DIA Biostatistics Industry and Regulator Forum

The causal estimand of interest in many randomized trials is the intention-to-treat effect, that is, the effect of being assigned to the treatment strategies of interest. However, in many randomized trials, patients and doctors are more interested in another causal estimand: the per-protocol effect, that is, the effect of following the assigned treatment strategies as indicated in the protocol during the follow-up period. Valid estimation of the per-protocol effect generally requires adjustment for pre- and post-randomization prognostic factors associated with adherence and loss to follow-up. This talk describes the relative advantages and disadvantages of intention-to-treat and per-protocol effects, and reviews several case studies in cardiovascular disease, infectious disease, and cancer.