Computer-aided detection (CADe) has been shown to improve outcomes in randomized trials. However, most studies occurred in controlled environments where endoscopists are aware of being monitored, potentially influencing performance and limiting real-world applicability. Our aim was to assess the impact of CADe in routine multicentre practice, with endoscopists unaware that their performance was being evaluated and serving as their own controls.

Prospective multicentre study across 13 units (March–September 2025) including adults aged ≥40 years undergoing colonoscopy. All six endoscopists executed procedures both in a centre where CADe use is compulsory and in others without CADe, unaware that their CADe and nonCADe performance was being compared. The primary outcome was adenoma detection rate (ADR). Secondary outcomes included combined adenoma plus clinically significant serrated polyp detection (A+CSSPDR), polyp detection (PDR), proximal neoplasia detection (PNDR), diminutive polyp detection (DPDR) and serrated detection rate (SDR).  

Of 3161 eligible procedures, 2973 were analysed (947 CADe; 2026 non-CADe). ADR was higher with CADe (34.4% vs 30.7%; p<0.05). A+CSSPDR (40.3% vs 35.8%; p<0.05), PDR, PNDR and DPDR were also significantly higher, while SDR showed a non-significant positive trend. CADe remained independently associated with higher ADR (aOR 1.31, 95%CI 1.11–1.55; p=0.002) and A+CSSPDR (aOR 1.33, 95%CI 1.13–1.56; p=0.001) after adjusting for age, sex and indication; the magnitude of effect was attenuated once endoscopist-level adjustment was applied.

In real-world practice, with fully blinded conditions, CADe improves detection of adenomas and clinically relevant serrated lesions, though the effect varies between endoscopists. This underscores its utility while reflecting operator-driven variability.