Bounding the per-protocol effect in randomized trials: an application to colorectal cancer screening.

Bounding the per-protocol effect in randomized trials: an application to colorectal cancer screening.