Primary Submission Category: Randomized Designs and Analyses
Causal Effects in Blinded Trials
Authors: Jipcy Amador, Michael Hudgens, Todd Schwartz,
Presenting Author: Jipcy Amador*
In randomized clinical trials (RCTs), blinding (or masking) is commonly employed to reduce bias that may be introduced due to knowledge or presumptions that participants or investigators may have about the treatment being evaluated. However, in many settings it may also be of interest to evaluate the treatment effect in the absence of blinding to better predict the effect of treatment when used in the real world. In some settings understanding the effect of blinding may also be of intrinsic interest. This paper considers identification of treatment effects (in the presence or absence of blinding) and blinding effects (in the presence or absence of treatment) in blinded RCTs, under different assumptions about the treatment and blinding effects. In instances where the effect of interest is partially identified, sharp bounds are provided. Both single-blind and double-blind RCTs are considered. The methods are applied to: (i) a single-blind RCT evaluating the effect of silver diamine fluoride on caries arrest and prevention in children, and (ii) a double-blind RCT evaluating the effect of rituximab in individuals with myasthenia gravis.