An endoscopic training model to improve accuracy of colonic polyp size measurement

Abstract

Purpose: Most studies of colonic polyps rely on visual estimation when regarding polyp size; however, the reliability of a visual estimate is questionable. Our study aims to develop a training model to improve the accuracy of size estimation of colonic polyps in vivo.

Methods: Colon polyps were recorded on 160 video clips during colonoscopy. The size of each polyp was estimated by visual inspection and subsequently measured with a flexible linear measuring probe. The study included a pretest, an intervention, and a posttest. The pretest included 160 video clips, which comprised the visual-estimation portion of the study. The intervention was an educational model consisting of 30 video clips which included a visual-estimation section and a linear-measuring-probe section, designed to help the endoscopists to compare their visual estimate of size with the measured size of the polyps. The posttest included the 160 video clips used in the pretest, presented in random order. Intraobserver agreement and diagnostic accuracy were compared before and after the training session.

Results: Eight beginners and four experienced colonoscopists were enrolled. The overall kappa (kappa) values of intraobserver agreement for pretest and posttest were 0.74 and 0.85 for beginner group as well as 0.83 and 0.88 for experienced group, respectively. The overall diagnostic accuracy improved from 0.52 to 0.78 for beginner group and 0.71 to 0.87 for experienced group (P < 0.05) after education with the training model.

Conclusions: This training model could help endoscopists improve the accuracy of measurement of polyps on colonoscopy in a short period. The durability of learning effect needs further investigation.