Patients with inflammatory bowel disease (IBD) have a higher risk of dysplasia/neoplasia, requiring intensified surveillance colonoscopies. Colitis associated neoplasia (CAN) has flatter morphology, making it harder-to-detect. Serrated adenomas in routine screening colonoscopy have similar flat morphology. How neoplasia detection rates during IBD surveillance exams change based on baseline average risk colonoscopy performance is not known.  Also, while there are well defined quality metrics for adenoma detection rate (ADR) and sessile serrated lesion detection rate (SSDR) in non-IBD colonoscopy, these metrics do not exist in IBD surveillance. We hypothesized that endoscopists with higher non-IBD SSDR have higher ADR and SSDR in IBD surveillance colonoscopies, while maintaining the same ADR trend. Thus, we have conducted a retrospective study to investigate baseline ADR and SSDRs compared to adenoma and serrated lesions detection rates during IBD surveillance colonoscopy.

We performed a retrospective study of patients who underwent colonoscopy for an indication of IBD surveillance. A validated natural language processing (NLP) model was then used to report adenomas and sessile lesions found within pathology reports. The number of IBD surveillance colonoscopies per endoscopist was reviewed, and the top quartile, by procedure volume, was included for additional evaluation. Among this quartile, the baseline ADR and SSDR for average risk colonoscopies were calculated. To evaluate whether IBD ADR and SSDR varied across screening quartiles, we fit linear models with quartile as the predictor and diagnosis rate as the outcome. Overall significance was based on the likelihood ratio test and parameter estimates were examined to assess differences between quartiles.

Our study included 56 endoscopists and 5093 IBD surveillance colonoscopies performed between 2017-2023. Physicians in the top quartile were those performing greater than 24 IBD surveillance colonoscopies in this time period. There were significant differences in IBD ADRs among physicians when stratified by screening ADR quartiles (p-value <0.0001). The ADR was 10% among IBD surveillance colonoscopies for the lowest ADR quartile, and this rate was similar for the second and third quartiles. However, the diagnosis rate was 10% higher for the fourth quartile of endoscopists when compared to the lowest quartile (p-value <0.0001). The mean baseline non-IBD screening ADR was 37% (SD 8.3) and the mean baseline screening SSDR was 13% (SD 5.3%). First quartile endoscopists had a baseline non-IBD ADR and SSDR of 27.2% and 6.5%, respectively. Fourth quartile endoscopists had a baseline non-IBD ADR and SSDR of 48.1% and 20%, respectively. Regarding SSDR in IBD surveillance, the differences in SSDR among IBD cases between the physicians based on screening ADR quartiles was significant (p-value <0.001). SSDR in IBD cases was 3% in first quartile physicians, similar to second and third quartiles. However, fourth quartile physicians had a higher rate of SSDR, 6% higher than first quartile physicians (p-value 0.0003). 

Endoscopists with the highest screening ADR and SSDR also achieved significantly higher detection of adenomas and sessile serrated lesions during IBD surveillance. While lower quartiles showed similar performance, top quartile physicians detected 10% more adenomas and 6% more sessile lesions. Thus, these findings suggest that provider-level detection performance in screening colonoscopy meaningfully translates to IBD surveillance quality, given the improved detection of subtle, flat dysplasia typical of IBD.