Did you know?
You can earn activity points by filling your profile with information about yourself (what city you live in, your favorite team, blogs etc.)
A problem I gave in assessing the reliability of the prespecified subgroup data analyses (all of them) is that Don has never to my knowledge shown how well (or how poorly) the placebo and treated patients within the subgroups were matched for relevant variables. Given that they knew in advance what subgroup analyses they were going to do, ideally they should have used a stratified randomisation procedure to ensure that the subgroups (not just the total groups) would be well matched for the things that matter. As it is, they left that to chance, and we don’t know how it worked it out.