Observer agreement in endoscopic assessment of ulcerative colitis

Abstract

Background: Colonoscopy is the investigation of choice to evaluate ulcerative colitis, but the reliability of the assessment of endoscopic signs is not clear.

Aims: The aim of this study was to evaluate interobserver agreement for the identification of endoscopic lesions typical of ulcerative colitis, and the influence of training.

Material and methods: Four experienced observers and 11 endoscopists under training assessed 49 still images selected from endoscopic video recordings.

Results: The agreement rate between experienced observers was excellent or good (k > 0.39) for recognition of 10 out of 14 signs or patterns (loss of vascular pattern, erythema, oedema, granular mucosa, blood, pseudopolyp, erosion, ulcer, normal pattern, severe activity), and was poor for pus, stricture, mild activity, moderate activity. The rates between endoscopists under training were excellent or good for 6 items (loss of vascular pattern, erythema, oedema, pseudopolyp, normal pattern, severe activity).

Conclusions: Trained observers can reproducibly record most endoscopic signs of ulcerative colitis. A reliable overall scoring of severity should be based on a simple three-grading scale, i.e. normal pattern, moderate activity, severe activity. Acceptable agreement rates can be obtained by endoscopists under training on some well-defined endoscopic appearances.