Clinical validation and accuracy assessment of the Capsule Endoscopy‐Crohn's Disease index (CE‐CD)

Abstract Objectives To compare the recently proposed Capsule Endoscopy‐Crohn's Disease index (CE‐CD) to pre‐existing capsule endoscopy (CE) scores, to measure its precision and accuracy to predict adverse clinical outcomes in children with Crohn's disease (CD). Methods Children with CD who underwent CE at diagnosis and had, at least, 1‐year follow‐up postprocedure were selected. Capsule study was viewed and the different indices were independently scored by two trained paediatric gastroenterologists. The relationship between pre‐existing scores and CE‐CD was assessed by linear regression analysis. Clinical outcomes prediction assessment was based on receiver operating characteristics curves, survival analysis and Cox regression. Finally, interobserver agreement was measured. Results Fifty‐nine patients were finally included. CE‐CD showed a strong positive correlation with the Lewis score ( ρ = 0.947) and the Capsule Endoscopy Crohn's Disease Activity Index (CECDAI) ( ρ = 0.982). Both CE‐CD and CECDAI were significant predictors of treatment escalation (hazard ratio 1.07 and 1.09, respectively, with both p ‐values < 0.01). However, no score predicted risk of hospital admission, surgery or clinical/endoscopic relapse. The presence of moderate‐to‐severe small bowel (SB) inflammation, defined as a score of ≥9 on CE‐CD, provided a hazard ratio of treatment escalation of 2.6 (95% confidence interval: 1.3–5.3). This cut‐off provided the optimal sensitivity/specificity pair: 48.4%/89.3%. No interobserver misclassification among inflammation categories given by CE‐CD were observed (kappa 100%). Conclusion CE‐CD is a useful tool to document SB inflammation in children with CD. It correlates strongly with classical scores, can better predict need for treatment escalation and shows good interobserver agreement.