Abstract:
|
One common goal of subgroup analyses is to determine the subgroup of the population for which a given treatment is effective. Many approaches involve testing—implicitly or explicitly—hypotheses about many types of patients which are not exchangeable, and methods of controlling family-wise Type I error rate inflation are available. Such methods are designed to control the rate of erroneously declaring at least one type of patient as benefiting and are therefore quite strict. We present a method for instead controlling a weighted false discovery rate (FDR) in the sense of controlling the expected proportion of patient types declared benefiting, weighted by their population prevalence, which do not in fact benefit from treatment. Such population-weighted FDR is analogous to maintaining positive predictive value of a diagnostic test for expected benefit. We minimize power loss by using a resampling approach that accounts for correlation among test statistics at similar covariate points. We illustrate our method with an analysis of a clinical trial of an Alzheimer’s disease treatment.
|