Comparative Efficacy Of Different Computer-Aided Detection Products in Colonoscopy: A Network Meta-Analysis Of Randomized Controlled Trials.

Abstract

Background and aims: Multiple computer-aided detection (CADe) systems have been developed to improve adenoma detection during colonoscopy, but their comparative efficacy and consistency of effect remain uncertain. This network meta-analysis aimed to compare the efficacy of different CADe systems across multiple detection outcomes. Methods: MEDLINE, Embase, Scopus, and the Cochrane Library were systematically searched for randomized controlled trials (RCTs) comparing CADe-assisted colonoscopy versus colonoscopy without CADe or other CADe systems. The primary outcome was adenoma detection rate (ADR). Secondary outcomes were adenomas per colonoscopy (APC) and adenoma miss rate (AMR). A random-effects network meta-analysis was performed to synthesize direct and indirect evidence. Results: Forty-one RCTs (31,789 patients; 14 CADe systems) were included. In the exploratory complete-network analysis, 10 of 14 systems significantly improved ADR (RR range 1.12–1.76), 9 increased APC, and 3 reduced AMR. In the FDA-cleared systems analysis, 3 of 6 improved ADR (RR range 1.14–1.25), whereas 3 did not. Several systems retained a significant benefit in studies meeting the control-arm ADR thresholds recommended by ESGE (≥25%) and ASGE/ACG (≥35%), although effects varied across platforms and attenuation at higher baseline ADR was not uniform. Baseline ADR was not associated with treatment effects in meta-regression. Conclusions: The effectiveness of CADe-assisted colonoscopy is not uniform across platforms and varies according to the specific product. These findings support a system-specific rather than class-based evaluation of CADe technologies. However, comparative estimates should be interpreted cautiously given the low certainty of evidence for several outcomes and the absence of head-to-head comparisons and closed loops within the network.