Impact of Artificial Intelligence on Trainee Polyp Miss Rates

Artificial Intelligence (AI), Colonoscopy, Colon Polyps

tandem colonoscopy, randomized controlled trial, adenoma miss rate

Based on prior studies, trainee and practicing gastroenterologists miss pre-cancerous polyps (adenomas and serrated polyps) during colonoscopy. The use of computer-aided detection (CADe) systems, a form of artificial intelligence (AI) has been shown to help identify colorectal lesions for practicing gastroenterologists. However, less is known how AI impacts polyp detection for trainees. The investigators are conducting a tandem colonoscopy study wherein a portion of the colon is examined first by the trainee and then the attending physician. For each procedure, randomization will occur which will determine whether or not the trainee will utilize AI for their examination of the colon. At the end of the study, the investigators will determine whether AI helps trainees miss fewer polyps during colonoscopy. The investigators will also conduct interviews with trainees to understand how AI impacts colonoscopy training.

Quality metrics in colonoscopy are useful in determining endoscopist performance. The adenoma detection rate (ADR) has historically been the primary colonoscopy quality indicator of interest, whereby those with a higher ADR have lower rates of interval colon cancers. Other important parameters include the adenoma miss rate (AMR) and the sessile serrated polyp miss rate (SMR). The AMR among practicing gastroenterologists is estimated at around 25% based on tandem-colonoscopy studies, while the SMR may be even higher. Prior tandem colonoscopy studies have also shown that gastroenterology trainees have an AMR between 30-41% and an SMR between 56-86%. As missed polyps may grow into colorectal cancers over time, efforts must be dedicated to reducing AMR and SMR. In 2021, the Food and Drug Administration (FDA) approved the first computer-aided detection (CADe) system, (GI-Genius; Medtronic, Minneapolis, MN) for identifying colorectal lesions in the United States. An initial tandem colonoscopy study among experienced endoscopists using this CADe system showed a reduction in AMR from 32.4% to 15.5%. However, it is unknown whether trainees would have a similar reduction in polyp miss rates. CADe has been introduced into gastroenterology trainee education in a largely unstructured fashion and trainees' perception of CADe's impact on training is largely unknown. Presently, trainees and attendings complete colonoscopies with and without CADe. It is at the discretion of attendings whether CADe is used for a procedure regardless of whether or not a trainee is involved. In this single center study, all Northwestern gastroenterology trainees will be invited to participate in the trial. First year gastroenterology trainees will initiate participation after 6 months of endoscopic training (approximately 100 colonoscopies completed). Trainees will be complete an hour-long meeting with the study team. The study protocol will be explained in detail and written informed consent will be obtained from the trainees. The medical records of patients scheduled for ambulatory screening or surveillance colonoscopy during select endoscopy blocks will be screened in advance for inclusion in the study. Participants will be prospectively enrolled. On the day of the patient's scheduled colonoscopy, a member of the study team will obtain written informed consent for both the colonoscopy and participation in the research study in a private bay in the pre-endoscopy area. The patient will then be transported to the procedure room and sedated. Sedation method (monitored anesthesia care or moderate sedation) will be determined prior to the scheduled appointment. If the patient is to receive moderate sedation, fentanyl and midazolam will be administered in incremental doses per the standard of care. Once the patient is sedated, the fellow and attending will perform the insertion as is standard for a colonoscopy in which a fellow is participating. Once the cecum is reached, the patient will then be randomized to whether the initial inspection of the right colon by the fellow is performed with or without the assistance of CADe in a 1:1 randomization. The fellow withdrawal (pass #1) will start. The fellow, with or without the use of CADe based on randomization, will identify polyps independently of the attending and remove them with attending assistance as appropriate until they feel that they have performed a complete examination of colonic mucosa from the cecum to the splenic flexure. During pass #1, the supervising attending will not provide verbal cues regarding the identification of colon polyps as would typically occur during colonoscopies performed by a fellow, however, the fellow can request attending assistance at any time. Additionally, the attending will provide standard assistance with insertion and polypectomy techniques. While gastroenterology fellow participation in colonoscopy with attending supervision is standard at an academic medical center, pass #1 will be considered research for the purposes of this study as the attending will not be providing verbal cues regarding polyp detection as the purpose of this study is to quantify trainee miss rates independent of attending input. Once the fellow completes pass #1, the attending physician will reinsert the colonoscope to the cecum and perform a second withdrawal (pass #2) with the assistance of the CADe system. A second proximal withdrawal time for pass #2 will be recorded until the splenic flexure is again reached and the attending feels that the colonic mucosa has been fully examined. Any polyps identified and resected by the attending physician during pass #2 will be considered missed polyps. The fellow and attending will then complete the exam from the splenic flexure to the rectum with the assistance of CADe, both providers identifying and resecting polyps as necessary, as is standard when a fellow performs colonoscopy with attending supervision.

No related papers found yet.

United States