| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychiatry and Neuropsychology, Maastricht University (F.P., N.A.N.), Maastricht, and Department of Clinical Epidemiology and Biostatistics, Free University Medical Center (J.B.), Amsterdam, The Netherlands.
Address reprint requests to: Frenk Peeters, MD, PhD, Department of Psychiatry, University Hospital Maastricht, P.O. Box 5800, 6202 AZ Maastricht, The Netherlands. Email: f.peeters{at}sp.unimaas.nl
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: Experience sampling methodology (self-reports of mood and events, with simultaneous saliva samples, 10 times each day for 6 consecutive days) and multilevel regression analysis were used to examine the relationship between events in daily life and salivary cortisol levels.
RESULTS: In contrast to healthy participants, depressed participants showed no increase in cortisol following negative events. Responses were even more blunted in depressed participants with a family history of mood disorders. Although the effects of negative events on cortisol responses appeared to be mediated by changes in mood, negative affect tended to be less closely associated with cortisol levels in depressed participants. Depressed women showed larger cortisol responses to negative events than depressed men. Positive events had no effect on cortisol levels in either group.
CONCLUSIONS: These results suggest that responses of the HPA axis to negative daily events and mood changes are blunted in MDD. Future studies will need to address whether these abnormalities disappear after clinical recovery.
Key Words: daily events, • depression, • gender, • mood, • salivary cortisol, • stress.
Abbreviations: BDI = Beck Depression Inventory;; ESM = Experience Sampling Method;; HPA = hypothalamic-pituitary-adrenal;; MDD = major depressive disorder;; NA = negative affect;; PA = positive affect;; SCID-I = structured clinical interview for DSM-IV.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Negative daily events ("hassles") are followed by increases in negative affect (NA) and decreases in positive affect (PA) (13–15). Mood changes have been found to mediate cortisol responses to everyday stressors (16–18). Given our recent finding that changes in NA and PA after daily hassles are blunted in depressed outpatients compared with healthy subjects (19), we hypothesize that cortisol responses to such events in MDD may also be blunted.
In the psychiatric and psychological literature, much attention has been paid to affective and biological responses to negative events, whereas the effects of positive daily events have been much less intensively investigated. Positive events are associated with lower NA and higher PA (13, 19–21). Evidence of effects of positive events on cortisol secretion is equivocal. Previous work reported decreases (22, 23), increases (24), or no changes (25–27) in cortisol levels after experimental induction of positive mood. Effects of positive events in daily life on cortisol have never been studied. Such knowledge is clinically potentially relevant, given suggestions that positive daily events and subsequent mood changes may act as a buffer against damaging effects of HPA axis abnormalities resulting from chronic stress and MDD (28).
In recent years, daily process designs such as the Experience Sampling Method (ESM) (29) have been shown to be valid and reliable techniques for the simultaneous assessment of events, mood, and cortisol in the natural environment (17, 18). In this study, we used ESM to examine cortisol responses to negative and positive events in the daily lives of depressed and healthy individuals. Our first aim was to test whether cortisol responses to daily events differ in magnitude between depressed and healthy individuals. Second, abnormalities of the HPA axis in MDD have been reported to be influenced by clinical characteristics such as severity (30), length of the current depressive episode (31, 32), number of previous episodes (33), and positive family history (34). Therefore, associations between these characteristics in depressed subjects and cortisol responses to events were tested. Third, associations between changes in negative and positive mood states and cortisol levels in both groups were investigated. Fourth, we examined whether the influence of daily events on cortisol is mediated by mood changes. Finally, we tested the influence of gender on cortisol responses, because stronger responses of the HPA axis to stress have been hypothesized to contribute to the higher prevalence of MDD in women (35).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
18 on the 17-item Hamilton Depression Rating Scale (37). Exclusion criteria were substance abuse in the last 6 months, psychotic symptoms, bipolar disorder, pregnancy, weight loss exceeding 15% in the previous 6 months, endocrine and rheumatic disorders, medications (including antidepressants) known to affect the HPA axis, and insufficient fluency in Dutch. In cases where previously prescribed antidepressants were judged to be clinically ineffective, these drugs were tapered off; subjects were then allowed to participate in the study after a medication-free interval of at least 1 week (this applied to 5 subjects, none of whom used fluoxetine). Use of previously prescribed low-dose benzodiazepines was allowed (8 subjects). Thirty-nine healthy subjects, matched as a group to the patient sample for gender and age, were recruited from available research pools and through a local newspaper advertisement. Additional exclusion criteria for healthy subjects were a lifetime history of any DSM-IV axis-I disorder, or a history of inpatient treatment for an axis I psychiatric disorder in a first-degree relative. The study was approved by a medical ethics committee, and written informed consent was obtained from all subjects.
Questionnaires
Patients as well as healthy subjects completed the Symptom Check List (SCL-90) (38). Patients also completed the Beck Depression Inventory (BDI) (39).
Ambulatory Sampling Procedure
The Experience Sampling Method (ESM) (29, 40) was used to collect data from participants at selected moments during their daily activities. Participants received auditory signals (beeps) from a wristwatch programmed to emit 10 beeps between 7:30 AM and 10:30 PM each day, at semirandom intervals of approximately 90 minutes. After each beep, participants completed self-report forms concerning current mood, any negative and any positive events that may have taken place since the last report, and various extraneous influences on cortisol secretion (for details see measures below). Participants completed ESM reports for 6 consecutive days, including a weekend. During a briefing session, study aims and procedures were explained. In a final session, the ESM booklets were checked for legibility and missing data. Compliance with the procedure was generally good. The criteria set for inclusion in the analyses (more than 20 ESM reports completed within 25 minutes after the programmed time of the beep) were met by all participants except 1 depressed participant.
Measures
Mood assessment
Momentary mood states were assessed with 16 adjectives rated on 7-point scales (1 = not at all, 7 = very). Factor analyses (principal components analysis with varimax rotation) on mean scores aggregated per subject and on within-subject z scores identified 2 mood factors with eigenvalues greater than 1. These factors accounted for 81.1% of the total variance in the analysis of the subject mean scores and 46.1% of the variance in the analysis of the within-subject z scores. Ratings on the items anxious, irritated, restless, tense, guilty, irritable, easily distracted, and agitated were averaged to form a negative affect (NA) scale (Cronbach’s alpha = 0.91 based on 4535 reports). Ratings on the items energetic, enthusiastic, happy, cheerful, talkative, strong, satisfied, and self-assured were averaged to form a positive affect (PA) scale (alpha = 0.95).
Event assessment
At each beep, participants were asked briefly to describe any positive and any negative event that may have taken place since the last ESM report. Although participants were instructed to report only events or situations that actually took place in their daily environment in the preceding interval, some event reports clearly referred to internal states (eg, current ruminations about past events, personal health concerns). Following preestablished criteria, the research team identified such internal events by consensus. We limited the analysis to the effects of external events on cortisol levels, because internal events reports are likely to confound event and mood measures. In the depressed group, 17% of all event reports referred to internal states compared with 7% in the healthy group.
Salivary cortisol
At the same time ESM forms were being completed, subjects collected saliva with cotton Salivettes (Sarstedt). They stored new saliva samples in their home freezers at the end of each day. On the ESM forms, the following information was also obtained to control for possible extraneous influences on cortisol secretion: recent physical exertion (rated on a 7-point scale) and any food intake, alcohol, coffee, smoking, or medication (eg, benzodiazepines) in the interval between the current and the previous beep. In the morning, subjects recorded the time of awakening and rated sleep quality the previous night (7-point scale).
Uncentrifuged saliva samples were stored at -20°C until analysis. Salivary free-cortisol levels were determined in duplicate by direct radioimmunoassay, using iodohistamine-125 coupled to cortisol-3CMO. The lower detection limit of the assay was 0.33 nmol/L, with mean inter- and intra-assay coefficients of variation of 8.3% and 4.3%, respectively. All samples from an individual were analyzed in the same assay to reduce sources of variability. Extreme cortisol values (>44 nmol/L) were excluded, which applied to 12 values from different individuals and all data (38 samples) from 1 depressed subject who had 11 extreme cortisol values.
Statistical Analysis
Because the ESM observations are nested within days within participants, we estimated the effects of daily events, mood states, and individual characteristics on cortisol secretion with multilevel or hierarchical linear analysis, a variant of multiple regression appropriate for nested data. We refer to the three levels in the model as beep-level, day-level, and person-level. The multilevel model was estimated using the program MLwiN (41).
A natural log transformation of the raw cortisol values yielded an unskewed response variable. The regression equation for cortisol at the beep level can be expressed as follows: 
|
|
where LogCort ijt is the transformed cortisol level of participant i at the t-th beep on day j. The intercept is denoted by b0m, xqijt is an entry of explanatory variable xq, bqm is the corresponding regression coefficient, and emijt is an error term. The index m indicates whether the participant is depressed (m = 1) or healthy (m = 0). By indexing all regression coefficients with label m, we estimated all effects separately for the depressed and healthy participants. The error term emijt is itself a function of error terms that capture the dependencies among the cortisol measurements. Such dependencies occur because two observations tend to be more similar if (1) taken at points closer in time on the same day, (2) taken on the same day rather than on 2 different days, and (3) taken from 1 participant rather than from different participants. To account for these sources of dependency, we (1) modeled the autocorrelation between 2 subsequent observations as an exponentially decaying function of the time interval between these observations, (2) included a random intercept and slope at the day level to account for day-to-day variability in the diurnal cortisol pattern, and (3) included a random intercept and slope at the person level to account for interindividual variability in cortisol levels. Ignoring these dependencies biases the estimated standard errors of the regression coefficients and may lead to incorrect inferences about the effects of the explanatory variables. In the presented model, separate autocorrelation terms are specified for the depressed and healthy participants. We do not specify separate within- and between-subject variances because they were not significantly different in the 2 groups.
We fitted the diurnal pattern of cortisol secretion by a fourth-degree polynomial, because this provided a better fit than the linear component alone. Fixed effects estimated at different levels included a number of potential confounders (beep-level: recent awakening, food intake, physical exertion, smoking, alcohol, coffee, benzodiazepines; day-level: self-reported sleep quality; person-level: age, use of oral contraceptives, and SCL-90 anxiety score). Of these, only the 2 variables with significant effects (recent awakening and food intake) were retained in the final model. The secretory peak in cortisol that occurs after morning awakening (42) was modeled by including a variable that took the value "1" if the saliva sample was collected less than 1 hour after awakening, and "0" otherwise. Food intake in the interval preceding a beep was also dummy-coded. Recent awakening (MDD; ß = 0.176, SE = 0.054 vs. healthy; ß = 0.198, SE = 0.052) and food intake (MDD; ß = 0.093, SE = 0.021 vs. healthy; ß = 0.118, SE = 0.019) resulted in comparable increases in cortisol levels in both groups.
To test our main hypotheses, we included negative events (1 if a negative event had occurred, and 0 otherwise) and positive events (1 if a positive event had occurred, and 0 otherwise) at the beep-level. We also included the following additional explanatory variables: gender (1 if female, -1 if male), severity of depression (BDI score centered around overall mean), duration of current depressive episode (centered around overall mean), history of previous depressive episode(s) (coded -1 or 1), and positive family history for mood disorders (coded -1 or 1). These variables were either centered around the grand mean or effect coded (ie, -1,1 instead of 0,1) so that inclusion in the model would not automatically change the previously estimated effects of events on cortisol. To examine the influence of gender and clinical characteristics on cortisol reactivity to events, we estimated the interaction effects of daily events (both negative and positive) with these variables. In a second step, we entered the variables NA (negative affect, scaled 0–6) and PA (positive affect, scaled 0–6), as well as their interactions with negative and positive events.
To test the significance of the regression coefficients, z-scores were calculated by dividing the estimated effect by its standard error (43). We also tested whether the regression coefficients were statistically different inpatient vs. control groups. Two-tailed tests were used, even when hypotheses were directional. Significance levels were set at alpha = 0.05.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
Effects of Diurnal Cycle, Depression, and Extraneous Influences on Cortisol Secretion
Figure 1 shows observed cortisol values during the day in MDD and control groups. Both groups displayed a clear diurnal pattern, with high cortisol values in the early morning declining during the rest of the day.
|
|
With respect to clinical characteristics in the depressed group, subjects with a family history of mood disorders showed additional blunting of cortisol responses (ß = -0.083, SE = 0.028, z = 2.96, p < .01). No other clinical characteristics significantly moderated cortisol responses.
The next model (see Model 2 in Table 2) added NA and PA as predictors to evaluate the effect of mood on cortisol. In general, increases in NA were associated with higher cortisol levels, with a trend toward a smaller effect in the depressed group. It should be noted that after addition of NA, the effect of negative events on cortisol was attenuated in healthy subjects. In other words, affect appears to mediate to some extent the relationship between negative events and cortisol found in the first model. Neither PA levels nor event-related changes in PA were associated with cortisol levels in either group (data not shown).
Gender Effects on Cortisol Reactivity
Finally, we examined whether depressed men and women differed with respect to cortisol reactivity. Gender did not influence basal cortisol levels in the depressed group (ß = 0.007, SE = 0.052, NS), but depressed women showed larger cortisol responses to negative events than depressed men did (gender by event interaction; ß = 0.067, SE = 0.029, p < .05). Cortisol responses to positive events were similar in depressed women and men. Within the group of healthy participants, no effects of gender on cortisol levels or reactivity were found (data not shown).
As allowed by the protocol (see Methods), 8 depressed subjects reported daytime use of benzodiazepines during the study (on 68 of 2039 ESM reports). To exclude the possibility that benzodiazepine use influenced cortisol levels, we repeated the analysis after excluding these 8 subjects. All findings remained the same.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
The observed increase in salivary cortisol after negative daily events in healthy subjects replicates earlier findings (17, 18) . The reduced cortisol reactivity to negative events observed in the depressed group requires explanation. Previous reports of blunted cortisol responses in MDD have been linked to higher basal cortisol levels (6, 7, 9). In the current outpatient sample, however, basal levels were normal, and individual differences in basal levels were controlled for in the analysis. In remitted depressed subjects with normal basal levels, Trestman et al. (7) also found a trend toward blunted cortisol responses to a laboratory stressor. The lack of cortisol response may reflect the blunted mood responses after negative events we have found in depressed subjects, even though they appraise such events as more unpleasant, important, and stressful than healthy individuals (19). The trend toward smaller effects of NA on cortisol responses in depressed subjects as compared with healthy subjects (as shown in Table 2) may also indicate hyporesponsivity of the HPA axis to external stimuli that activate negative affect.
We found more erratic cortisol levels in MDD (44), but the current results indicate that these are not directly attributable to daily life experiences. Cortisol responses to acute stressors are thought to be highly adaptive to the organism, ensuring physiological, affective, cognitive, and behavioral changes, followed by a rapid return to homeostasis (45). In this sense, blunted responses are evidence of a general dysregulation of the HPA axis in MDD (46). Even greater blunting in depressed subjects with a family history of mood disorders suggests that altered responsiveness of the HPA axis may reflect a genetic vulnerability; this is in agreement with previously reported genetic influences on MDD-related abnormalities in basal cortisol levels (47), and in feedback mechanisms of the HPA axis (34).
Positive events and changes in PA had no effects on cortisol. As reported elsewhere (19), mood changes after positive events in daily life were smaller than the mood changes seen after negative events and may thus be insufficient to influence the HPA axis. Moreover, characteristics of stimuli capable of lowering cortisol are poorly understood.
Although they remained less reactive than healthy participants did, depressed women showed larger cortisol responses to negative events than depressed men did. This cannot be explained by larger mood responses after daily events, because these were similar in depressed women and men in the same sample (19). It is possible that additional variables such as social support (48), cognitive processes (eg, rumination) (49), modulatory effects of gonadal steroids (35), and behavioral responses (50) contribute to the observed gender difference in cortisol reactivity in the depressed group.
This study has some limitations. Even in a time-sampling approach like ESM, retrospective bias cannot be entirely ruled out. Event assessment was based on self-reports only, because there was no practical means of verifying event occurrence. Second, the causal association between events, mood, and cortisol cannot be firmly established, because data were collected at the same points in time. However, prior events, controlling for effects of current events, were associated with persistent changes in both mood states, which supports our assumption that events influenced mood and cortisol and not vice versa (19). Third, we did not obtain data on childhood abuse, which has been shown to alter HPA responses to a laboratory stressor in adulthood in women (51).
The finding that small hassles and mood fluctuations in daily life influence cortisol levels differently in depressed and healthy subjects points to differences in both psychological experience and biological systems in their daily lives. More generally, our results fit well with a dysregulation hypothesis of MDD (46); a dysregulated system is characterized by impairment of homeostasis, erratic output, disrupted periodicities, abnormal responsivity to environmental stimuli, slower return to basal activity after perturbation, and normalization after effective treatment. The clinical significance of these psychological and biological alterations for illness course and response to treatment requires further examination. Finally, future studies should investigate whether abnormal responses to daily events in MDD normalize after clinical recovery.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication March 18, 2002.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  L. Z. FitzGerald, P. Kehoe, and K. Sinha Hypothalamic--Pituitary-- Adrenal Axis Dysregulation in Women With Irritable Bowel Syndrome in Response to Acute Physical Stress West J Nurs Res, November 1, 2009; 31(7): 818 - 836. [Abstract] [PDF]
|
|
|
|
 |
|
 S. A. Vreeburg, W. J. G. Hoogendijk, J. van Pelt, R. H. DeRijk, J. C. M. Verhagen, R. van Dyck, J. H. Smit, F. G. Zitman, and B. W. J. H. Penninx Major Depressive Disorder and Hypothalamic-Pituitary-Adrenal Axis Activity: Results From a Large Cohort Study Arch Gen Psychiatry, June 1, 2009; 66(6): 617 - 626. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Kimhy, P. Delespaul, H. Ahn, S. Cai, M. Shikhman, J. A. Lieberman, D. Malaspina, and R. P. Sloan Concurrent Measurement of "Real-World" Stress and Arousal in Individuals With Psychosis: Assessing the Feasibility and Validity of a Novel Methodology Schizophr Bull, May 8, 2009; (2009) sbp028v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. C. Wichers, I. Myin-Germeys, N. Jacobs, G. Kenis, C. Derom, R. Vlietinck, P. Delespaul, R. Mengelers, F. Peeters, N. Nicolson, et al. Susceptibility to Depression Expressed as Alterations in Cortisol Day Curve: A Cross-Twin, Cross-Trait Study Psychosom Med, April 1, 2008; 70(3): 314 - 318. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. K. Adam, L. C. Hawkley, B. M. Kudielka, and J. T. Cacioppo Day-to-day dynamics of experience-cortisol associations in a population-based sample of older adults PNAS, November 7, 2006; 103(45): 17058 - 17063. [Abstract] [Full Text] [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]
|
|
|
|
|
|
G. E. Miller, N. Rohleder, C. Stetler, and C. Kirschbaum Clinical Depression and Regulation of the Inflammatory Response During Acute Stress Psychosom Med, September 1, 2005; 67(5): 679 - 687. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. M. Burke, L. C. Fernald, P. J. Gertler, and N. E. Adler Depressive Symptoms Are Associated With Blunted Cortisol Stress Responses in Very Low-Income Women Psychosom Med, March 1, 2005; 67(2): 211 - 216. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
