Disclosing the true impact of screening endoscopy on colorectal cancer incidence

Objectives Randomized trials have demonstrated reduction of colorectal cancer (CRC) incidence by screening endoscopy. However, measured reduction underestimates true reduction due to inclusion of preclinical cases already present at recruitment. We aimed to quantify the true impact of screening endoscopy on reducing the CRC incidence. Design Simulation study replicating reported CRC incidence by SCORE, a large, randomized screening sigmoidoscopy trial, and deriving expected incidence after excluding cases that manifested during follow-up but were already prevalent at baseline. Setting Offer of a single flexible sigmoidoscopy in an organised, population-based screening setting. Participants Simulated, sex- and age-matched SCORE trial population (intervention group, N=17,136, control group, N=17,136, 50% women, ages 55-64 at baseline). Interventions Screening flexible sigmoidoscopy versus no screening. Main outcome measure ‘True’ (i.e., unbiased, excluding prevalent cancers at baseline) and ‘apparent’ (i.e., as reported) incidence rate ratios (IRR) for screening versus no screening. Results In the initial years after randomization, apparent cumulative incidence in the screening group was higher than in the control group due to inclusion of a large proportion of prevalent cancers. In the longer run, apparent cumulative incidence was lower in the screening group, but this incidence reduction was still much lower than true incidence reduction due to inclusion of prevalent cases in calculation of apparent cumulative incidence. In intention-to-screen analysis, apparent/true risk reductions after 8, 11 and 15 years of follow-up were 16%/31%, 20%/28%, and 21/25%, respectively. In per-protocol analyses, respective apparent/true risk reductions were 28%/54%, 34%/49%, and 35%/44%. Estimated underestimation of true incidence was similar among men and women and among age groups 55-59 and 60-64. Conclusions The preventive effect of screening endoscopy is likely much stronger than reflected in the reported apparent IRRs. Published findings of randomized screening trials underestimate the true preventive effective of screening endoscopy even after 15 year or longer follow-up. Summary Box What is already known on this topic - Several large-scale randomized trials have demonstrated substantial reduction of colorectal cancer incidence by endoscopic screening. - In these trials, the preventive effect of screening endoscopy only transpires after 4-6 years, as screen-detected, prevalent cancers (which can no longer be prevented) dominate the measured incidence in the first years of follow-up. - The true impact of screening endoscopy on CRC incidence is therefore essentially unknown. What this study adds This modelling study derives estimates of the apparent and true impact of screening sigmoidoscopy on reducing the CRC incidence by accounting for prevalent preclinical cancers at baseline. After careful calibration, the model closely predicts observed effects on CRC incidence in the SCORE trial, a randomized trial of flexible sigmoidoscopy conducted in Italy, and demonstrates that the endoscopy screening effect on incidence might be substantially larger when accounting for prevalent preclinical cancers at baseline.