Abstract:
|
There is effect modification if the magnitude or stability of a treatment effect varies systematically with the level of an observed covariate. A larger or more stable treatment effect is typically less sensitive to bias from unmeasured covariates, so it is important to recognize effect modification when it is present. We first illustrate a recent proposal for conducting a sensitivity analysis that empirically discovers effect modification by exploratory methods, but controls the family-wise error rate in discovered groups. The proposal is illustrated in a matched pair study of the effect of a patient requiring surgery being treated at a hospital with superior vs. inferior nursing. Then, we develop an extension of the method to allow for multiple controls. We show that making use of multiple controls substantially increases the power of the procedure.
|