DOP033. Evaluation of site versus central assessment of endoscopies in Crohn's disease: Comparison using data from EXTEND
P. Rutgeerts1, W. Reinisch2, W.J. Sandborn3, G. D'Haens4, J.-F. Colombel5, J. Petersson6, Q. Zhou6, R.B. Thakkar6, 1KU Leuven, Division of Gastroenterology, Leuven, Belgium, 2McMaster University, Department of Medicine, Hamilton, Canada, 3University of California, San Diego, Division of Gastroenterology, La Jolla, United States, 4Academic Medical Center, Department of Gastroenterology, Amsterdam, Netherlands, 5Icahn School of Medicine at Mount Sinai, Division of Gastroenterology, New York, United States, 6AbbVie Inc, GPRD, North Chicago, United States
Interobserver variation in the endoscopic assessment of disease activity has been reported for ulcerative colitis clinical trials . We performed a post hoc analysis to compare the agreement between site and central readings of endoscopies from the adalimumab Crohn's disease clinical trial EXTEND .
All randomised patients (pts) from EXTEND who had endoscopies reviewed by one central reviewer and by site reviewers were included in this analysis. Endoscopies were performed at baseline (BL), week (wk) 12, and 52 and scored using SES-CD at all 19 sites and CDEIS at six sites. Agreement between site and central reading was determined by calculating the kappa (κ) coefficient for categorical variables and the Pearson (r) coefficient for continuous variables.
Of the 129 randomized pts in EXTEND, 52 pts at BL, 49 at wk 12, and 34 at wk 52 had CDEIS scored endoscopies by both the one central reviewer and the site reviewer, whereas 129, 122, and 84 pts, respectively, had SES-CD scored endoscopies by both reviewers. A high degree of agreement on total CDEIS score or total SES-CD score occurred between central and site reviewers at all three time points (Table). Site and central readers showed good to very good agreement (κ [or r] = 0.6–1.0) in scoring CDEIS deep ulceration and surface involved by the disease at all three time points and in all segments, except right colon, where fair to moderate agreement was observed regardless of time point (κ [or r] = 0.4–0.6). The weakest agreement between site and central reading was observed at BL for CDEIS ulcerated surface score and SES-CD large ulceration score compared to the agreement at wks 12 and 52 (CDEIS ulcerated surface r = 0.67, 0.88, 0.89 at BL, wk 12, wk 52; SES-CD large ulceration κ = 0.04, 0.53, 0.42 at BL, wk 12, wk 52, respectively). A high degree of correlation was observed between central and site reading in the assessment of change in total CDEIS or SES-CD score from BL at wks 12 and 52, with stronger agreement for reductions at wk 52.
|Pearson coefficient||Total score,|
|Central assessment||Site assessment||Central assessment||Site assessment|
Overall, very good agreement between site and central reading of endoscopies occurred in EXTEND, with the weakest agreement at study BL.
1. Feagan, B et al. 2013. Gastroenterol. 145:149–57.
2. Rutgeerts, P, et al. 2012. Gastroenterol. 142:1102–11.