Observer agreement in endoscopic assessment of ulcerative colitis

Background. Colonoscopy is the investigation of choice to evaluate ulcerati ve colitis, but the reliability of the assessment of endoscopic signs is no t cleat: Aims. The aim of this study was to evaluate interobserver agreement for the identification of endoscopic lesions typical of ulcerative colitis, and th e 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 rare between experienced observers was excellent or good (k>0.39) for recognition of 10 out of 14 signs or patterns (loss of va scular pattern, erythema, oedema, granular mucosa, blood, pseudopolyp, eros ion, ulcer normal pattern, severe activity), and was poor for pus, strictur e, mild activity moderate activity. The rates between endoscopists under tr aining were excellent or good for 6 items (loss of vascular pattern, erythe ma, oedema, pseudopolyp, normal pattern, severe activity). Conclusions. Trained observers can reproducibly record most endoscopic sign s of ulcerative colitis. A reliable overall scoring of severity should be b ased on a simple three-grading scale, i.e. normal pattern, moderate activit y severe activity Acceptable agreement rates can be obtained by endoscopist s under training on some well-defined endoscopic appearances.