While the use of artificial intelligence (AI) during colonoscopy may contribute to improved cancer prevention, it may also add to patient burden in terms of increased colonoscopy frequency and, in turn, health care costs, a new study suggests.
The study, published online in Clinical Gastroenterology and Hepatology, found that colonoscopy plus AI (vs. colonoscopy alone) increased the proportion of patients requiring intensive postpolypectomy colonoscopy surveillance by roughly 35% in the United States and Japan and by about 20% in Europe.
“It’s certainly possible that AI will lead to more frequent surveillance for some patients, but it may balance itself out given that recent surveillance guidelines have pushed off the surveillance interval depending on the size of the polyp,” senior author Seth A. Gross, MD, professor of medicine and clinical chief of the division of gastroenterology and hepatology at New York University Langone Health, said in an interview.
Impact on intensive colonoscopy surveillance
AI tools have been shown to increase the adenoma detection rate (ADR) during colonoscopy, but what impact this has on the frequency of surveillance colonoscopy is unknown.
To investigate, Dr. Gross and an international team conducted a pooled analysis of nine randomized, controlled trials comparing colonoscopy with or without AI detection aids. Five trials were done in China, two in Italy, one in Japan, and one in the United States.
The primary outcome was the proportion of patients recommended to undergo intensive surveillance (that is, 3-year interval).
Among a total of 5,796 patients (mean age, 53 years; 51% male), 2,894 underwent AI-assisted colonoscopy and 2,902 underwent standard colonoscopy without AI assistance.
Higher ADRs in the AI-assisted colonoscopy groups were observed in all of the trials.
When following the U.S. and Japanese guidelines, the proportion of patients recommended for intensive surveillance increased from 8.4% (95% confidence interval, 7.4%-9.5%) in the non-AI group to 11.3% (95% CI, 10.2%-12.6%) in the AI group, which is an absolute difference of 2.9% (95% CI, 1.4%-4.4%) and a risk ratio of 1.35 (95% CI, 1.16-1.57). When following the European guidelines, the increase was from 6.1% (95% CI, 5.3%-7.0%) to 7.4% (95% CI, 6.5%-8.4%), which is an absolute difference of 1.3% (95% CI, 0.01%-2.6%) and a RR of 1.22 (95% CI, 1.01-1.47).
The increases are primarily the result of reallocation of patients from low-risk to intermediate- or high-risk categories, the investigators said. That shift is likely caused by the AI-related increase in adenomas per colonoscopy, which may lead to more effective cancer prevention.
“AI does show us that there’s always an opportunity for improvement when we do screening and surveillance colonoscopy,” Dr. Gross said. “The goal is the same, which is to offer the highest quality colonoscopy exam and the best possible outcome for our patients, and I think that’s what we’re starting to see.”
Cost analysis needed
Dr. Gross noted that he sees a cost-effectiveness analysis of AI in colonoscopy in the future.
“As this becomes more and more part of regular clinical practice, if it’s not being done already, someone will look at the cost-effectiveness of incorporating AI, just like they would for other technologies that come into the area of endoscopy,” Dr. Gross said.
Commenting on the study, Atsushi Sakuraba, MD, PhD, a gastroenterologist with University of Chicago Medicine, said he believes that the “benefit of improved adenoma detection and resulting reduction in colon cancer would outweigh the downsides of increased colonoscopy frequency and cost.”
“However, an economic impact study would need to be performed to answer this question,” added Dr. Sakuraba, who wasn’t involved with this study.
The study had no specific funding. Dr. Gross has served as a consultant for Olympus, Cook, Pentax, Ambu, and Iterative Scopes, and served on the advisory board for Docbot. Dr. Sakuraba reported no relevant financial relationships.
A version of this article first appeared on Medscape.com.