In the era of widespread endoscopy use, randomized trials may strongly underestimate the effects of colorectal cancer screening

Objectives: Although randomized controlled trials (RCTs) are considered the most valid approach to evaluate screening effects, they also face a number of methodological challenges, including nonadherence and contamination. We aimed to quantify their potential impact in RCTs evaluating endoscopic screening for colorectal cancer (CRC).

Study design and setting: We carried out model calculations using plausible levels of nonadherence and contamination.

Results: Assuming medium values of adherence (70%) and contamination, that is, use of lower gastrointestinal endoscopy other than the one offered for screening (30%), true reductions in CRC incidence or mortality of 70%, 50%, or 30% would be expected to be attenuated to reductions of 43%, 29%, and 16%, respectively, in intention-to-screen analyses. In case of low levels of adherence (50%) and high but realistic levels of contamination (50%), even more severe attenuation of screening effects to estimated reductions of 27%, 17%, and 9% would be expected. The estimates are only slightly modified in sensitivity analyses, additionally allowing for differential adherence and contamination according to CRC risk.

Conclusion: In the era of widespread endoscopy use even outside the screening programs, RCTs may strongly underestimate the effects of CRC screening. Additional analyses accounting for nonadherence and contamination are crucial for disclosing the true screening effects.