By Prateek Sharma, MD, with Jillian Mock
In this month’s column, I cover three studies on different topics: the rate of detection of adenomas and sessile serrated lesions in the colon after a positive stool test, inappropriate use of proton pump inhibitors, and the use of artificial intelligence to improve the quality of colonoscopy.
In the first study, researchers examining colonoscopy after a positive stool test—either the fecal immunochemical test (FIT) or multitarget stool DNA (mt-sDNA) test—found that endoscopists who had lower adenoma detection rates (ADRs) were more likely than endoscopists with higher ADRs to say the test was a false positive, which the researchers reported could indicate missed polyps. The researchers suggested that ADRs should be at least 40% in patients who have a positive stool test and the clinically significant serrated polyp detection rate (CSSDR) should be at least 20% in those with a positive mt-sDNA test.
PPIs are prescribed frequently for the treatment of upper GI disorders. However, many patients may continue to take them for inappropriate indications, as was shown in this second study I highlight. This review urges clinicians to reevaluate the indications for and dosing of PPIs when seeing patients in the clinic.
In the third study, researchers found that AI can automate the evaluation of the Boston Bowel Preparation Scale. AI determined that inadequate bowel prep was associated with a high adenoma miss rate (AMR), demonstrating that AI tools can be associated with clinically relevant outcomes.
Detection Skills
Am J Gastroenterol 2024 May 10. doi:10.14309/ajg.0000000000002817
Researchers in the United States examined how the skill level of the endoscopist performing a colonoscopy after a positive stool-based test affected whether or not they determined the stool test result was false.
They obtained data from the New Hampshire Colonoscopy Registry between February 2015 and June 2023. The sample included 864 patients with a positive mt-sDNA result and 497 patients with a positive FIT, followed by a colonoscopy. They also divided the 65 endoscopists who performed the colonoscopies into four quartiles based on their ADR and CSSDR. Endoscopists in the fourth quartile had the highest detection rates and those in the first quartile had the lowest.
The results showed that endoscopists with higher ADR and CSSDR had a significantly lower frequency of negative colonoscopies after a positive stool test. For endoscopists in the fourth ADR quartile, the detection rate for any adenoma was 63.3% after a positive FIT and 62.8% after a positive mt-sDNA test. Endoscopists in the fourth CSSDR quartile found sessile serrated lesions (SSLs) in 29.2% of exams after a positive mt-sDNA test and 13.5% of exams after a positive FIT. In comparison, the ADR was 48.7% following mt-sDNA tests and 35.8% following FIT in the first quartile, and the CSSDR rates were 12.8% and 10.2%, respectively.
Endoscopists with lower ADR and CSSDR were more likely to determine the stool test results were a false positive after a negative colonoscopy. This finding implies that important polyps may have gone undetected in exams performed by endoscopists in the lowest quartiles, the researchers reported.
Based on these findings, the researchers concluded that a benchmark (ADR =40% for either FIT or mt-sDNA test or CSSDR =20% for mt-sDNA) could help endoscopists gauge whether they are missing polyps in patients with positive stool tests.
PPI Use
Gut 2024 Apr 26. doi:10.1136/gutjnl-2024-332154
In this systematic review and meta-analysis, researchers in India conducted a global literature survey of studies of the inappropriate overuse of PPIs.
The review included 79 studies with a total of 20,050 patients. The researchers found no significant differences based on study design, setting, sample size estimation, study quality or other factors related to the studies themselves.
Ultimately, the researchers concluded that PPI overuse is a global problem and that in about 60% of cases, PPIs were prescribed without a valid indication (95% CI, 0.55-0.65; I2=97%). Overprescribing can increase costs, pill burden and patients’ risks for adverse effects.
To solve this problem, the study authors suggested adopting PPI stewardship practices, similar to those for antibiotics, and identifying effective deprescribing strategies.
AI in Bowel Preparation
Gastrointest Endosc 2024 Apr 16. doi:10.1016/j.gie.2024.04.015
In this prospective, single-center observational study, researchers in China evaluated the usefulness of AI in evaluating bowel preparation before a repeat colonoscopy.
The study was conducted at Eighth Affiliated Hospital at Sun Yat-sen University, in Guangzhou, China, from Oct. 8, 2021, to Nov. 9, 2022. An AI system evaluated bowel preparation using an e-Boston Bowel Preparation Scale (e-BBPS) developed by the study investigators, while the endoscopist performing the colonoscopy evaluated preparation using the BBPS. If both the BBPS and the e-BBPS scores—only visible to a research assistant—deemed preparation to be adequate, the patient immediately received a second colonoscopy. If not, the patient underwent a re-cleansing before the second colonoscopy.
Among the 393 patients included in the analysis, endoscopists detected 72 adenomas larger than 5 mm but missed 27 adenomas of that size. The AMR for adenomas 5 mm or larger was significantly higher in patients the AI system deemed to be unqualified based on inadequate bowel preparation, than in patients the AI system deemed qualified (35.71% vs. 13.19%; odds ratio, 0.2734; 95% CI, 0.1139-0.6565; P=0.0056).
The researchers concluded that the e-BBPS’s bowel assessments reached expert levels and were largely in agreement with the assessments of experienced endoscopists. They wrote that the study supports the widespread adoption of AI assessments for bowel preparation, but more studies are needed to determine the global application and effectiveness of such a system.
Dr. Sharma is a member of the Gastroenterology & Endoscopy News editorial board.
This article is from the July 2024 print issue.