Colonoscopy setting and impact on statistical significance in Artificial Intelligence trials

We read with interest the article by Ahmad et al. [1] investigating the impact of artificial intelligence (AI) on adenoma detection in a fecal immunochemical test (FIT)-based screening colonoscopy setting.

In our opinion the colonoscopy setting of the study is the main reason for the lack of statistically significant difference found in the adenoma detection rate (ADR). The FIT-based colonoscopy setting has always been considered a challenging area in which to test technologies aimed at improving the ADR. This is due to the higher prevalence of adenomas expected in this setting [2], reducing the relative statistical weight of any absolute difference that may be found. We previously investigated the impact of using AI in a FIT-based colonoscopy screening program in Italy, obtaining a statistically significant absolute increase of 8 % in ADR in the AI group (53.6 % vs. 45.3 %) [3]. This result appears to be similar to the 6 % absolute difference described in the Ahmad et al. study.

The higher cutoff for FIT positivity in the UK (80–150 µg Hb/g) [4] compared with Italy (20 µg Hb/g) selects a cohort of patients with a higher prevalence of adenomas; furthermore, the extensive use of the mucosal exposure device in the study minimized the risk of missing lesions [5]. Both these factors increase the expected baseline ADR in this study, marginalizing the relative (statistical) impact of AI. This might have been avoided by calculating the sample size based on an appropriate baseline ADR, with consequently a huge number of colonoscopies needed to prove any difference. By calculating the sample size based on the polyp detection rate, as Ahmad et al. did, the impact of AI on adenoma detection performance was probably underestimated.

Publication History

Article published online:
28 November 2023

• References

• 1 Ahmad A, Wilson A, Haycock A. et al. Evaluation of a real-time computer-aided polyp detection system during screening colonoscopy: AI-DETECT study. Endoscopy 2023; 55: 313-319
• 2 Hassan C, Piovani D, Spadaccini M. et al. Variability in adenoma detection rate in control groups of randomized colonoscopy trials: a systematic review and meta-analysis. Gastrointest Endosc 2023; 97: 212-225.e7
• 3 Rondonotti E, Di Paolo D, Rizzotto ER. et al. AIFIT Study Group. Efficacy of a computer-aided detection system in a fecal immunochemical test-based organized colorectal cancer screening program: a randomized controlled trial (AIFIT study). Endoscopy 2022; 54: 1171-1179
• 4 Cancer Research UK. Bowel cancer screening programmes at a glance. 2023 Available at (Accessed: July 11, 2023): https://www.cancerresearchuk.org/sites/default/files/bowel_cancer_screening_programmes_at_a_glance_2023.pdf
• 5 Patel HK, Chandrasekar VT, Srinivasan S. et al. Second-generation distal attachment cuff improves adenoma detection rate: meta-analysis of randomized controlled trials. Gastrointest Endosc 2021; 93: 544-553.e7