Wang CADe
2019
Real-time automatic detection system increases colonoscopic polyp and adenoma detection rates: a prospective randomised controlled study
In 1,058 patients undergoing diagnostic colonoscopy, real-time deep-learning polyp detection increased the adenoma detection rate compared with standard colonoscopy.
Result
- Absolute difference
- 8.8 points
- NNT
- 12
Absolute difference and NNT are derived from the event rates.
Also reported: Adenomas per patient 0.53 vs 0.31 (P<0.001). Gain driven by diminutive adenomas (185 vs 102; P<0.001) with no difference in larger adenomas (77 vs 58; P=0.075); hyperplastic polyps also increased (114 vs 52; P<0.001).
- Subtopic
- Adenoma detection
- Design
- Open-label RCT
- Relevance
- Important: one of several pillars.
- Source
- Wang P, et al. Gut. 2019;68(10):1813-1819.doi 10.1136/gutjnl-2018-317500
Discussion & critique
First randomised trial of real-time CADe and the proof of concept for the whole cluster. Open, non-blinded design in a low-ADR population; the gain came almost entirely from diminutive adenomas and hyperplastic polyps, so the clinical value, and the cost of the extra resections, was left for later trials to settle.