Cortisol and Depressive Symptoms in a Population-Based Cohort of Midlife Women

Departments of Internal Medicine and Psychiatry, Rush University Medical Center, Chicago, Illinois, USA.
Psychosomatic Medicine (Impact Factor: 3.47). 11/2010; 72(9):855-61. DOI: 10.1097/PSY.0b013e3181f4ab87
Source: PubMed


To determine whether there is a relationship between depressive symptoms and cortisol assessed at first morning awakening, 6 PM, and 9 PM in a population-based sample of midlife women. If this relationship is not linear, we aim to test whether this relationship is nonlinear, only present in those with more severe depressive symptoms, better accounted for by diurnal slope, or only apparent under uncontaminated conditions.
We investigated the cross-sectional association between cortisol and depressive symptoms, assessed by the Center for Epidemiological Studies Depression scale (CES-D) in 408 midlife women (45.7% African Americans, 54.3% white; mean age, 50.4 years) participating in the Chicago site of the Study of Women's Health Across the Nation.
Diurnal cortisol slope is significantly flatter for women with higher CES-D scores than for less depressed women (p < .05 for the interaction). This relationship remains significant even after adjusting for age, smoking status, race, education, income, menopausal status, hormone replacement therapy, body mass index, medications, and wake time, as well as possibly contaminating factors, including physical activity, smoking, eating, or caffeine or alcohol consumption before saliva collection. Results using depression assessed categorically (CES-D cutoff ≥16) were similar to those using continuous depression in both unadjusted and adjusted analyses (p = .005 for the interaction of CES-D by time).
In this population-based sample of midlife women, greater depressive symptoms were associated with a significantly flatter diurnal cortisol slope than those with fewer symptoms, even after adjusting for covariates and possibly contaminating behaviors.

Download full-text


Available from: Elizabeth F Avery,
14 Reads
  • Source
    • "Scores of 27 or more indicate a major depression (Zich et al. 1990). The CES-D is a frequently used depression scale with high validity and stability in clinical and community samples (Knight et al. 2010; Morin et al. 2011). The CES-D scores in both samples were significantly correlated with other scales measuring mood state at 6 months postpartum, e.g. the Edinburgh Postnatal Depression Scale (EPDS) (Cox et al. 1987) [r = "
    [Show abstract] [Hide abstract]
    ABSTRACT: Mothers vary in duration of breastfeeding. These individual differences are related to a variety of demographic and individual maternal factors including maternal hormones, mood and early experiences. However, little is known about the role of genetic factors. We studied single nucleotide polymorphisms (SNPs) in the OXT peptide gene (rs2740210; rs4813627) and the OXT receptor gene (OXTR rs237885) in two samples of mothers from the Maternal adversity, Vulnerability and Neurodevelopment study (MAVAN), a multicenter (Hamilton and Montreal, Canada) study following mothers and their children from pregnancy until 7 years of age. Data from the Hamilton site was the primary sample (n = 201) and data from Montreal the replication sample (n = 151). Breastfeeding duration, maternal mood (measured by the CES-D scale) and early life adversity (measured by the CTQ scale) were established during 12 months postpartum. In our primary sample, polymorphisms in OXT rs2740210, but not the other SNPs, interacted with early life adversity to predict variation in breastfeeding duration (overall F (8,125) = 2.361, p = .021; interaction effect b = -8.12, t = -2.3, p = .023) and depression (overall F(8, 118) = 5.751, p ≤ .001; interaction effect b = 6.06, t = 3.13, p = .002). A moderated mediation model showed that higher levels of depression mediated the inverse relation of high levels of early life adversity to breastfeeding duration, but only in women possessing the CC genotype (effect a' = -3.3401, 95%CI=-7.9466 to -.0015) of the OXT SNP and not in women with the AA/AC genotype (a' = -1.2942, ns). The latter findings (moderated mediation model) were replicated in our Montreal sample (a' = -.277, 95%CI=-.7987 to -.0348 for CC; a' = -.1820, ns, for AA/AC).
    Genes Brain and Behavior 08/2013; 12(7). DOI:10.1111/gbb.12069 · 3.66 Impact Factor
  • Source
    • "Future work should carry out a more fine-grained analysis investigating how different aspects of physical health are associated with deviation from the normative diurnal cortisol curve. Previous research shows strong links between stress-eliciting environments and psychological well-being (e.g., Slavich et al., 2009), and between flatter diurnal cortisol slope and symptoms (Knight et al., 2010) and diagnosis (Jarcho et al., 2013) of depression. Thus, a careful examination of potential links between psychological wellbeing and diurnal cortisol profiles is another important direction for future studies to pursue. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Diurnal cortisol is a marker of HPA-axis activity that may be one of the biological mechanisms linking stressors to age-related health declines. The current study identified day-centered profiles of diurnal cortisol among 1101 adults living in the United States. Participants took part in up to four consecutive days of salivary cortisol collection, assessed at waking, 30min post-waking, before lunch, and before bedtime. Growth mixture modeling with latent time basis was used to estimate common within-day trajectories of diurnal cortisol among 2894 cortisol days. The 3-class solution provided the best model fit, showing that the majority of study days (73%) were characterized by a Normative cortisol pattern, with a robust cortisol awakening response (CAR), a steep negative diurnal slope, coupled with low awakening and bedtime levels. Relative to this profile, diurnal cortisol on the remainder of days appeared either elevated throughout the day (20% of days) or flattened (7% of days). Relative to the normative trajectory, the elevated trajectory was distinguished by a higher morning cortisol level, whereas the flattened trajectory was characterized by a high bedtime level, with weaker CAR and diurnal slope parameters. Relative to the normative profile, elevated profile membership was associated with older age and cigarette smoking. Greater likelihood of the flattened cortisol pattern was observed among participants who were older, male, smoked cigarettes, used medications that are known to affect cortisol output, and reported poorer health. The current study demonstrates the value of a day-centered growth mixture modeling approach to the study of diurnal cortisol, showing that deviations from the classic robust rhythm of diurnal cortisol are associated with older age, male sex, use of medications previously shown to affect cortisol levels, poorer health behaviors, and poorer self-reported health.
    Psychoneuroendocrinology 06/2013; 38(10). DOI:10.1016/j.psyneuen.2013.05.003 · 4.94 Impact Factor
  • Source
    • "This cortisol secretory pattern is referred to as the diurnal cortisol decline (DCD), and the slope of the decline between the morning and evening cortisol values is used as a measure of the DCD. Many studies have found altered, i.e., flattened or heightened, diurnal cortisol slopes in patients suffering from pulmonary tuberculosis (Baker et al., 2000), sepsis (Bronstein et al., 1998), rheumatoid arthritis (Dekkers et al., 2000), and psychosomatic or psychiatric disorders, such as fatigue (Jerjes et al., 2005) or depression (Knight et al., 2010), as well as work and situational stress (Oginska et al., 2010; Ulhóa et al., 2011; Wirth et al., 2011). The DCD profile in patients with metastatic cancer was found to be variable according to the type of cancer and additional factors. "
    [Show abstract] [Hide abstract]
    ABSTRACT: A flatter diurnal rhythm of cortisol has been reported to be associated with early mortality in patients with metastatic breast cancer. The clinical stage of disease at the time of diagnosis and the patient's performance status (PS) are known to be important prognostic factors for lung cancer (LC) survival. The authors examined the relationship between diurnal cortisol rhythms and these prognostic factors in patients with advanced LC. Cortisol concentrations were measured in saliva samples collected from 52 patients (37 males/15 females) with advanced LC and from 56 healthy subjects (32 males/24 females) to characterize the diurnal cortisol rhythm, specifically the cortisol awakening response (CAR) and diurnal cortisol decline (DCD). Variations of CAR and DCD in the patients were analyzed according to their clinical disease stage and PS score, and the differences in CAR and DCD between patients and healthy controls were compared. The patient group showed significantly reduced diurnal cortisol secretory activity and rhythmicity, compared with healthy controls. When the patients were subgrouped according to their clinical disease stage, patients with stage 4 disease showed significantly reduced CAR and flatter DCD compared with the healthy controls. However, the CAR and DCD in patients with stage 3a and 3b disease were comparable to those of healthy controls. Neither the CAR nor the DCD showed stepwise changes as the disease stage worsened. When patients were subgrouped according to their Eastern Cooperative Oncology Group (ECOG) PS score, there was stepwise reduction in the CAR and flattening of the DCD as the PS score increased. Both an abolished CAR and a flattened DCD were common in patients with ECOG PS scores of 3 and 4. These results indicate that alteration of the diurnal cortisol rhythm in patients with advanced LC is more closely associated with their PS score than with their clinical disease stage. Gradual alteration of the CAR and DCD, indicative of loss of 24-h cortisol rhythm, in concert with increase in PS score implies that endogenous circadian rhythms may also be disintegrating as the PS score worsens in these patients. (Author correspondence: ).
    Chronobiology International 08/2012; 29(8):1109-20. DOI:10.3109/07420528.2012.706767 · 3.34 Impact Factor
Show more