A few things I didn’t know about handling non-adherence in randomized trials and a few other things I still don’t know
In randomized clinical trials, subjects often do not comply with their randomized treatment arm. Although one can still unbiasedly estimate the causal effect of assignment to treatment using the common Intention-to-Treat (ITT) estimator, there is now potential confounding of the causal effect of actually *receiving* treatment. Basic alternative estimators such as the per protocol or as treated estimators have been used, but are generally biased for estimating the causal effect of interest. Balke and Pearl (1997) and Angrist, et al. (1996) independently proposed an instrumental variable (IV) estimator that would estimate the causal effect (the Complier Average Causal Effect) of receiving treatment in a subpopulation of people who would comply with treatment assignment (i.e. the compliers). Other authors (Robins in several papers, Frangakis and Rubin and many others) have proposed more complex methods for dealing with the problem that allow one to incorporate more information, but these methods lose the simplicity associated with the IV estimator and see relatively little usage in the general statistical community.
In this talk, I will dissect the instrumental variable estimator in order to compare it to the per protocol and as treated estimators. I will also provide some suggested influence plots that I’ve found useful to help understand the influence of certain parameters on inference. Finally, I will talk about how thinking about the problem in this way introduces some new questions and directions for future research.