Abstract
Abstract Background There is no clear consensus on the definition of endoscopic remission in Crohn’s disease (CD). Indeed, existing endoscopic scoring systems for CD focus primarily on disease activity rather than on remission. The CREDO programme aims to define an objective measure (depth or grade) of endoscopic remission in CD. This inter-reader variation study aims to assess reproducibility in identifying endoscopic lesions that are assumed broadly agreed to be associated with remission. Methods Sixteen centers in Belgium and France were involved in the study of patients with CD in clinical and biochemical remission (CDAI <150 and CRP <5 mg/l and fecal calprotectin <250 µg/g). Lesions potentially related or not to remission in CD were predefined by the principal authors, then discussed and refined during a consensus meeting with 12 central readers (CRs) and the local endoscopists. Fifteen patients per center were planned to be selected with an endoscopy representing equally complete remission, almost complete remission, not complete, nor an almost complete remission according to local endoscopist appraisal. Finally, 210 patients were enrolled with 15 endoscopy videos recorded within each of the 14 centers. The 14 centers and 12 CRs were splitted in two balanced groups. Within each group, readings of the 15 videos per center were randomly assigned to the 6 CRs using a balanced incomplete block design, each video being read by 4 CRs and each CR reading 10 videos. In each group, this design was replicated 7 times leading to 105 videos randomly assigned to 6 CRs, 4 CRs reading each video and 70 videos being read by each CR (efficiency factor of 0.90). Inter-reader reproducibility level of each lesion was evaluated within each group by the intraclass correlation coefficient (ICC)a. Benchmark ICC limits widely accepted are 0.4-0.6, 0.6-0.8, 0.8-1.0 to define moderate, substantial, almost perfect agreement, respectivelyb. Results Highly variable inter-reader reproducibility level was observed for the different lesions in group 1 (see ICC estimate with its 95% confidence interval in table). Level of reproducibility was substantial for passable stenosis, pseudopolyp type white and red, mucosal defect and ulcer. It was moderate for stenosis, ulcerated stenosis, surelevated scarred lesion, raised trabecular scar, erosion with halo, erosion, punctiform lesion, deep ulcer, superficial ulcer and alteration of vascular pattern type white. Results of reproducibility levels estimated in group 2 were similar in spite of less frequent lesions. Conclusion Our inter-reader variation study findings provide a foundation for developing an objective and standardized definition of endoscopic remission in CD. References a/ Fleiss J L. Balanced incomplete block design for intra-rater reliability studies. Applied Psychological Measurement 1981; 5: 105-112. b/ Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33:159-75.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call