Effects of mood and rumination on cortisol levels in daily life: An ambulatory assessment study in remitted depressed patients and healthy controls

Psychoneuroendocrinology (2013) 38, 2258—2267

Available online at www.sciencedirect.com

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / p s y n e u e n

Effects of mood and rumination on cortisol levels in daily life: An ambulatory assessment study in remitted depressed patients and healthy controls Silke Huffziger a,*, Ulrich Ebner-Priemer b, Vera Zamoscik c, Iris Reinhard d, Peter Kirsch c, Christine Kuehner a a

Research Group Longitudinal and Intervention Research, Department of Psychiatry and Psychotherapy, Central Institute of Mental Health, Medical Faculty Mannheim, Heidelberg University, Germany b Karlsruhe Institute of Technology, Germany c Department of Clinical Psychology, Central Institute of Mental Health, Medical Faculty Mannheim, Heidelberg University, Germany d Department of Biostatistics, Central Institute of Mental Health, Medical Faculty Mannheim, Heidelberg University, Germany Received 21 December 2012; received in revised form 8 April 2013; accepted 19 April 2013

KEYWORDS HPA axis; Basal cortisol; Remitted depression; Mood; Rumination; Daily life; Ecological momentary assessment; Experience sampling

Summary The influence of naturally occurring emotional and cognitive experiences on hypothalamic—pituitary—adrenal axis (HPAA) activity is still underinvestigated, particularly in clinical populations. The present study examined effects of mood and rumination on cortisol levels in daily life in remitted depressed patients with recurrent episodes or a chronic precourse (n = 31) and healthy controls (n = 32). Ambulatory assessment of subjective variables (valence, calmness, energetic-arousal, ruminative self-focus), daily stressors, and saliva cortisol samples was performed five times a day on two consecutive workdays, whereby cortisol was collected 20 min after the subjective assessments. In addition, depressive symptoms and trait rumination (brooding, reflection) were measured retrospectively. Multilevel models revealed that remitted depressed patients showed lower cortisol activity compared to healthy controls. Depressive symptoms and trait rumination did not predict HPAA activity, whereas, by controlling for daily stressors, higher daily means of ruminative self-focus and lower daily means of valence, energetic arousal and calmness were associated with higher daily cortisol levels. Separate analyses per group revealed that mean daily ruminative self-focus predicted higher cortisol in both samples. In contrast, lower daily means of calmness, but also of valence and energetic arousal, were significantly linked to higher cortisol output only in healthy controls, but not in the patient sample. These findings indicate that naturally occurring rumination and low mood are associated with increased

* Corresponding author at: Research Group Longitudinal and Intervention Research, Department of Psychiatry and Psychotherapy, Central Institute of Mental Health, Medical Faculty Mannheim/Heidelberg University, J5, 68159 Mannheim, Germany. Tel.: +49 621 1703 6058; fax: +49 621 1703 1205. E-mail address: [email protected] (S. Huffziger). 0306-4530/$ — see front matter # 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.psyneuen.2013.04.014

Effects on cortisol levels in daily life

2259

activation of the HPAA in daily life. Moreover, our data revealed a potentially reduced mood— cortisol coupling in remitted recurrent depression, possibly indicating that during the course of recurrent depression HPAA activation might become less responsive toward subtle emotional experiences in natural contexts. # 2013 Elsevier Ltd. All rights reserved.

1. Introduction Major depression as a stress-related disorder has been linked to alterations of the hypothalamic—pituitary—adrenal axis (HPAA). In acute depressive states, a substantial subgroup of patients shows hyperactivation of the HPAA with elevated basal cortisol (Stetler and Miller, 2011). However, cortisol downregulation has also been observed, indicating the existence of different depression pathophysiologies (Fries et al., 2005; Lamers et al., 2012). Furthermore, consistent with the notion that HPAA abnormalities are not just state-dependent but linked to depression vulnerability (Pariante and Lightman, 2008), there is evidence that after remission of acute depression HPAA activity remains dysregulated. Here, empirical findings involve both reduced and increased basal cortisol levels. Young et al. (2000) used repeated assessments of morning (8 am) and evening (before bedtime) saliva cortisol samples over 14 days and identified a trend for higher mean cortisol levels in monozygotic twins with a history of depression compared to never-depressed twins. Similarly, Lok et al. (2012), by using two consecutive morning (8 am) and one single evening (10 pm) sample showed that highly recurrent remitted depressed patients displayed higher mean cortisol concentrations than nondepressed individuals. In contrast, Ahrens et al. (2008) collected individual morning (8 am) and afternoon (4 pm) samples over seven days and reported reduced morning cortisol in women with remitted depression compared to healthy women, whereas afternoon cortisol levels did not differ between groups. Furthermore, lower cortisol levels have been linked to indicators of a more severe previous depression course. Gex-Fabry et al. (2012) examined average cortisol exposure over the day by collecting seven saliva cortisol probes per day at six occasions during a 14 months period and found that a longer illness duration predicted lower cortisol output (AUC). Similarly, in the study by Lok et al. (2012), more previous depressive episodes predicted lower mean cortisol output in remitted depressed patients. Finally, Bockting et al. (2012) found that lower morning cortisol (sampled at 8 am on two consecutive days) predicted earlier recurrence of depression. To explain hypocortisolism in recurrent depression, it has been suggested that lower cortisol levels might occur after prolonged hyperactivation of the HPAA due to its eventual downregulation (Ahrens et al., 2008). While HPAA downregulation can have beneficial effects by protecting against allostatic overload (Fries et al., 2005; Miller et al., 2007), hypocortisolism also points to adrenal exhaustion and can imply detrimental effects, for example through overactive immune responses and inflammation (Raison and Miller, 2003). Very little is known about factors that influence HPAA functioning in daily life. However, the identification of such factors might help to explain the large inter- and intraindividual variability in cortisol activity and to dissolve related diverging findings. While laboratory stressors have been

rather consistently identified to elicit cortisol increases (Kudielka and Wu ¨st, 2010), few studies have investigated possible associations of naturally occurring emotional and cognitive processes with cortisol activity in real life. Recently, Kudielka et al. (2012) have stressed the importance to transfer research on HPAA regulation from the lab to ecologically valid contexts, and there is evidence that cortisol responses in the lab do not necessarily converge with cortisol responses in natural settings (Wolfram et al., 2012). To study HPAA regulation in daily life, ambulatory assessment (AA) methods can be applied that capture real-time, realworld experiences, preferably with electronic devices (Trull and Ebner-Priemer, 2013). Using AA, it becomes thus possible to repeatedly measure both subjective experiences and cortisol activity over the day in real life which enhances reliability and generalizability of identified associations. A small number of previous AA studies have addressed possible influences of emotional experiences on cortisol activity during normal daily life. These studies found that higher momentary negative affect was linked to higher cortisol in community samples (Smyth et al., 1998; Hanson et al., 2000; Adam, 2006; Jacobs et al., 2007; Matias et al., 2011), in pregnant women (Giesbrecht et al., 2012), and in siblings of psychotic patients (Collip et al., 2011). To our knowledge, there is only one clinical study that compared associations between affect and cortisol in daily life in acutely depressed patients and healthy controls (Peeters et al., 2003). This study identified similar cortisol levels over the day in the two groups. Moreover, negative affect predicted higher cortisol levels in both groups; however, this effect was marginally weaker in patients compared to controls. Furthermore, patients also showed reduced cortisol reactivity to daily events compared to controls in this study. Thus, this study suggests that in depressed individuals HPAA activity in response to daily experiences is reduced. Similar findings of reduced stress—cortisol coupling in daily life have been reported in highly recurrent bipolar patients (Havermans et al., 2011). Thus, specific alterations of psychoneuroendocrine regulation within the flow of subjective experiences in daily life might represent an important pathophysiologic marker of depressive and bipolar disorders. However, systematic research on this topic is still missing. In addition to emotional experiences, cognitive processes, specifically rumination, might also influence cortisol activity. Rumination has been defined as repeatedly thinking about one’s negative mood and its possible causes and consequences, and has been postulated as an important cognitive vulnerability factor for depression (Nolen-Hoeksema et al., 2008). Longitudinal studies indicate that rumination predicts later depression particularly in nonclinical samples (NolenHoeksema et al., 2008; Huffziger et al., 2009). According to the perseverative cognition hypothesis by Brosschot et al. (2005), rumination entails an ongoing mental representation of a stressor, which should be associated with prolonged

2260

S. Huffziger et al.

activation of stress-related physiological systems such as the HPAA. However, findings on associations between rumination and cortisol have been inconsistent, partly due to different conceptualizations and assessments of both rumination and cortisol (see review by Zoccola and Dickerson, 2012). For example, individuals with high levels of habitual rumination in response to negative mood showed an attenuated cortisol awakening response (Kuehner et al., 2007) and reduced cortisol reactivity in the lab (Zoccola et al., 2008). In contrast, stress-related trait rumination was positively correlated with evening cortisol (Rydstedt et al., 2009). Findings on laboratory-assessed cortisol and state rumination have been more consistent. Experimental laboratory studies found that rumination measured after a stress task was linked to higher subsequent cortisol levels (Zoccola et al., 2008) and that induced rumination led to reduced cortisol decline in individuals with higher depressive symptoms (Kuehner et al., 2009). However, all studies on rumination and cortisol so far lack ecological valid assessments and were restricted to nonclinical samples. Thus, we do not know whether associations between rumination and cortisol also occur in daily life and whether individuals with and without a history of depression might differ in their cortisol response to rumination in this context. The present study used AA to examine natural cortisol activity in daily life and its possible determinants in remitted depressed patients and healthy controls. First, we investigated possible differences in overall cortisol levels between the two groups. Second, we examined effects of depressive symptoms and trait rumination — measured with retrospective assessments — on cortisol levels over the day. Third, we examined whether daily life mood and rumination would influence cortisol activity over the day. In this context, we were interested in the (net-)effects of these state characteristics beyond those of possible confounders such as daily stressors.

2. Method 2.1. Participants The study initially included 32 remitted depressed patients aged 18—55 with at least two previous major depressive Table 1

episodes (n = 30) or a previous chronic major depressive episode of at least two years according to DSM-IV (n = 2). All patients had to be in a state of partial or full remission, i.e. did not fulfill the criteria of a Major Depressive Episode according to DSM-IV, for at least two months. Furthermore, 32 age-, sex-, and education-matched healthy controls without current or lifetime diagnosis for a depressive disorder were included. Exclusion criteria for both samples involved bipolar and psychotic disorders, substance dependence, current substance abuse, general anxiety disorder, current obsessive-compulsive, posttraumatic stress, and eating disorder according to DSM-IV as well as hormonal medication affecting the HPAA. Psychopathology-related in- and exclusion criteria were assessed with the Structured Clinical Interview for DSM-IV axis I (SCID-I; Wittchen et al., 1997). After the SCID-I, one patient withdraw from the AA resulting in a final patient sample of n = 31. Almost half of the patients (n = 15, 48.4%) reported 4 previous depressive episodes, 7 (22.6%) and 6 (19.4%) patients reported 2 and 3 previous episodes. The mean age of depression onset was 22.8 years (SD = 11.10). One patient fulfilled criteria for a comorbid current agoraphobia and another patient for a comorbid current social phobia. Demographic characteristics of both samples are presented in Table 1. The study was approved by the local ethics committee of the University of Heidelberg. All participants gave written informed consent.

2.2. Procedure Participants were recruited using advertisements indicating a study on ‘‘thoughts and feelings’’, which were placed in local newspapers as well as on the homepage of the Central Institute of Mental Health where the study was carried out. After a telephone prescreening, a trained clinical psychologist administered the SCID-I and the Montgomery and Asberg Depression Rating Scale (MADRS) in an individual session. Here, participants also received detailed instructions for the AA including the cortisol sampling and filled in several questionnaires (see below). Directly following the individual session, participants performed the AA on two consecutive days and a few days later returned to our lab where they also underwent an fMRI experiment, which is not subject of the present analyses.

Demographic and baseline characteristics for remitted depressed patients and healthy controls.

Age Gender (male:female) Education (% with high school degree) Work situation (% in regular job or education) Marital status (% married or living together) BDI-II MADRS Previous inpatient treatment due to depression Current psychotherapy Current psychotropic medication Current non-psychotropic medication a

Patients (n = 31)

Controls (n = 32)

Test statistic

p

45.42 (7.98) 9:22 64.5% 80.6% 48.4% 9.61 (8.27) 5.45 (4.90) 71.0% 35.5% 25.8% 22.6%

44.50 (7.86) 10:22 62.5% 81.3% 81.3% 3.41 (3.93) 1.31 (2.33) — — — 25.0%

t = 0.46 Chi2 = 0.04 Chi2 = 0.03 Chi2 = 0.004 Chi2 = 8.71 t = 3.78 t = 4.26

.647 .848 .868 .951 .003 <.001 <.001

BDI-II, Beck Depression Inventory-Revised; MADRS, Montgomery and Asberg Depression Rating Scale. a Mainly blood pressure medication.

Chi2 = 0.05

.822

Effects on cortisol levels in daily life

2261

2.3. Ambulatory assessment (AA)

2.4. Saliva cortisol

The AA was conducted over two consecutive weekdays with personal digital assistants (PDAs, Palm Tungsten E2, Palm Inc.). The procedure was implemented using IzyBuilder software (IzyData Ltd., Fribourg, CH). There were ten subjective assessments per day, with the first assessment at 8 am and the last at 8 pm. We realized a fixed sampling scheme for all participants with inter-assessment intervals that varied between 75 and 85 min. Assessments could be delayed by 15 min. At each subjective assessment, which was indicated by a beep, participants rated momentary mood and ruminative self-focus and indicated whether they had experienced a stressful event since the last beep. Momentary mood was assessed with six bipolar items which were collapsed into the three scales positive valence, calmness, and energetic-arousal (valence: ‘‘content—discontent’’, ‘‘unwell—well’’; calmness: ‘‘agitated—calm’’, ‘‘relaxed—tense’’; energetic—arousal: ‘‘tired—awake’’, ‘‘full of energy—without energy’’). Item scores ranged from 0 to 6. Scale scores were partly recoded so that higher scores indicated more positive mood states. The present mood scales are explicitly recommended for use in AA research and exhibit good reliability and sensitivity for change (Wilhelm and Schoebi, 2007). Momentary ruminative self-focus was assessed with the two items ‘‘At the moment, I am thinking about my problems’’ and ‘‘. . .I am thinking about my feelings’’ which were averaged. Items were rated on a scale from 0 (not at all) to 7 (very much) and have been shown to be suitable for an AA-context (Moberly and Watkins, 2008; Huffziger et al., 2012). In addition to momentary scores of mood and ruminative self-focus, aggregated scores were calculated (daily means and overall means). Stressful events were assessed with the question ‘‘have you experienced a stressful event since the last beep?’’. If participants indicated ‘‘yes’’, they also noted the type of event from a predefined list of categories (interpersonal stressor, demand-related stressor, other stressors). In addition to the ten subjective assessments of mood and ruminative self-focus, participants collected five saliva cortisol samples per day. Participants could choose their individual wake times but were requested to get up by no later than 7 am. Cortisol samples were collected 20 min after the second (around 9.40 am), third (around 11 am), sixth (around 3 pm), eighth (around 5.50 pm), and tenth (around 8.20 pm) subjective assessment. The cortisol samples were signaled by a beep of the PDA which indicated the respective sample number. Importantly, at each subjective assessment that preceded a cortisol sample participants were instructed not to eat, drink, smoke, or brush their teeth in the upcoming 20 min (compliance instructions). After collection of the samples, participants indicated on the PDA whether they had followed the compliance instructions and how much they had physically exercised. By realizing a time-lag of 20 min between subjective assessments and cortisol samples, we were thus able to control for confounding effects and, more importantly, to examine the actual influence of subjective variables on cortisol, since cortisol reacts with a time lag of 10—30 min (Schlotz et al., 2008).

Saliva cortisol samples were collected using cotton swabs (‘‘Salivette’’, Sarstedt). After storage in the participants’ home freezers, samples were stored at 20 8C in our laboratory until biochemical analysis at the laboratory of Prof. Kirschbaum (Dresden, Germany). There, samples were centrifuged at 3000 rpm for 5 min, which resulted in a clear supernatant of low viscosity. Salivary cortisol concentrations were measured using commercially available chemiluminescence-immunoassay with high sensitivity (IBL International, Hamburg, Germany). The intra- and interassay coefficients for cortisol were <8%. In total, participants collected 610 out of 630 lagged cortisol samples. According to the PDA recorded compliance reports, 7 samples (3 in the patient group) were excluded due to recent eating, 3 samples (2 in the patient group) due to recent drinking, and 1 sample due to delayed collection. Furthermore, 6 samples (2 in the patient group) were excluded since they were 4 SDs above the mean value for that sample across participants, leaving 593 samples for final analyses. Thus, overall compliance for cortisol assessment reached 94.1%.

2.5. Interviews and questionnaires Inclusion and exclusion criteria were assessed with the SCID-I (see above), which has demonstrated good reliability and validity (Wittchen et al., 1997). Depressive symptoms during the previous two weeks were assessed with the self-rated Beck Depression InventoryRevised (BDI-II; German version Hautzinger et al., 2006) and the interviewer-rated Montgomery and Asberg Depression Rating Scale (MADRS; Montgomery and Asberg, 1979). Both scales exhibited acceptable Cronbachs as in the present samples (BDI-II: patients .91, controls .78; MADRS: patients .80, controls .73). For statistical analyses, we collapsed the two symptom measures into an overall composite score for depressive symptoms (DEP) by averaging the z-standardized BDI-II and MADRS scores (z-standardization was performed with respect to the total sample). Trait rumination in response to negative mood was assessed with the two 5-item subscales brooding and reflection of the Response Styles Questionnaire (Treynor et al., 2003; German version Huffziger and Ku ¨hner, 2012). Brooding describes the tendency to passively ruminate on one’s deficits and problems (‘‘why can’t I handle things better’’) and is supposed to represent a dysfunctional rumination component. Reflection means to neutrally think about one’s self (‘‘I go someplace alone to think about my feelings’’). In the present samples, both scales had acceptable Cronbachs as (brooding: patients .83, controls .69; reflection: patients .84, controls .78).

2.6. Statistical analyses Cortisol data were analyzed with multilevel models with assessments/samples (level-1) nested within participants (level-2). In our final analyses, we did not include day as a separate level since preliminary analyses revealed that random intercepts on the day level were nonsignificant and the

2262

S. Huffziger et al.

Table 2 Intercorrelations, means, and standard deviations of depressive symptoms, habitual brooding, habitual reflection, dailylife mood (valence, calmness, energetic arousal), and daily-life ruminative self-focus in remitted depressed patients and healthy controls. DEP

Brooding Reflection Valence a Calmness a EnergeticA a Rumination a M (SD) Patients (n = 31) Controls (n = 32)

DEP Brooding Reflection Valence a Calmness a EnergeticA a Rumination a

.23 .58 ** .35 .49 ** .41 * .32 .44 *

.33 .50 ** .37 * .29 .47 **

*

.42 .59 ** .23 .24 .29 .30

.54 .01 .08

**

.92 ** .70 ** .84 **

*

.44 .02 .02 .90 ** .68 ** .73 **

**

.55 .08 .22 .84 ** .72 ** .60 **

**

.70 .15 .47 ** .60 ** .50 ** .53 **

0.43 10.87 10.97 3.97 3.67 3.48 1.32

(1.07) (3.84) (3.67) (1.01) (0.96) (0.89) (1.17)

0.46 8.13 8.53 4.68 4.55 4.26 0.82

(0.47) (2.43) (3.31) (1.15) (1.03) (0.86) (0.87)

Note. Correlations above the diagonal and in bold font refer to the patient sample, correlations below the diagonal to the controls. DEP, composite score for depressive symptoms; EnergeticA, energetic arousal. a Daily-life momentary variables were aggregated over the two days. * p < .05. ** p < .01.

two-level models yielded a better fit to the data than the three-level models, justifying the use of the more parsimonious two-level solution.1 Moreover, it has been stressed by Nezlek (2012) that two days would be too less to reliably estimate the corresponding variance components. Cortisol values, which were log-transformed (natural log, lncort) due to their right-skewed distribution, were entered in the models as dependent variable. After performing an empty multilevel model to extract the intraclass correlation coefficient (ICC), all models included time (coded as daily sample number) as fixed and random effect in addition to a random intercept. In a next step, combining both groups, we added fixed effects of day and possible confounding variables (psychotropic medication, other medication, oral contraceptive use, habitual smoking, menstrual cycle phase (0 = male or menopause, 1 = follicular phase = menstrual cycle day 1—13, 2 = luteal phase = menstrual cycle day > 13), relationship status, waking time, and physical activity). If any of these variables were significant, they were retained in the following models. In a further step, we tested whether sex and group status (patients vs. controls) had a significant effect on diurnal patterns or overall concentrations of cortisol. Therefore, we entered main effects of group and sex as well as the interactions group  time, sex  time, group  sex, and group  sex  time to the model. Sex and its interactions were only retained in the models when significant. Finally, in our main analyses, we separately included our key predictors (on level-2: depressive symptoms, brooding, reflection, on level-1: ruminative self-focus, mood) as main and interaction effects with group. Importantly, in the models with level-1 predictors, we included daily means (aggregated scores) and momentary scores (cluster-mean centered) of the respective subjective variables to separate the effect of individuals’ daily means from the momentary effect. Cluster-mean centering of the momentary scores (centering on participants’ means; cf. Enders and Tofighi, 2007) allowed us to thoroughly disentangle the within-person

1 Importantly, by using the corresponding three-level models we identified comparable results for the examined fixed effects.

from the between-person (aggregated scores) effects. Furthermore, the models with level-1 predictors also controlled for daily life stressful events. All models were estimated with Maximum Likelihood (ML). Analyses were performed with IBM SPSS Version 20.

3. Results 3.1. Descriptive and baseline cortisol analyses As depicted in Table 1, remitted depressed patients and controls did not significantly differ with respect to age, gender, education, work situation, and current non-psychotropic medication. However, a lower percentage of patients compared to controls was married or living with a partner.2 Moreover, patients had significantly higher BDI-II and MADRS scores (see Table 1) as well as higher scores of DEP ( p < .001), brooding ( p = .001), reflection ( p = .007), and momentary rumination ( p = .058) and lower scores of valence ( p = .012), calmness ( p = .001), and energetic arousal ( p = .001) than controls. Means and SDs of the latter variables are presented in Table 2. We first specified empty multilevel models with log-transformed cortisol as dependent variable and identified an ICC of .16 in the total sample, and of .09 in the patients and .21 in the controls. Thus, in the total sample, 16% of the variance in cortisol (9% in patients, 21% in controls) was accounted for by the person level, while the remaining variance was due to within-person variability. In the next step, we added the fixed and random effects for time as well as fixed effects of possible confounding variables. The following variables were added: day (0, 1), psychotropic medication, other medication, oral contraceptive use,

2 Since relationship status might influence daily life experiences, we analyzed whether there were differences between married/ cohabiting and single participants in momentary rumination, valence, calmness, and energetic arousal. However, t-tests for independent samples revealed no significant group differences (all ps > .05).

Effects on cortisol levels in daily life

2263

Table 3 Results of separate multilevel models for the effects of daily-life mood and rumination on momentary cortisol in remitted depressed patients and healthy controls. Patients (n = 31) Fixed effect

B

Controls (n = 32) SE

p

B

SE

p

Mean daily valence Momentary valence

0.08 0.02

0.04 0.03

.065 .518

0.16 0.02

0.04 0.03

<.001 .444

Mean daily calmness Momentary calmness

0.03 0.007

0.05 0.03

.532 .832

0.19 0.003

0.05 0.03

<.001 .928

Mean daily energetic-arousal Momentary energetic-arousal

0.08 0.02

0.05 0.03

.098 .550

0.11 0.04

0.05 0.02

.020 .083

Mean daily rumination Momentary rumination

0.09 0.02

0.04 0.02

.015 .322

0.15 0.02

0.05 0.03

.003 .343

Note. The dependent variable cortisol was entered log-transformed. Momentary scores were group-mean centered. Models were based on 291 observations in the patients and 302 in the controls. Models included random intercepts as well as random and fixed effects for time, fixed effects for stressful events, and in the patients, fixed effects for psychotropic medication.

habitual smoking, menstrual cycle phase (0 = male or menopause, 1 = follicular phase = menstrual cycle day 1—13, 2 = luteal phase = menstrual cycle day > 13), relationship status, waking time, and physical activity (the latter as level-1 variable). For all possible confounders, main and interaction effects with time were added. The analysis revealed nonsignificant interaction effects (all ps > .10), which were consequently excluded from the model. The remaining model identified a significant main effect of time (B = 0.33, SE = 0.02, p < .001) together with a significant main effect of psychotropic medication (B = 0.34, SE = 0.12, p = .005), while all other possible confounding variables were nonsignificant (all ps > .05). That is, cortisol significantly decreased over the day and the intake of psychotropic medication was associated with higher cortisol. Subsequent analyses therefore controlled for the effects of time and of psychotropic medication. We then tested whether group (controls as reference) and sex (females as reference) as well as the interactions group  time, sex  time, group  sex, and group  sex  time influenced cortisol output over the day. The analysis revealed that all interaction terms were non-significant ( ps > .05). After excluding these interactions, we found that sex did not exhibit a significant effect on cortisol (B = 0.13, SE = 0.09, p = .136). In contrast, group status was significantly associated with cortisol output (B = 0.19, SE = 0.09, p = .032), indicating that patients showed lower cortisol concentrations compared to controls. The following cortisol mean scores (in nmol/l, raw cortisol) were identified: patients M = 6.98, SD = 5.57 (291 cortisol samples); controls M = 7.94, SD = 5.33 (302 cortisol samples).

3.2. Effects of depressive symptoms and trait rumination on cortisol To examine the effects of depressive symptoms (composite score, DEP), brooding, and reflection on cortisol concentrations over the day, we added these variables as main and interaction effects with group to the multilevel models. The three predictors were individually entered, resulting in three separate models. Each model thus included fixed effects of

time, psychotropic medication, group (0 = controls, 1 = patients), predictor (DEP, brooding, or reflection), and the interaction of group by predictor in addition to a random intercept and random slope for time. Analyses revealed that in all three models, the interactions with group were nonsignificant (group  DEP: B = 0.13, SE = 0.13, p = .319; group  brooding: B = 0.05, SE = 0.03, p = .081; group  rereflection: B = 0.01, SE = 0.02, p = .580). That is, effects of depressive symptoms, brooding, and reflection on momentary cortisol did not differ between remitted depressed patients and controls. In a further set of models, we therefore omitted these non-significant interactions and found that in the total sample, neither DEP (B = 0.09, SE = 0.05, p = .077), nor brooding (B = 0.02, SE = 0.01, p = .078), nor reflection (B = 0.02, SE = 0.01, p = .110) significantly influenced cortisol output over the day.3

3.3. Effects of mood and rumination in daily life on cortisol To examine the effects of daily-life ruminative self-focus, valence, calmness, and energetic-arousal on momentary cortisol, we added aggregated (daily means) and momentary scores (cluster-mean centered) of these variables, both as main and interaction effects with group, together with stressful events to the multilevel models. We performed four separate models, one for each set of daily-life variables (ruminative self-focus, valence, calmness, energetic arousal). Analyses revealed that the interactions with group were nonsignificant for ruminative self-focus (group  mean daily ruminative self-focus B = 0.07, SE = 0.06, p = .226; group  momentary ruminative self-focus B = 0.05, SE = 0.04, p = .179), valence (group  mean daily valence B = 0.09,

3

Given the documented sex differences in depression and rumination, we included depressive symptoms, brooding, and reflection also as interacting with sex in a further set of analyses. These analyses revealed that sex did not moderate the influence of these predictors on cortisol output (all ps > .05).

2264 SE = 0.06, p = .155; group  momentary valence B = 0.01, SE = 0.04, p = .885), and energetic arousal (group  mean daily energetic arousal B = 0.04, SE = 0.07, p = .536; group  momentary energetic arousal B = 0.03, SE = 0.04, p = .343). This indicates that the two groups did not significantly differ with respect to the cortisol effects of ruminative selffocus, valence, and energetic arousal. Therefore, we excluded the non-significant interactions with group and found that higher daily means of ruminative self-focus (B = 0.11, SE = 0.03, p < .001) and lower daily means of valence (B = 0.12, SE = 0.03, p < .001) and of energetic arousal (B = 0.10, SE = 0.03, p = .004) were significantly associated with higher cortisol in the total sample.4 In contrast, we identified a significant interaction effect of group  mean daily calmness predicting cortisol over the day (B = 0.17, SE = 0.06, p = .012), while group  momentary calmness was non-significant (B = 0.02, SE = 0.04, p = .612). Consequently, the effect of daily calmness on cortisol output significantly differed between patients and controls. Finally, we also analyzed the effects of daily-life ruminative self-focus, valence, calmness, and energetic arousal in separate models for each group (see Table 3). Although we did not identify statistically significant interactions of group status with daily life ruminative self-focus, valence, and energetic arousal, we analyzed the effects of these variables comparably to those of daily calmness. Previous evidence suggests that patients and controls might differ in the closeness of the associations between subjective experiences and cortisol (Peeters et al., 2003), and it is possible that due to potentially limited power of our models to detect cross-level interactions (i.e. interactions of level-1 predictors and group), we might not have been able to identify quantitative group differences in cortisol responsiveness toward subjective experiences. As shown in Table 3, separate models for patients and controls revealed that in both samples, higher daily means of ruminative self-focus were significantly associated with higher cortisol. In contrast, lower daily calmness, but also lower valence and lower energetic arousal significantly predicted higher daily cortisol output only in healthy controls but not in remitted depressed patients (Table 3).

4. Discussion The present ambulatory assessment (AA) study found that remitted depressed patients with at least two previous depressive episodes or with a chronic precourse displayed higher trait and momentary rumination, lower momentary mood and lower cortisol levels over the day than healthy controls. By examining factors that might influence HPAA activity in daily life, the present study revealed that depressive symptoms and trait rumination did not predict cortisol levels. However, higher daily means of ruminative self-focus as well as lower daily means of valence and energetic arousal were associated with higher cortisol in the total sample, while the effects of mean daily calmness significantly

4 We also tested whether sex  mean daily rumination and sex  momentary rumination significantly predicted cortisol. The analysis revealed that sex did not moderate the effect of daily-life ruminative self-focus on cortisol ( ps > .05).

S. Huffziger et al. differed between the two groups. Separate analyses revealed that higher mean daily ruminative self-focus predicted higher cortisol in both samples. Furthermore, lower daily means of calmness, but also of valence and energetic arousal, were linked to higher daily cortisol output only in healthy controls but not in the patient sample. Our study found that compared to healthy controls, remitted depressed patients showed lower cortisol levels over the day. While previous studies on basal HPAA activity in remitted depression have revealed both elevated (Lok et al., 2012) and lower cortisol concentrations (Ahrens et al., 2008), accumulating recent evidence indicates that reduced cortisol output might specifically occur when remitted depression is associated with a more severe and longstanding illness course (Bockting et al., 2012; Gex-Fabry et al., 2012; Lok et al., 2012). Relatedly, our clinical sample included remitted patients with at least two previous major depressive episodes or a chronic episode of at least two years, and over 70% of the present sample had been in inpatient treatment for depression. Moreover, reduced basal cortisol activity has also been found in other stressrelated disorders such as posttraumatic stress disorder (PTSD), fibromyalgia, or chronic fatigue syndrome (Fries et al., 2005). According to a developmental model, which may also apply to remitted depressed patients with a longstanding illness history, hypocortisolism is likely to occur after prolonged stress periods associated with enhanced cortisol secretion (Fries et al., 2005; Ahrens et al., 2008). In this context, future studies with larger samples should thoroughly address specific conditions in which remitted depression might be associated with a hypocortisolemic state, by taking possible moderators into account such as illness history, the predominance of a specific depression subtype, or genetic risk. Examining the effects of daily life mood on cortisol levels in natural contexts, we first analyzed whether group status moderated the effects of the three mood facets addressed. We found that the effects of mean daily calmness differed between remitted depressed patients and healthy controls, while valence and energetic arousal did not exhibit significant different effects between the two groups. However, since our moderate sample size might have impeded the possibility to detect significant cross-level interactions, we performed separate analyses per group with respect to all three mood facets. Indeed, these analyses revealed that lower daily means of valence, calmness and energetic arousal were significantly linked to higher cortisol output only in healthy controls but not in remitted depressed individuals. Thus, although our patient sample seemed to exhibit more intraindividual variability in cortisol levels than healthy controls (as indicated by the lower ICC), daily mood variations could not explain substantial parts of the within-person variability in cortisol secretion in this group. Consequently, our results suggest that emotional experiences exhibit regulatory influence on natural cortisol levels in healthy individuals, while remitted depressed patients’ HPAA activity might be less responsive to subtle emotional experiences in daily life. This potentially reduced responsiveness might develop over the course of recurrent depression. Relatedly, there is evidence from previous AA studies that acutely depressed individuals (Peeters et al., 2003) and recurrent bipolar patients with many episodes (Havermans et al., 2011)

Effects on cortisol levels in daily life displayed reduced cortisol responses to natural stressors. Taken together, we therefore think that reduced responsiveness of psychoendocrinological activity in daily life might represent a pathophysiologic marker of recurrent depression, but also of other types of ill mental health conditions, that should be examined in more detail in future studies. Our observation that lower mood was associated with higher cortisol levels in daily life, particularly in healthy individuals, is consistent with previous AA studies on the topic (e.g. Jacobs et al., 2007; Matias et al., 2011). Importantly, however, we also extended previous findings. First, by including different time-related components of mood (daily and momentary mood levels; see Doane and Adam, 2010) we were able to examine the specific time-related patterns of the mood—cortisol association, since respective associations could occur over the whole day and/or within the moment. We found that lower daily means of all three mood facets were associated with higher cortisol output over the day, while momentary mood scores were not. This indicates that slight mood changes from one moment to the next were not able to provoke discrete measurable cortisol reactions, whereas the extent of low mood over a whole day appeared to exert an effect on HPAA activity. As a further extension of previous research on the topic, which mostly focused on broadly assessed negative and positive affect (Giesbrecht et al., 2012; Jacobs et al., 2007; Peeters et al., 2003; for an exception, see Schlotz et al., 2006), our study examined individual mood facets of positive valence, calmness — the opposite pole of tense arousal- and energetic arousal. Herewith, we could show that HPAA activity in daily life was linked to various emotional states. Specifically, increased cortisol occurred in the context of low levels of the mood components positive valence and calmness, which both can be regarded to reflect overall well-being (Wilhelm and Schoebi, 2007), as well as in the context of low energetic arousal, the latter depicting the level of energetic wakefulness. Interestingly, our finding that increased cortisol was predicted by low levels of energetic arousal but high levels of tense arousal (i.e. low calmness) implicates that it makes sense to differentiate between these two states of emotion-associated arousal when examining HPAA activity. Taken together, our findings support the assumption of a covariation between naturally occurring affect and basal cortisol activity during everyday life and confirm evidence indicating a functional coupling of the HPAA with emotion circuits in the brain (cf. Pruessner et al., 2010). Importantly, this coupling is not sufficiently attributable to the experience of stressful events or hassles during the day since we controlled for the influence of daily stressors in our models. To our knowledge, the present study is also the first to address the possible influence of naturally occurring ruminative self-focus on cortisol activity over the day in natural settings. We found that in both samples, higher daily levels of rumination were linked to increased HPAA activity over the day. This is in accordance with the perseverative cognition hypothesis suggesting that rumination entails prolonged activation of bodily stress systems which may ultimately lead to disease (Brosschot et al., 2005). Previously, consistent positive associations between (state) rumination and HPAA activity have predominantly been described during stress tasks in the lab (see review by Zoccola and Dickerson, 2012).

2265 However, these studies could not disentangle the effect of the cognitive ruminative process from the stressor itself (Zoccola and Dickerson, 2012). Moreover, the ecological significance of the rumination—cortisol association remained unclear. Therefore, the present findings add important evidence to the perseverative cognition model by showing that subtle ruminative processes in natural settings predicted higher cortisol levels in daily life. Importantly, this effect occurred in both healthy individuals and remitted depressed patients. Moreover, by again controlling for stressful events in our analyses we could show that the effect of ruminative self-focus on cortisol output was independent of daily life stressors. The present findings also revealed that in contrast to the AA-based indices of subjective experiencing in daily life, neither retrospectively assessed depressive symptoms nor trait rumination was able to predict cortisol levels in everyday life in our samples. Consistent with the view of Conner and Barrett (2012) we conclude that momentary variables, representing immediate experiences, appear to be more reliably and closely linked to HPAA activation compared to retrospectively assessed symptoms and traits. One can speculate that symptom and trait assessments, which depend on memory processes and are therefore more prone to biases, might show less consistent convergence with more objective indicators of functioning such as cortisol output than AA measures of immediate experiences. The present study has some limitations. First, our AA design involved only two days. Therefore, the present findings might be biased if the two sample days were not representative. Second, about a fourth of the patients took antidepressant medication which is known to influence the HPAA (Granger et al., 2009). While we included patients with psychotropic medication to increase generalizability and statistically controlled for the effect of antidepressants, future studies with larger samples should examine whether antidepressants might influence associations between subjective measures and cortisol levels. Moreover, future studies should also address other possible moderators to account for idiographic relations between subjective experiences and cortisol (see also Kuppens et al., 2012). Third, although capturing different mood states provided additional information to previous studies, our three mood states valence, calmness, and energetic arousal were clearly not independent but overlapped (see also Schimmack and Reisenzein, 2002; Wilhelm and Schoebi, 2007). Fourth, despite the fact that the 20 min time-lag between the AA-based subjective measures and the cortisol samples in the present study represents an advantage over studies with pure cross-sectional assessments, we cannot infer clear causal effects of subjective variables on cortisol levels. To address such causal effects, studies with adequate experimental designs are needed (e.g. Kuehner et al., 2009). To sum it up, our results revealed that both in healthy and remitted depressed individuals, naturally occurring rumination was linked to higher cortisol levels over the day, thereby transferring findings supporting the perseverative cognition hypothesis (Brosschot et al., 2005) into the daily life context. Moreover, lower mood in daily life predicted higher cortisol levels in healthy individuals, while in remitted depressed individuals with recurrent episodes, daily cortisol levels were

2266 generally attenuated and exhibited only a nonsignificant responsiveness toward mood in daily life. Future studies could examine whether this potentially reduced mood—cortisol coupling in remitted recurrent depression represents a reliable phenotype of daily life HPAA regulation with prognostic value for the further course of illness.

Role of the funding source This study was supported by grants from the German Research Foundation to CK and PK (DFG KU1464/4-1, DFG KI576/12-1). The study sponsor had no further role in the study design, in the collection, analysis, and interpretation of data, in writing the report, and in the decision to submit the paper for publication.

Conflict of interest statement All authors declare that they have no conflict of interest.

Acknowledgement The authors gratefully acknowledge the persons who participated in this study.

References Adam, E.K., 2006. Transactions among adolescent trait and state emotion and diurnal and momentary cortisol activity in naturalistic settings. Psychoneuroendocrinology 31, 664—679. Ahrens, T., Deuschle, M., Krumm, B., von der Pompe, G., den Boer, J.A., Lederbogen, F., 2008. Pituitary—adrenal and sympathetic nervous system response to stress in women remitted from recurrent major depression. Psychosom. Med. 70 (4) 461— 467. Bockting, C.L.H., Lok, A., Visser, I., Assies, J., Koeter, M.W., Schene, A.H., 2012. Lower cortisol levels predict recurrence in remitted patients with recurrent depression: a 5.5 year prospective study. Psychiatry Res. 200 (2—3) 281—287. Brosschot, J.F., Pieper, S., Thayer, J.F., 2005. Expanding stress theory: prolonged activation and perseverative cognition. Psychoneuroendocrinology 30 (10) 1043—1049. Collip, D., Nicolson, N.A., Lardinois, M., Lataster, T., van Os, J., MyinGermeys, I., 2011. Daily cortisol, stress reactivity and psychotic experiences in individuals at above average genetic risk for psychosis. Psychol. Med. 41 (11) 2305—2315. Conner, T.S., Barrett, L.F., 2012. Trends in ambulatory self-report: the role of momentary experience in psychosomatic medicine. Psychosom. Med. 74 (4) 327—337. Doane, L.D., Adam, E.K., 2010. Loneliness and cortisol: momentary, day-to-day, and trait associations. Psychoneuroendocrinology 35 (3) 430—441. Enders, C.K., Tofighi, D., 2007. Centering predictor variables in crosssectional multilevel models: a new look at an old issue. Psychol. Methods 12 (2) 121—138. Fries, E., Hesse, J., Hellhammer, J., Hellhammer, D.H., 2005. A new view on hypocortisolism. Psychoneuroendocrinology 30 (10) 1010—1016. Gex-Fabry, M., Jermann, F., Kosel, M., Rossier, M.F., Van der Linden, M., Bertschy, G., Bondolfi, G., Aubry, J.-M., 2012. Salivary cortisol profiles in patients remitted from recurrent depression: one-year follow-up of a mindfulness-based cognitive therapy trial. J. Psychiatr. Res. 46, 80—86.

S. Huffziger et al. Giesbrecht, G.F., Campbell, T., Letourneau, N., Kooistra, L., Kaplan, B., 2012. Psychological distress and salivary cortisol covary within persons during pregnancy. Psychoneuroendocrinology 37 (2) 270— 279. Granger, D.A., Hibel, L.C., Fortunato, C.K., Kapelewski, C.H., 2009. Medication effects on salivary cortisol: tactics and strategy to minimize impact in behavioral and developmental science. Psychoneuroendocrinology 34, 1437—1448. Hanson, E.S., Maas, C.M., Meijman, T.F., Godaert, G.L.R., 2000. Cortisol secretion throughout the day, perceptions of the work environment, and negative affect. Ann. Behav. Med. 22 (4) 316— 324. Hautzinger, M., Keller, F., Kuehner, C., Beck, A.T., 2006. Das Beck Depressionsinventar II. Deutsche Bearbeitung und Handbuch zum BDI II. Frankfurt a.M., Harcout Test Services. Havermans, R., Nicolson, N.A., Berkhof, J., de Vries, M.W., 2011. Patterns of salivary cortisol secretion and responses to daily events in patients with remitted bipolar disorder. Psychoneuroendocrinology 36 (2) 258—265. Huffziger, S., Ku ¨hner, C., 2012. Die Ruminationsfacetten Brooding und Reflection. Z. Klin. Psychol. Psychother. 41 (1) 38—46. Huffziger, S., Reinhard, I., Kuehner, C., 2009. A longitudinal study of rumination and distraction in formerly depressed inpatients and community controls. J. Abnorm. Psychol. 118 (4) 746—756. Huffziger, S., Ebner-Priemer, U., Koudela, S., Reinhard, I., Kuehner, C., 2012. Induced rumination in everyday life: advancing research approaches to study rumination. Pers. Individ. Dif. 53, 790—795. Jacobs, N., Myin-Germeys, I., Derom, C., Delespaul, P., van Os, J., Nicolson, N.A., 2007. A momentary assessment study of the relationship between affective and adrenocortical stress responses in daily life. Biol. Psychol. 74 (1) 60—66. Kudielka, B.M., Gierens, A., Hellhammer, D.H., Wu ¨st, S., Schlotz, W., 2012. Salivary cortisol in ambulatory assessment–—some dos, some don’ts, and some open questions. Psychosom. Med. 74 (4) 418—431. Kudielka, B.M., Wu ¨st, S., 2010. Human models in acute and chronic stress: assessing determinants of individual hypothalamus—pituitary—adrenal axis activity and reactivity. Stress 13 (1) 1—14. Kuehner, C., Holzhauer, S., Huffziger, S., 2007. Decreased cortisol response to awakening is associated with cognitive vulnerability to depression in a nonclinical sample of young adults. Psychoneuroendocrinology 32 (2) 199—209. Kuehner, C., Huffziger, S., Liebsch, K., 2009. Rumination, distraction and mindful self-focus: effects on mood, dysfunctional attitudes and cortisol stress response. Psychol. Med. 39 (02) 219—228. Kuppens, P., Tuerlinckx, F., Russell, J.A., Barrett, L.F., 2012. The relation between valence and arousal in subjective experience. Psychol. Bull., http://dx.doi.org/10.1037/a0030811. Lamers, F., Vogelzangs, N., Merikangas, K.R., de Jonge, P., Beekman, A.T.F., Penninx, B.W.J.H., 2012. Evidence for a differential role of HPA-axis function, inflammation and metabolic syndrome in melancholic versus atypical depression. Mol. Psychiatry, http:// dx.doi.org/10.1038/mp.2012.144. Lok, A., Mocking, R.J.T., Ruhe ´, H.G., Visser, I., Koeter, M.W., Assies, J., Bockting, C.L.H., Olff, M., Schene, A.H., 2012. Longitudinal hypothalamic—pituitary—adrenal axis trait and state effects in recurrent depression. Psychoneuroendocrinology 37, 392—902. Matias, G.P., Nicolson, N.A., Freire, T., 2011. Solitude and cortisol: associations with state and trait affect in daily life. Biol. Psychol. 86 (3) 314—319. Miller, G.E., Chen, E., Zhou, E.S., 2007. If it goes up, must it come down? Chronic stress and the hypothalamic—pituitary—adrenocortical axis in humans. Psychol. Bull. 133 (1) 25—45. Moberly, N.J., Watkins, E.R., 2008. Ruminative self-focus and negative affect: an experience sampling study. J. Abnorm. Psychol. 117 (2) 314—323. Montgomery, S., Asberg, M., 1979. A new depression scale designed to be sensitive to change. Br. J. Psychiatry 134, 382—389.

Effects on cortisol levels in daily life Nezlek, J., 2012. Multilevel modeling analyses of diary-style data. In: Mehl, M.R., Conner, T.S. (Eds.), Handbook of Research Methods for Studying Daily Life. Guilford, New York, pp. 357—383. Nolen-Hoeksema, S., Wisco, B.E., Lyubomirsky, S., 2008. Rethinking rumination. Perspect. Psychol. Sci. 3 (5) 400—424. Pariante, C.M., Lightman, S.L., 2008. The HPA axis in major depression: classical theories and new developments. Trends Neurosci. 31 (9) 464—468. Peeters, F., Nicholson, N.A., Berkhof, J., 2003. Cortisol responses to daily events in major depressive disorder. Psychosom. Med. 65 (5) 836—841. Pruessner, J.C., Dedovic, K., Pruessner, M., Lord, C., Buss, C., Collins, L., Dagher, A., Lupien, S.J., 2010. Stress regulation in the central nervous system: evidence from structural and functional neuroimaging studies in human populations — 2008 Curt Richter Award Winner. Psychoneuroendocrinology 35 (1) 179—191. Raison, C.L., Miller, A.H., 2003. When not enough is too much: the role of insufficient glucocorticoid signaling in the pathophysiology of stress-related disorders. Am. J. Psychiatry 160, 1554— 1565. Rydstedt, L.W., Cropley, M., Devereux, J.J., Michalianou, G., 2009. The effects of gender, long-term need for recovery and trait inhibition-rumination on morning and evening saliva cortisol secretion. Anxiety Stress Coping 22 (4) 465—474. Schimmack, U., Reisenzein, R., 2002. Experiencing activation: energetic arousal and tense arousal are not mixtures of valence and activation. Emotion 2, 412—417. ¨st, S., Schlotz, W., Kumsta, R., Layes, I., Entringer, S., Jones, A., Wu 2008. Covariance between psychological and endocrine responses to pharmacological challenge and psychosocial stress: a question of timing. Psychosom. Med. 70 (7) 787—796. Schlotz, W., Schulz, P., Hellhammer, J., Stone, A.A., Hellhammer, D.H., 2006. Trait anxiety moderates the impact of performance

2267 pressure on salivary cortisol in everyday life. Psychoneuroendocrinology 31 (4) 459—472. Smyth, J., Ockenfels, M.C., Porter, L., Kirschbaum, C., Hellhammer, D.H., Stone, A.A., 1998. Stressors and mood measured on a momentary basis are associated with salivary cortisol secretion. Psychoneuroendocrinology 23 (4) 353—370. Stetler, C., Miller, G.E., 2011. Depression and hypothalamic—pituitary—adrenal activation: a quantitative summary of four decades of research. Psychosom. Med. 73 (2) 114—126. Treynor, W., Gonzalez, R., Nolen-Hoeksema, S., 2003. Rumination reconsidered: a psychometric analysis. Cognitive Ther. Res. 27 (3) 247—259. Trull, T.J., Ebner-Priemer, U.W., 2013. Ambulatory assessment. Annu. Rev. Clin. Psychol. 9, 4.1—4.27. Wilhelm, P., Schoebi, D., 2007. Assessing mood in daily life: structural validity, sensitivity to change, and reliability of a short-scale to measure three basic dimensions of mood. Psychol. Assess. 23 (4) 258—267. Wittchen, H.U., Wunderlich, U., Gruschwitz, S., Zaudig, M., 1997. SCID: Structured Clinical Interview for DSM-IV Axis I Disorders. Hogrefe, Goettingen, Germany. Wolfram, M., Bellingrath, S., Feuerhahn, N., Kudielka, B.M., 2012. Cortisol responses to naturalistic and laboratory stress in student teachers: comparison with a non-stress control day. Stress Health, http://dx.doi.org/10.1002/smi.2439. Young, E.A., Aggen, S.H., Prescott, C.A., Kendler, K.S., 2000. Similarity in saliva cortisol measures in monozygotic twins and the influence of past major depression. Biol. Psychiatry 48 (1) 70—74. Zoccola, P.M., Dickerson, S.S., 2012. Assessing the relationship between rumination and cortisol: a review. Psychosom. Res. 73 (1) 1—9. Zoccola, P.M., Dickerson, S.S., Zaldivar, F.P., 2008. Rumination and cortisol responses to laboratory stressors. Psychosom. Med. 70 (6) 661—667.