In their recently published study, Araujo et al1 come to the conclusion that in men, endogenous sex steroid levels appear to have relatively weak albeit significant associations with mortality. Nevertheless, even this weak connection should be regarded with caution because several confounding factors, which may further limit the significance of the finding, were not included in their analysis.
First, only sex steroid measurements were given, although it is more appropriate to measure testosterone and interpret findings in conjunction with measurements of plasma-luteinizing hormone.2 Moreover, although the circadian variation of testosterone is markedly dampened with advancing age,2 no adjustment was made for previous nighttime sleep quantity, which seems to be associated with a substantial part of the variability in the morning testosterone levels of older men.3 Second, the presence or absence of depressive symptoms, as a covariate, apparently was not included in the analysis by Araujo et al,1 although relevant data are available from the Massachusetts Male Aging Study.4 In their sample of men aged 40 to 70 years, a considerable prevalence of depression might be expected, which is a condition very often accompanied by complaints of poor health and probably linked with testosterone deficiency5 and increased mortality rates. Third, several medical conditions, besides cardiovascular disease, diabetes mellitus, and dyslipidemia, as well as medications known to affect testosterone plasma levels2 were not presented in their analysis. Last but not least, the severity of alcohol consumption, which is usually associated with low testosterone level and changes in the levels of other reproductive hormones,6 was not adequately gauged.