Paparrigopoulos et al seem to misunderstand the conclusion of our article,1 which is that sex steroids have virtually no relationship with mortality. As there is no relationship to be confounded, their subsequent comments regarding uncontrolled or residual confounding lack foundation. Furthermore, the many carefully selected covariates included in our analysis (age, body mass index, waist circumference, high-density lipoprotein cholesterol level, systolic blood pressure, race, alcohol consumption, calories expended in physical activity, smoking, self-assessed health, and self-reported heart disease, hypertension, and diabetes mellitus) represent a very comprehensive set of variables that could have confounded an association between testosterone and mortality. Additional factors affecting testosterone levels undoubtedly exist, but those mentioned by Paparrigopoulos et al were either controlled for (ie, alcohol consumption), based on scant data (ie, the impact of sleep duration on testosterone levels in 12 subjects2), or found to be unassociated with testosterone levels in the Massachusetts Male Aging Study (ie, depressive symptoms3).
Araujo AB, Kupelian V, Page ST, Handelsman DJ, Bremner WJ, McKinlay JB. The Relationship Between Testosterone and Mortality in Men: A Debatable Issue—Reply. Arch Intern Med. 2008;168(3):330. doi:10.1001/archinternmed.2007.93