Abstract
BACKGROUND AND AIMS
Accurate endoscopic assessment in Crohn's disease (CD) is crucial, influencing treatment decisions and prognosis. However, endoscopic scores are complex and not easily applicable in routine practice. This study aimed to assess interobserver reproducibility of CDEIS, SES-CD and their subsections, in inflammatory bowel disease (IBD) experts and non-experts, to improve future endoscopic assessment.
METHODS
Observational, prospective study including CD patients who underwent a routine colonoscopy in an IBD unit, excluding patients with poor preparation (Boston<6). 7 endoscopists from 4 centers, 4 specialized in IBD, independently scored the videos using CDEIS and SES-CD. Inter-observer variability was assessed, comparing between IBD experts and non-experts, and correlating endoscopic scores with clinical activity and biomarkers.
RESULTS
Overall ICC was 0.835 for CDEIS and 0.772 for SES-CD, indicating substantial agreement. The lowest correlations were deep ulcers in ileum, descending colon, and rectum (CDEIS), and ulcer size in ileum and detection strictures in descending colon (SES-CD). Non-IBD experts showed a highest interobserver agreement (ICC: 0.915 for CDEIS, 0.881 for SES-CD) compared to IBD experts (ICC: 0.795 for both). No significant correlation was found between endoscopic scores and Harvey-Bradshaw index or CRP. SES-CD showed significant correlation with fecal calprotectin (0.582, p=0.011) and CDEIS trended towards significance (r=0.445, p=0.064). 22% of patients were inconsistently classified regarding endoscopic remission.
CONCLUSIONS
Although CDEIS and SES-CD are highly reproducible without specialized training, they have limitations detecting mild inflammatory activity. An ideal score should emphasize ileal activity and simplify the assessment of ulceration severity and stenosis to improve clinical utility.
