Agreement Between Crohn’s Disease Endoscopic Severity Scores Derived from Live, Video-recorded, and Central Readings

Evidence- and consensus-based recommendations from the International Organization for the Study of Inflammatory Bowel Diseases (IOIBD) and the FDA stress the importance of endoscopic endpoints in both clinical practice and clinical trials.1 Central reading of endoscopy in inflammatory bowel disease (IBD) patients has become the gold standard for inclusion and evaluation of endoscopic endpoints in clinical trials.2 Although several publications have confirmed the good inter- and intra-rater reliability of the Crohn’s Disease Endoscopic Index of Severity (CDEIS) and the Simple Endoscopic Score for Crohn’s Disease (SES-CD),3 and agreement between site and central readings,4 the agreement derived from live local, delayed local video-recorded, and central readings has never been studied. This is relevant since it can be suspected that lesions seen during live endoscopies may be missed on reviewing videotapes. The goal of this study was to assess the agreement among live evaluation, video-recorded and central reading of Crohn’s disease (CD) endoscopic scores, CDEIS, and SES-CD. In a prospective study conducted at Mount Sinai Hospital in New York, Crohn’s disease patients were recruited between April and December 2015. The research ethics board at Mount Sinai Hospital approved the study. Three endoscopists with substantial IBD practices performed live local (L) readings and delayed local video-recorded (D) readings at least 3 months after the L reading. Two central readers (C1 and C2) read all videos. All local readers were not using endoscopic scoring in the past, but underwent standardized training for the CDEIS and SES-CD based on thorough review of published criteria.5, 6 Intraclass correlation coefficients (ICC) and Bland and Altman plots were estimated to assess the agreement between severity scores derived from L, D, C1, and C2 readings.7 The primary outcome was the agreement between severity scores derived from live local (L) and central (C1 and C2) readings. Secondary outcomes were the agreement between severity scores derived from live local (L) and delayed local (D) readings and severity scores derived from D and C1 and C2. A total of 45 CD patients satisfied the inclusion criteria. These patients had a mean age of 41 ± 15 years, disease duration of 17 ± 11 years, and HBI of 3.1 ± 3.9. Forty-two percent of patients had an ileocolonic phenotype, and 60% had experienced prior surgery. ICC coefficient estimations were higher than 0.80 for the primary outcome (L vs C1 and L vs C2) and for the majority of secondary outcomes (Table 1). Limits of agreement, measured by mean differences ±2 SD, were close to 10 for our primary outcome (Fig. 1). The scoring items stenosis and affected area were identified as particular areas of disagreement in a subgroup of patients and could explain this wider range. Intra-class Correlations Between Various Readings for the CDEIS and SES-CD C1 – Central Reader 1, C2 – Central Reader 2 Intra-class Correlations Between Various Readings for the CDEIS and SES-CD C1 – Central Reader 1, C2 – Central Reader 2 Blue dots: endoscopist 1; green dots: endoscopist 2; red dots: endoscopist 3. Middle line: mean differences. Upper and lower lines: mean differences ±2 SD. Central readers have become instrumental in clinical trials to reduce variability in interpreting disease severity; however, specific training in the SES-CD and CDEIS for our local readers translated into endoscopic scores that had a high level of agreement with central readers. This could be explained by the lack of inherent bias to overestimate disease severity, which could lead to discrepancy in scoring in clinical trials. Importantly, even if endoscopists involved in this study have substantial IBD practices, they were not using Crohn’s endoscopic scores consistently before participating in this study, which makes our data more generalizable. In conclusion, our results showed that after standardized training in Crohn’s disease endoscopic scoring, the agreement level between local readers and central readers is high in a clinical setting. This should motivate all gastroenterologists involved in the care of IBD to learn and apply these scores in routine clinical practice to improve the quality of endoscopic reports. Also, this data further validates the reliability of central reading in clinical trials. Conflicts of Interest: None of the authors has any conflict of interest to declare. Author Contributions: JCD was involved in the study design, data collection, data analysis, drafting, and critical revision of the letter. FP was involved in the data collection, data analysis, drafting, and critical revision of the letter. KK was involved in the study design. TU, JM, PL, and BC were involved in the data collection and critical revision of the letter. JFC was involved in the study design, data analysis, drafting, and critical revision of the letter. This work was presented at ECCO in Barcelona in February 2017 and at DDW in Chicago in May 2017.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.