VANCOUVER, British Columbia—Computer-aided detection in colonoscopy proved to be beneficial for a wide spread of colonoscopy providers, improving adenoma per colonoscopy rates among several subsets of clinicians, according to a new study presented at the 2023 annual meeting of the American College of Gastroenterology.
“Our results show that CADe helps gastroenterologists in both academic and community practices, at all times of the day, with the largest increase for physicians with more than 20 years of experience and those with ADR [adenoma detection rate] lower than 45%,” Aasma Shaukat, MD, the Robert M. and Mary H. Glickman Professor of Medicine at the NYU Grossman School of Medicine, in New York City told Gastroenterology & Endoscopy News.
The detection of adenomas can be influenced by certain endoscopist’s characteristics and, multiple studies have shown, can be enhanced with the use of CADe (World J Gastroenterol 2021;27[47]:8103-8122).
To explore how these factors interrelate, Dr. Shaukat and her co-investigators analyzed the association of provider- and procedure-level factors associated with adenomas per colonoscopy (APC) in the randomized SKOUT registration study of CADe versus standard colonoscopy (abstract P0549). The population included 1,423 participants and 22 endoscopists practicing in community and in academic settings.
Colonoscopy factors included practice type, procedure time of day (before versus after 12 pm), procedure withdrawal time (8-minute threshold) and quality of the bowel prep (inadequate/poor, fair/good and excellent). Endoscopist factors included baseline ADR and years of experience.
During procedures with adequate bowel preparation and withdrawal times of at least 8 minutes, CADe improved APC rates. For physicians with more than 20 years of experience, the APC rate was 1.035 with CADe versus 0.756 with standard colonoscopy without CADe, a difference of 0.279 (95% CI, 0.077-0.484). The difference was smaller for clinicians with 11 to 20 years of experience (0.241; 95% CI, –0.034-0.510). There was no improvement in APC for physicians with one to 10 years of experience (–0.035; 95% CI, –0.367-0.316). The use of CADe improved APC by 0.297 (95% CI, 0.112-0.484) in clinicians with an ADR lower than 45%, compared to only 0.098 (95% CI, –0.172-0.355) in those with ADR of at least 45%, the study found.
Dr. Shaukat said the findings could help practices decide whether to invest in artificial intelligence approaches and could be informative as to which physicians within practices might benefit most from its use.
—Caroline Helwick
Dr. Shaukat reported financial relationships with Freenome, Iterative Heath and Motus GI.
