and NBI.

Results: Forty-four of 50 participants completed the study (22 academic and 22 community). Compared with the criterion standard, participants did not identify 28% (HD-WLE) and 31% (NBI) of VLs. Community endoscopists had more experience (>5 years in practice: community 85% vs academic 54.5%, P = .041; >5 surveillance endoscopies a month: community 85% vs academic 31.8%, P = .046). Across all participants, VL detection using NBI improved significantly with a minimum of 5 surveillance endoscopies per month (area under the curve = .72; 95% confidence interval, .56-.85; P = .006).

Conclusions: Despite improved endoscope resolution and availability of virtual chromoendoscopy, the overall rate of VL detection remains low. Identification of VLs using NBI may be volume dependent. Further education and training efforts focused on VL detection during BE surveillance endoscopy are needed.