Salivary cortisol and depression in public sector employees: Cross-sectional and short term follow-up findings

Abstract

Increased cortisol levels have been suggested to play a role in the development of depression. An association has been shown in some studies but not consistently. The timing of an association is uncertain, and long-term follow-up studies may miss associations in narrower time windows. In the present study, we examined the association of several cortisol measures and depression in a repeated cross-sectional and short-term follow-up design. Depression was assessed by both self-reported symptoms of depression and clinical interviews. In 2007, 10,036 public sector employees received a questionnaire along with salivary cortisol test tubes for home administration. Morning (30min after awakening) and evening (2000h) salivary samples were collected. Questionnaires and valid saliva samples were returned from 3536 employees. Approximately 3.6 months later a subsample of the participants collected three morning saliva samples (at awakening, 20min and 40min after awakening) plus an evening sample (2000h); participants with high baseline scores of self-reported depressive symptoms, burnout and perceived stress were invited to a standardized interview (SCAN) to detect clinical depression; and the symptom questionnaire was repeated for subsample participants. The study was repeated in 2009 with questionnaires and salivary test tubes (n=2408). In four cross-sectional and two short-term follow-up analyses odds ratios of depressive symptoms and of clinical depression were estimated by logistic regression for morning, evening, mean and the difference between morning and evening cortisol (slope). For the subsample, awakening response (CAR) and area under the curve (AUC) cortisol measures were calculated. We adjusted for sex, age, income, education, family history of depression, physical activity and alcohol consumption. None except one of the measures of salivary cortisol were associated with self-reported depressive symptoms or clinical depression, neither in the four cross-sectional analyses nor in the two short term follow-up analyses. E.g. in 2007, the adjusted odds ratios (OR) of depressive symptoms by a one unit increase in morning and evening cortisol (ln(nmol/litre saliva)) were 1.01 (95% CI: 0.88-1.17) and 1.05 (0.93-1.18), respectively. The one exception was significant at p=0.04 and was considered as due to chance. In this large study, salivary cortisol was not associated with self-reported symptoms of depression or with clinical depression.