RESUMEN
Assessing the amount of bioavailable cortisol in saliva with immunoassays and thus sampling an endocrine marker of hypothalamus-pituitary-adrenal axis activity is of major interest in both research and clinical practice. However, absolute cortisol concentrations obtained with different immunoassays (IAs) are barely comparable precluding direct comparison between studies or individuals whenever cortisol analyses were not based on the same IA. The present technical report aims to solve this problem by evaluating the validity of, as well as agreement between the most commonly used immunoassays in psychoneuroendocrinological research (i.e., IBL, DRG, Salimetrics, DSL, and DELFIA) and a reference method (LC-MS/MS) in a sample of 195 saliva specimen covering the whole range of cortisol concentrations in adults. A structural equation modelling framework is applied to decompose systematic assay variance and estimate cortisol reference values, which are adjusted for measurement error and interference of salivary cortisone. Our findings reveal nonlinear relations between IAs and LC-MS/MS, which are discussed in terms of IA cross-reactivity with saliva matrix components. Finally guidelines for converting cortisol concentrations being obtained by these immunoassays into comparable reference values are proposed by providing conversion functions, a conversion table, and an online conversion tool.