Abstract
Background: Colonoileoscopy is increasingly used to evaluate Crohn's disease, but the reproducibility of endoscopic findings is not clear. Methods: The interobserver variation of endoscopic findings and the influence of experience on assessments were investigated in 82 colonoileoscopies in Crohn's disease. Results: In colonic assessment there was excellent agreement for most endoscopic features (kappa values <0.75;p < 0.001). In ileal assessment agreement was excellent with regard to detection of large ulcers and strictures and endoscopic staging on the basis of ulcer size and stricture (kappa <0.76; p < 0.001). Observer experience was an important factor in ileal assessments: agreement was excellent for 13 features within experienced pairs, compared with 3 when only 1 of a pair was experienced. In colonic assessment experience was less important. Conclusion: The study showed that acceptable agreement can be obtained on some well-defined inflammatory lesions in Crohn's disease even when investigator experience is limited. Endoscopic staging on the basis of ulcer size and stricture, being excellently reproducible, can serve as a simple summarizing assessment.