| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the University of California, San Francisco, San Francisco, CA (H.M.B.); the University of California, Berkeley, Berkeley, CA (L.C.F.); the University of California, Berkeley, Instituto Nacional de Salud Publica, Berkeley, CA (P.J.G.); and Mexico University of California, San Francisco, CA (N.E.A.).
Address correspondence and reprint requests to Heather M. Burke, PhD, Health Psychology Program, Department of Psychiatry, University of California, San Francisco, 3333 California St., Ste. 465, San Francisco, CA 94143. E-mail: hburke{at}itsa.ucsf.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Methods: Adult women (N = 1109) between the ages of 18 and 44 years (mean age, 29) were identified in a house-to-house survey in low-income areas (income <20th percentile nationally) of urban Mexico. An interview containing the Spanish version of the Center for Epidemiologic Studies—Depression Scale (CES-D) was administered to all women. The naturalistic stressor was defined as the unexpected arrival of a team of researchers at the participants’ homes followed by an in-depth interview and physical assessment, with saliva samples taken at time of arrival (baseline), 25 minutes, and 50 minutes after arrival.
Results: The mean CES-D score was 19.42 (range, 0–53). Results of hierarchical linear modeling analyses revealed no effect of depressive symptoms on baseline salivary cortisol levels. However, a significant depressive symptom by time interaction revealed that women with elevations in depressive symptoms (CES-D scores = 35) failed to exhibit a cortisol response to the stressor. In contrast, in women with lower CES-D scores, cortisol levels significantly increased in response to the stressor.
Conclusion: Consistent with research on individuals with major depressive disorder, results of this study demonstrate that women with very high levels of depressive symptoms exhibit blunted cortisol responses to a naturalistic psychological stressor. Results also contribute to previous research by generalizing findings to a high risk, underserved population of women.
Key Words: depression • salivary cortisol • stress • poverty • HPA axis
Abbreviations: HPA = hypothalamic-pituitary-adrenal; MDD = major depressive disorder; SES = socioeconomic status; CES-D = Center for Epidemiological Studies—Depression Scale; PSS = Perceived Stress Scale.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
The HPA axis is a principal transducer of psychosocial stress into physiological action (11). There is substantial evidence linking depression to alterations in HPA activity (12–21). Most of what is known about HPA functioning in depressed individuals is based on studies of basal HPA activity (16,22) and pharmacological challenge studies, such as the DST, designed to test the negative feedback mechanisms in the HPA axis (12). Some findings have suggested that a subset of individuals with major depressive disorder (MDD) have elevated basal cortisol and CRH levels coupled with cortisol elevations in response to dexamethasone administration (15,17,21).
Although provocative, the results of pharmacological challenge studies have some important limitations. First, the levels of dexamethasone, a synthetic glucocorticoid, used in pharmacological challenge studies are specifically designed to mimic the highest extreme of glucocorticoid functioning in order to suppress endogenous cortisol release (12). Such high levels of glucocorticoids may not accurately reflect the magnitude of endogenous HPA responses to psychosocial stressors. Furthermore, pharmacological and neuroendocrine challenge tests typically do not take into account suprahypothalamic factors in limbic-HPA functioning. Given that suprahypothalamic circuits exert important influences on HPA functioning, their omission in pharmacological and neuroendocrine challenge studies is a limitation of the existing literature. Thus, psychological stress challenges offer the advantage of reflecting endogenous activity of the entire HPA system.
To address the issue of generalizability, researchers have compared changes in cortisol in response to psychologically stressful stimuli in depressed with nondepressed individuals. Results of a recent meta-analysis revealed that clinically depressed individuals exhibit blunted cortisol stress reactivity and impaired cortisol recovery after laboratory stressors compared with their nondepressed counterparts (23). In addition, depression severity (as measured by elevated Hamilton scores) enhanced the blunted reactivity and impaired recovery effects. One study that compared cortisol responses to naturalistic psychosocial stressors between depressed and nondepressed patients also revealed blunted cortisol stress responses in the depressed sample (24). Thus, although results of pharmacological and neuroendocrine challenge studies have revealed evidence of HPA hyperactivity in depressed individuals, results of a much smaller number of psychological stress studies have revealed evidence of blunted cortisol stress responses.
One factor that might moderate the depression and HPA axis activity association is severity of depression. Clinical depressive disorders, including MDD, and depressive symptoms are considered by some to lie on a continuum, distinguished from one another by degree of symptom severity (25). There exists some evidence that depression-associated HPA abnormalities, such as DST nonsuppression, may vary as a function of symptom severity (26). However, few studies have examined the effect of depressive symptom severity on HPA responsiveness to psychosocial stressors. Studies that did examine responses to psychosocial stressors have focused exclusively on clinically depressed patient samples, with little variability in depressive symptoms between studies. Thus, despite the greater prevalence of depressive symptoms than clinical depression, little is known about the association between depressive symptoms and cortisol stress responses.
Clinical depression and depressive symptoms are not exclusive to Western culture; worldwide, they are a significant source of disability and distress (1). However, the majority of studies investigating the effect of depression on cortisol responses have focused on clinically depressed Western samples. To enhance the generalizability of previous research on depression and cortisol stress responses, we recruited a sample of very poor Mexican women at high risk for physical and mental illness. As female gender (27,28) and low socioeconomic status (SES) (29–32) are independent risk factors not only for the development of a major depressive disorder but also the negative health consequences of such a disorder, this represents a particularly vulnerable sample of women. Thus, the purpose of this study was to characterize the association of depressive symptoms with the pattern of cortisol responses to a naturalistic psychosocial stressor. Consistent with a symptom severity model of cortisol activity, we hypothesized that in this very low-income population, elevated depressive symptoms, as measured by the Center for Epidemiological Studies—Depression Scale (CES-D), would be related to higher baseline cortisol levels and blunted cortisol responses to a naturalistic stressor. Furthermore, this study also tests the generalizability of findings to a more disadvantaged and vulnerable group.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Materials
Depressive Symptoms
Depressive symptoms were assessed using the Spanish version of the CES-D, a 20-item questionnaire designed to assess intensity of depressive symptoms in the general population (34,35). It is important to note that the CES-D is not a clinical instrument to diagnose MDD but is instead a screen for the probability that a disorder exists given certain scores. Items are scored to reflect the frequency of symptom occurrence during the past week on a scale of 0 to 3, with 0 reflecting "not at all or less than 1 day" and 3 reflecting "5 to 7 days." Symptoms assessed by this scale include depressed mood (dysphoria), loss of interest and/or pleasure in activities (anhedonia), fatigue, concentration difficulties, excessive guilt and/or worthlessness, vegetative symptoms such as sleep and appetite disturbances, and social difficulties. Scores range from 0 to 60, with a generally accepted cutoff score of 16 in the United States for high risk of clinical depression (35). The CES-D has been validated for use in Spanish-speaking populations (36,37). However, a higher cutoff score has been proposed for adult Mexican women in rural areas, given the higher prevalence of depressive symptomatology. For instance, scores higher than 35 among rural women in Mexico have been considered as indicators of the presence of mood disorder (37). In the current study, Cronbach’s 
for the total CES-D score was 0.86.
Stress
In this study, the naturalistic stressor was defined as the unexpected arrival of a team of researchers at the participants’ homes followed by an assessment of physical, demographic, and psychosocial factors. Researchers arrived unannounced and asked the participants detailed and personal questions about demographic characteristics, psychosocial functioning, and health. Although we do not have data to directly assess whether the visit was stressful to the participants, an unexpected home visit is a face valid stressor, encompassing recognized stressful components such as uncontrollability, unpredictability, and social evaluation (38). In addition, as we will describe later, on average, cortisol levels increased when one would instead expect a decline because of diurnal rhythm. Taken together, these findings suggest that the visits were stressful to the participants. General perceived stress over the past week was assessed using a 10-item Spanish version of Cohen’s Perceived Stress Scale (PSS) (39), with a Cronbach’s of 0.66.
Salivary Cortisol
Saliva samples (100–200 µl) were collected at three time points during testing. The first sample was a baseline measure taken approximately 5 minutes after the interview team arrived at the home, and thus reflects the time before their arrival. To simplify the presentation of results, we will refer to this sample as time 0. Two subsequent cortisol samples were collected 25 and 50 minutes after the onset of the interview, respectively. The samples were purposely scheduled 25 minutes apart to account for the delay of approximately 20 to 25 minutes between cortisol secretion and subsequent salivary cortisol levels (40). All samples were collected in sterile collection tubes (Salivettes; Sarstedt, Newton, NC) and kept at room temperature. In order to determine the cortisol concentration in the saliva, the samples were sent via express mail to the Biochemical laboratory in the Psychobiology Department at the University of Trier, where they were stored at –20°C until analysis. After thawing, the samples were assayed in duplicate analyses using a standard time-resolved fluorescence immunoassay. Cortisol levels were determined using a competitive solid phase time-resolved fluorescence immunoassay with fluorometric end point detection. With a computer-controlled program, a standard curve was generated, and the cortisol concentrations of the samples were calculated. Synthetic saliva mixed with cortisol in a range from 0 to 100 nmol/l served as standards. The intra-assay coefficient of variation was between 3.28% and 4.72%. To meet normality assumptions, all salivary cortisol data were transformed using a natural log transformation.
Procedures
All participating households were visited unexpectedly by a team of health professionals who conducted an interview lasting approximately 1 hour with each woman. Informed consent to participate in the study was received on arrival of the research team. The interviews were conducted at any time throughout the day from 0800 to 1600 and included the assessment of several measures of physical and mental health, including the CES-D, an assessment of depressive symptomatology (35). Three samples of saliva were obtained at 0 minutes, 25 minutes, and 50 minutes after arrival to measure salivary cortisol levels. Because posture has effects on the HPA axis, all women remained seated throughout the visit (41).
Statistical Analysis
All data were analyzed using hierarchical linear modeling procedures with SPSS 11.5 Advanced Models MIXED software (SPSS Corp., Chicago, IL). Hierarchical linear modeling analytic techniques account for the hierarchical, nested structure of the data (in this case, three observations are nested within each individual), allow for modeling the nonindependence of residuals and distinguish between fixed and random effects (42,43). Two-tailed tests were used in all cases and significance levels were set at = 0.05. To address the hierarchical structure of the data, two levels of equations were estimated. On the first level, we estimated each individual’s baseline cortisol level (eg, intercepts) and cortisol reactivity (eg, slopes). On the second level, we estimated equations to predict differences in level 1 intercepts and slopes from level 2 predictors such as hours since awakening, age, and depressive symptoms.
To ease the interpretation of the results, several adjustments were made to the data. First, time was anchored at baseline (at baseline, time = 0) so that the cortisol intercepts (ß00) would reflect the average individual’s cortisol level at baseline (43). Second, all level 2, between-person variables such as age and depression scores were centered at their grand mean, producing an average of 0 (42,43). Third, because cortisol follows a diurnal rhythm, with higher levels after awakening and lower levels later in the day, we wanted to make an adjustment for this rhythm. Several recent studies have demonstrated that the time of day and time of awakening are independent predictors of cortisol responsivity (44,45). To address these issues, we created an individual-level covariate (time since awakening) that incorporates both sources of variability. This variable was also centered at its grand mean (6.5 hours from waking).
After these transformations were performed on the data, model fit indices, cortisol intercept and slope variability, and alternate residual variance-covariance matrix structures were examined. An autoregressive (AR-1) error matrix structure, in which all variance in the dependent variable caused by the previous score is removed, was determined to best fit the data. These procedures produced the following multilevel model:
|
|
where t is time of observation, i is individual, ß00 is the estimated intercept representing the average individual’s baseline level of cortisol, ß01 is the estimated average slope of cortisol levels over time (eg, cortisol reactivity), ß02 is the estimated effect of time since awakening on baseline cortisol, ß03 is the estimated effect of age on baseline cortisol, ß04 is the estimated effect of depressive symptoms on baseline cortisol, ß14 is the estimated effect of depressive symptoms on cortisol reactivity (change in cortisol over time), µ0i is the estimated variability between individuals in baseline cortisol, 
ti is the autoregressive term, and rti is the residual variance of cortisol between observations.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
|
Baseline
Our first hypothesis addressed whether elevations in depressive symptoms would be associated with higher baseline cortisol levels. Support for this hypothesis would be revealed by a significantly positive ß04 DEPRESSIONi term, which estimates the relationship between depressive symptoms and cortisol levels at baseline, after controlling for the effects of the other covariates. Results do not fully support this hypothesis (ß04 = 0.004; SE = 0.002; p < .10; Table 1), because there was a statistically nonsignificant trend for elevated baseline cortisol levels in women with higher levels of symptoms. There was an expected negative effect of time since awakening on baseline cortisol levels because women later in the day, who were further into their diurnal cycle, had lower baseline cortisol levels. There was no effect of age on baseline cortisol levels.
|
Stress Reactivity
The second hypothesis addressed whether elevations in depressive symptoms would be associated with blunted cortisol responses to stress, and results support this hypothesis. There was a significantly negative ß14 DEPRESSION*TIMEti interaction term (ß14 = –0.003; SE = 0.001; p < .05,), which estimates the effect of the depressive symptoms on cortisol changes over time. These results suggest that after controlling for the other covariates, women with more severe depressive symptoms exhibited more blunted cortisol responses to stress than those with less severe symptoms (Table 1). There were no effects of time since awakening or age on cortisol reactivity.
The depression by time interaction was probed further using simple effects tests (Figure 1) with three separate regression lines plotted at CES-D scores of 35 and 16, the recommended cutoffs for adult Mexican women and US samples, respectively (46), and a comparison group of CES-D scores of 10 (46). Results demonstrate that women with CES-D scores of 35 failed to mount a cortisol response to the naturalistic stressor (ßSIMPLE SLOPE, CES-D=35 = 0.028; SE = 0.023; NS). In contrast, women with CES-D scores of 16 (ßSIMPLE SLOPE, CES-D=16 = 0.085; SE = 0.017) or 10 (ßSIMPLE SLOPE, CES-D=10 = 0.103; SE = 0.017) exhibited significant increases in salivary cortisol in response to the stressor (both significant at p < .001). Subsequent analyses revealed that cortisol responses in women with CES-D scores of 16 and 10 did not significantly differ from one another. Thus, it appears that women with elevations in depressive symptoms exhibit blunted responses to a naturalistic stressor.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Although in the predicted direction, the failure to detect a statistically significant association between depressive symptoms and baseline cortisol levels was unexpected, given the previous literature and our large number of participants. Although our results were not statistically significant, there was a trend (p = .08) in the predicted direction (ie, higher baseline cortisol levels in more depressed women). Because we had only one baseline sample, we were not able to make a determination or reliable assessment of basal HPA functioning. Our results contrast with some of the psychiatric literature demonstrating elevated basal cortisol levels in depressed individuals, yet replicate those of another naturalistic study which also found no differences in basal cortisol levels between clinically depressed and nondepressed individuals (24). The baseline cortisol levels of this study (3.96 nmol/l) were slightly lower than the basal levels observed in both the clinically depressed (4.79 nmol/l) and nondepressed (4.87 nmol/l) samples in the aforementioned naturalistic study (24). Two possible explanations for the lower salivary cortisol levels in this sample compared with those observed in Peeters et al. (24) are that this sample consisted entirely of women, who tend to have lower levels of free, unbound cortisol as a result of higher CBG levels. Another possible explanation is that baseline/basal cortisol levels correspond to depressive symptom severity. Indeed, results of this study and others suggest that the severity of depressive symptoms may moderate cortisol responses to stress (23). The majority of previous research demonstrating elevated basal HPA activity has used patient samples. However, no clinical evaluations of MDD diagnosis were made in this study to allow us to distinguish between elevations in depressive symptoms and MDD.
These results also extend previous research that has revealed blunted cortisol responses to laboratory and naturalistic psychosocial stressors in clinically depressed samples to subclinical depressive symptoms. Before this study, most of what was currently known about depression and cortisol stress responses was based on patient samples. Although informative, for various reasons, patient samples may be biased in favor of more severely depressed individuals, patients more willing to get help, or patients with the means to obtain help (eg, high SES). This is the first known study to find this pattern of responsivity in subclinically depressed women. However, given the high levels of depressive symptomatology, it is likely that a subset of the sample was clinically depressed.
One question that often arises in research on depression and cortisol stress responses is whether blunted cortisol stress responses are a result of a ceiling effect of baseline cortisol elevations. However, results of this study revealed that a subset of women with elevations in depressive symptoms (CES-D of 35) had baseline cortisol levels similar to women with lower levels of depressive symptoms, yet still exhibited blunted cortisol responses to the stressor. This finding is consistent with the findings of Peeters et al. (24), who also failed to find basal cortisol differences between clinically depressed and nondepressed individuals. Taken together, the results of this study and those of Peeters et al. (24) suggest that the blunted cortisol responses observed in these women are not simply a result of a ceiling effect. Instead, it appears that high levels of depressive symptoms and clinical depression, regardless of baseline cortisol levels, may share cortisol nonresponsiveness to naturalistic stressors as a common characteristic.
At present, the etiology of blunted cortisol stress responses in women with depressive symptoms is unknown. One possibility is that women with high levels of depressive symptoms experience psychological disengagement from environmental stressors—in this case the unexpected arrival of researchers. In other words, might women with more depressive symptoms be blunted psychologically and physiologically? Unfortunately, there was no manipulation check to assure that the unexpected visit and interview actually provoked a psychological stress response. Because this study was conducted to take advantage of another study already in progress, the unexpected arrival of the research team, although face-valid as a stressor, was not planned a priori as a stressor. We expected that the cortisol response to this type of stressor would be of smaller magnitude than those observed using the TSST, given that a previous comparison of laboratory and field stressors revealed smaller cortisol responses in the field paradigm (47). However, the women in this study exhibited an average 27% increase from baseline levels after 25 minutes, when one would expect levels to decrease because of diurnal factors, suggesting that the stressor was successful in eliciting a cortisol response.
Although we do not have data to address whether the more depressed women were more or less reactive psychologically to the research visit and interview, we do have data on overall perceived stress. Women with higher levels of depressive symptoms also had the highest levels of perceived stress, as measured by Cohen’s PSS (39). Reanalyzing the data with depressive symptoms and perceived stress as independent predictors of cortisol responses did not change the results of this study (data not shown). There were no main or interactive effects of perceived stress on cortisol responses. However, the relatively low reliability of the PSS measure in this sample suggests that it may be a less sensitive index of overall stress in this population. Given these considerations, it is unlikely that the blunted cortisol responsivity observed in this study is a result of psychological blunting.
Achieving a naturalistic environmental stressor where women were visited unexpectedly without the control of the laboratory introduces some limitations to the interpretations of the study results. We did not control for nor have data available concerning several potential confounding variables such as food eaten, caffeine intake, smoking, medications, oral contraceptive use, menstrual cycle phase, body mass index, health problems, early life stress, or current posttraumatic stress disorder. It is possible that one or a combination of these confounding variables may have influenced the results. However, we believe that the large number of participants compensates for these potential confounds.
In sum, results of this study demonstrate that women with very high levels of depressive symptoms exhibit blunted cortisol responses to a naturalistic psychological stressor. Furthermore, this study extends the generalizability of its findings to a more disadvantaged and vulnerable group of individuals than typically studied. Future studies should compare cortisol stress responses between clinically depressed, subclinically depressed, and nondepressed individuals across a wide range of depressive symptoms. A question to be addressed by future research is whether it is depressive symptom severity or specific categories of depressive symptoms that predict stress responsivity (or lack thereof). Furthermore, because there is a growing body of evidence suggesting that cortisol levels per se may not be as important as the effectiveness of cortisol on its target receptors, future studies should also investigate the sensitivity of target cells to glucocorticoids (48). Finally, the potential health implications of the stress response pattern associated with clinical depression and depressive symptoms warrant further study. With further research in these areas, this line of work may yield insight into the mechanisms for the depression and disease association.
The authors thank the John D. and Catherine T. MacArthur Foundation Network on SES and Health, the Fogarty International Center of the National Institutes of Health, and the Mexican Government for supporting this research, in addition to Ryo Shiba, Francisco Papaqui, Gustavo Olaiz, Aurora Franco, Mauricio Hernandez, Stefano Bertozzi, and Juan Pablo Gutierrez at Mexico’s National Institute of Public Health, Andrea Gierens at the University of Trier, and Megan Gunnar at the University of Minnesota.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Received for publication July 3, 2004; revision received October 15, 2004.
DOI:10.1097/01.psy.0000156939.89050.28
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. E. Minton, M. Hertzog, C. R. Barron, J. A. French, and R. Reiter-Palmon The First Anniversary: Stress, Well-Being, and Optimism in Older Widows West J Nurs Res, December 1, 2009; 31(8): 1035 - 1056. [Abstract] [PDF]
|
|
|
|
 |
|
 A. Nierop, A. Bratsikas, R. Zimmermann, and U. Ehlert Are Stress-Induced Cortisol Changes During Pregnancy Associated With Postpartum Depressive Symptoms? Psychosom Med, November 1, 2006; 68(6): 931 - 937. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Giese-Davis, F. H. Wilhelm, A. Conrad, H. C. Abercrombie, S. Sephton, M. Yutsis, E. Neri, C. B. Taylor, H. C. Kraemer, and D. Spiegel Depression and stress reactivity in metastatic breast cancer. Psychosom Med, September 1, 2006; 68(5): 675 - 683. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
