Phil B.FontanarosaMD, Deputy EditorIndividualAuthorStephen J.LurieMD, PhD, Fishbein FellowIndividualAuthor
To the Editor: It is unfortunate that the randomly
assigned placebo and androstenedione groups in the study by Dr King and colleagues1 had a difference in free testosterone levels of
almost 50%. Why were the 20 subjects not randomly regrouped in an attempt
to equilibrate this difference, especially considering that free testosterone
levels are perhaps the most important measure of androstenedione efficacy?
Granted, this approach would taint the randomized design of the study, but
with only 20 individuals, chance will ensure that groups are created that
do not permit valid conclusions. The incongruence between the 2 study groups
is all the more important considering that free testosterone levels did appear
to increase in the androstenedione group. The authors claim that "a sample
size of 160 would have been required to detect an effect of this size."1 Perhaps the sample size of 20 would have sufficed
had the groups been comparable.
O'Gara T. Effects of Androstenedione in Young Men. JAMA. 2000;283(6):741-743. doi:10.1001/jama.283.6.741