How to use and interpret hormone ratios

Abstract

Hormone ratios have become increasingly popular throughout the neuroendocrine literature since they offer a straightforward way to simultaneously analyze the effects of two interdependent hormones. However, the analysis of ratios is associated with statistical and interpretational concerns which have not been sufficiently considered in the context of endocrine research. The aim of this article, therefore, is to demonstrate and discuss these issues, and to suggest suitable ways to address them. In a first step, we use exemplary testosterone and cortisol data to illustrate that one major concern of ratios lies in their distribution and inherent asymmetry. As a consequence, results of parametric statistical analyses are affected by the ultimately arbitrary decision of which way around the ratio is computed (i.e., A/B or B/A). We suggest the use of non-parametric methods as well as the log-transformation of hormone ratios as appropriate methods to deal with these statistical problems. However, in a second step, we also discuss the complicated interpretation of ratios, and propose moderation analysis as an alternative and oftentimes more insightful approach to ratio analysis. In conclusion, we suggest that researchers carefully consider which statistical approach is best suited to investigate reciprocal hormone effects. With regard to the hormone ratio method, further research is needed to specify what exactly this index reflects on the biological level and in which cases it is a meaningful variable to analyze.