A commentary on “Testosterone and cortisol jointly modulate risk-taking” by P.H. Mehta, K.M. Welker, S. Zilioli, J.M. Carre, Psychoneuroendocrinology, 2015, 56, 88–99

Dear Editor, We read with interest the work of Mehta et al. (2015) who investigated the influence of salivary testosterone (sal-T) and salivary cortisol (sal-C) on risk-taking behaviors in a mixed gender and ethnicity sample. Authors report a positive relationship between basal sal-T and risk-taking, which was measured using the Balloon Analog Risk Task of Lejuezet al. (2002). While we found the topic of interest to the psychoendocrinology community, we have concerns with the methodology employed to derive a relationship between sal-T and risk-taking behavior. Firstly, the authors state saliva samples were collected between 10.30 and 17.30 h, which is a sizeable time-frame when considering the known circadianoscillation of steroid hormones in saliva (Riad-Fahmy et al.,1983). Moreover, an anticipatory effect of sal-T and sal-C is known prior to a task (Hayes et al., 2015) and therefore, it is possible that samples analyzed were measures of the anticipatory rise (rather than basal concentrations) of sal-C and sal-T. Furthermore, the mixed gender and ethnicity of the sample may have further confounded potential error ratesas a result of increased heterogeneity in androgen status (Litman et al., 2006). Moreover, we recently demonstrated high variability in sal-T and sal-C in a highly controlled laboratory environment (Hayes et al., 2014), which was epitomized by the large standard deviations (SD) observed in thisstudy (the SD of sal-C was greater than the mean). Furthermore, when sal-C was high (+1 SD), the association betweensal-T and risk taking was non-significant. These issues, considered alongside the innate variability and pulsatile nature of steroid hormones, suggest the interaction between risk-taking behavior and sal-T may be artifactual and the authorsmay have fallen victim to classic type I error. Overall, we find the authors’ conclusion that there were ‘‘consistent positive slopes between testosterone andrisk-taking only among low-cortisol individuals’’ to be unsub-stantiated as the association was not statistically significant(p = 0.051) and only self-reported and informant-reported risk-taking was statistically significant (p < 0.05) suggesting that perceived rather than observed risk-taking was associated with sal-T. Finally, we find the absence of ‘‘salivary’’ in title, abstract, and conclusion is misleading as salivary hormones do not consistently demonstrate acceptable agreement with serum values (Granger et al., 2004). With respect to the aforementioned issues with salivary hormone measurement cited herein, the conclusions of Mehta et al.(2015) which infers a cause-and-effect relationship betweensal-T and risk-taking are speculative.