Levels and confounders of morning cortisol collected from adolescents in a naturalistic (school) setting

Salivary cortisol is widely used in research but little is known about the typical, or expected, functioning of the HPA-axis in adolescents in naturalistic settings, nor whether the extensive array of confounders documented in the literature is applicable in this situation. In a school-based study, 2995 15-year-old pupils provided two saliva samples, 30 min apart, in morning sessions timed to capture peak cortisol decline. The collection protocol was a balance between the large sample size obtainable in a school situation and a limited number of samples, constrained by the school timetable. In addition, pupils completed a questionnaire containing items previously shown to be associated with cortisol levels (e.g. time since awakening and life events), and their height and weight were measured. Outcome measures were cortisol levels at Times 1 and 2, and change (per minute) in cortisol between the two time points. Median (IQR) cortisol levels for males and females were 10.5 (8.1) and 11.6 (9.3) nmol/L at Time 1, and 8.2 (6.0) and 8.1 (6.5) nmol/L at Time 2. 73% had a decline in cortisol level of more than 10% across the two time points, compatible with the expected diurnal pattern. In bivariate analyses, cortisol sampled on Monday, times of measurement and since awakening, prior smoking and several life events were associated with cortisol levels at Times 1 and 2 in both sexes. However, in multivariate analysis, few of these variables remained after controlling for times of measurement and since awakening and, in addition, the final models differed between the sexes. Two events (friend dying and splitting with a boy/girlfriend) predicted cortisol levels in both sexes while age, maturity, recent eating and smoking were predictors only in males. Several factors associated with cortisol change differed from those observed for absolute levels. Further adjustment for school clustering affected some associations, particularly time of measurement. This study managed many of the problems found in naturalistic research on cortisol and provides norms for morning cortisol levels in 15-year-old adolescents