nominally significantP <.05
To illustrate the broader applicability of the method, we have applied it to the data from START randomized controlled trial.We showed that theCVRSmethod applied with an additional selection step, identified a sensitive group that conferred a nominally significant interaction effect ( P <.05) between the treatment and the sensitivity status, in the case where there was no overall significant treatment effect.