Pragmatic Implementation Study of a Computer-Aided Detection Device on Colonoscopy Outcomes at a Trainee-Dominated Academic County Hospital.

Purpose: Multiple randomized controlled trials have shown that real-time computer-aided detection (CADe) devices can significantly increase adenoma and sessile serrated polyp detection. Unfortunately, there have been several pragmatic implementation studies that have questioned the utility of these devices in everyday practice. This study aimed to understand the effects of the pragmatic implementation of a CADe device on colonoscopy metrics at a trainee-dominated academic county hospital.

Methods: We collected data on all colonoscopies performed 1 year before (1665 colonoscopies) and after implementation of the CADe device (2301 colonoscopies). CADe use was not mandated. We excluded colonoscopies which were incomplete and had inadequate bowel preparation (Boston Bowel Prep Score (BBPS)) < 6 or < 2 in any colonic segment.

Results: The baseline adenoma detection rate (ADR) and sessile serrated lesion detection rate (SSLDR) were 31.5% and 1.3%, respectively, and increased by + 7.6% (95% CI 4.6-10.6) and + 2.9% (95% CI 1.9-3.8) after the implementation of CADe (p value < 0.001). After adjusting for other covariates (age, gender, BBPS, withdrawal time, positive fecal immunochemical test indication), the odds of adenoma detection (aOR 1.29, 95% CI 1.12-1.51, p = 0.001) and sessile serrated lesion detection (aOR 3.10, 95% CI 1.93-4.99, p < 0.001) were higher after CADe implementation.

Conclusion: The implementation of a CADe device irrespective of CADe device usage significantly improved ADR and SSLDR in a trainee-dominated academic county hospital setting. The benefits of CADe devices may be more apparent, and implementation of CAD devices should be considered in practices where the ADR and/or SSLDR is not already exceedingly high.