ABSTRACT
Despite these obvious benets, the extrapolation of the results of clinical trials to particular subgroups is surprisingly unreliable and contentious. Even if the eect sizes of the treatment dierences appear to be substantial and potentially important, standard approaches to subgroup analyses are so statistically insensitive that they may be unable to demonstrate them.3 Conversely, a highly signicant eect, using appropriate statistical methods, may still be inadequate for providing good evidence about subgroup-specic treatment preferences.3
