Introduction:
Adenoma detection rate (ADR) is the most important quality indicator for colonoscopy and is directly associated with reduced colorectal cancer risk. Although randomized trials have shown that artificial intelligence–based computer-aided detection (CADe) improves ADR, its effectiveness in routine clinical practice has remained uncertain. This large real-world quality improvement study evaluated the impact of CADe implementation across the Veterans Health Administration.
Why was this study needed?
- Randomized trials have shown improved ADR with CADe, but real-world evidence has been inconsistent.
- The effectiveness of CADe across different endoscopist experience levels remained unclear.
- Quality improvement data are needed before widespread implementation.
- Increasing ADR is a proven strategy for colorectal cancer prevention.
- AI-assisted colonoscopy is rapidly entering routine clinical practice.
Results:
- Implementation of CADe significantly increased adenoma detection rates, with a 22% higher likelihood of detecting at least one adenoma compared with standard colonoscopy.
- The benefit was consistent across all endoscopists, including those who already had high baseline adenoma detection rates.
- CADe did not increase unnecessary biopsies, prolong withdrawal time, or alter colorectal cancer detection, supporting its efficient integration into routine practice.
Clinical Impact:
This large real-world study confirms that AI-assisted colonoscopy improves adenoma detection outside clinical trials and benefits both average and high-performing endoscopists. These findings support broader implementation of CADe as a quality improvement tool to enhance colorectal cancer prevention.
Bottom Line:
Computer-aided detection (CADe) improves adenoma detection in everyday clinical practice regardless of baseline endoscopist performance. AI is becoming an effective quality-enhancement tool that complements, rather than replaces, the expertise of the endoscopist.