Direct effect on validity of response run-in selection in clinical trials

A run-in is a period prior to randomization during which potential participants who have met all entry criteria for a randomized clinical trial are assigned the same regimen, either the control (possibly placebo) or the experimental treatment. Typically, the intention is to exclude from the subsequent study (i.e., randomization) some segment of this cohort, based on their experiences during the run-in period. Selecting patients based on the run-in thereby forces differential representation of certain subpopulations relative to others. Previous studies have addressed the potential for a run-in to jeopardize validity through unintended mechanisms. While these concerns are valid, they leave open the possibility that modifications to the design of the run-in might be able to preserve validity, even if patient selection is based on the results of the run-in. As such, we address the potential for selecting patients based on response during a run-in period to jeopardize validity directly, through its intended effect of overrepresenting (relative to the screened population) some segments in the randomized portion of the trial and underrepresenting others.