Elsevier

Psychoneuroendocrinology

Volume 69, July 2016, Pages 16-25
Psychoneuroendocrinology

The temporal dynamics of cortisol and affective states in depressed and non-depressed individuals

https://doi.org/10.1016/j.psyneuen.2016.03.012Get rights and content

Highlights

  • Cortisol may be a crucial link between daily life experiences and emotions.

  • We assessed cortisol and affect dynamics in depressed and non-depressed individuals.

  • We conducted intra-individual analyses using 90 measurements per individual (n = 30).

  • A relationship between cortisol and affect was found in the majority of the sample.

  • There were large individual differences in the direction, sign and timing of effects.

Abstract

Objectives

Cortisol levels have been related to mood disorders at the group level, but not much is known about how cortisol relates to affective states within individuals over time. We examined the temporal dynamics of cortisol and affective states in depressed and non-depressed individuals in daily life. Specifically, we addressed the direction and timing of the effects, as well as individual differences.

Methods

Thirty depressed and non-depressed participants (aged 20–50 years) filled out questionnaires regarding their affect and sampled saliva three times a day for 30 days in their natural environment. They were pair-matched on age, gender, smoking behavior and body mass index. The multivariate time series (T = 90) of every participant were analyzed using vector autoregressive (VAR) modeling to assess lagged effects of cortisol on affect, and vice versa. Contemporaneous effects were assessed using the residuals of the VAR models. Impulse response function analysis was used to examine the timing of effects.

Results

For 29 out of 30 participants, a VAR model could be constructed. A significant relationship between cortisol and positive or negative affect was found for the majority of the participants, but the direction, sign, and timing of the relationship varied among individuals.

Conclusion

This approach proves to be a valuable addition to traditional group designs, because our results showed that daily life fluctuations in cortisol can influence affective states, and vice versa, but not in all individuals and in varying ways. Future studies may examine whether these individual differences relate to susceptibility for or progression of mood disorders.

Introduction

The hypothalamic-pituitary-adrenal (HPA) axis adapts the body to stress conditions by mobilizing energy and inhibiting non-emergency processes such as sleep, sexual activity, and growth (Chrousos, 2009). This system is vital for survival, and also thought to play an important role in the onset and progression of depression (e.g. Holsboer and Ising, 2010). By some, the HPA axis is even referred to as the final common pathway for most of the symptomatology of depression (e.g. Bao et al., 2008). This proposition is supported by the indirect evidence that stressful life events, which are capable of triggering the HPA axis, often precede depressive episodes (e.g. Kendler et al., 2000).

Cortisol is an end product of the HPA axis. Because of its numerous targets in the body, among which are brain areas involved in emotional reactivity, it is much researched in the context of depression. Many of these studies have been observational, of which the results were assessed in a meta-analysis by Stetler and Miller (2011). They concluded that, overall, there seems to be a tendency towards increased cortisol levels in depressed samples, albeit with considerable heterogeneity between studies. Observational studies with a single or a few measurements are useful to detect general patterns and traits in the population. However, cortisol has a highly dynamic nature (e.g. Booij et al., 2015), and the negative affect experienced by depressed individuals shows daily and context-dependent variation as well (e.g. Myin-Germeys et al., 2009, Wirz-Justice, 2008). Hence, these studies do not reveal information about the dynamic relationship between cortisol and depressive symptoms in daily life (Hamaker, 2012).

Depressive symptoms are characterized by increases in negative affect and decreases in positive affect (Clark and Watson, 1991). Improvement of depressive symptoms might therefore rely on lowering negative affect as well as on increasing positive affect in depressed individuals. Peeters et al. (2003) studied associations between negative affect and cortisol in daily life using an experience sampling protocol (ESM) (Csikszentmihalyi and Larson, 1987). Besides having high ecological validity, ESM has the added benefit that the small time interval between measurements reduces recall bias (de Vries and Csikszentmihalyi, 2006, Telford et al., 2012). In short, Peeters et al. (2003) found that negative affect was related to higher cortisol levels in the healthy but not the depressed group. van Eck et al. (1996) did a similar study in healthy individuals with low and high levels of perceived stress, including measures of both positive and negative affect. Regardless of the perceived stress level, they found a positive association between negative affect and cortisol, and no associations with positive affect. In a larger sample from the general population, however, Smyth et al. (1998) did find a negative association between positive affect and cortisol.

The design and analytical methods that were used in these studies did not allow statistical inference about the direction of the relationships. Experimental evidence showing that exposure to stressful situations increases cortisol levels (Dickerson and Kemeny, 2004, Kirschbaum and Hellhammer, 1989) favors the idea that increases in negative and/or reductions in positive affect precede increases in cortisol. As such, in many ESM studies the results are interpreted in this way (e.g. Adam, 2006, Peeters et al., 2003, van Eck et al., 1996). On the other hand, studies in rats have shown that increased cortisol levels may precede increases in negative affect (Lightman and Conway-Campbell, 2010). Thus, the relationship may also be reversed to what is generally assumed, or reciprocal.

In the above-mentioned ESM and experimental studies, effects were estimated over a limited time frame (e.g. minutes up to several hours). This restrains knowledge about the dynamics of the relationship and about the net effect of cortisol on affect over a longer time frame, and vice versa. Studies into the autocorrelation of negative and positive emotions (emotional inertia) in depressed versus non-depressed individuals suggest that depressed individuals show relatively large autocorrelation in both positive and negative affect, compared to non-depressed individuals (e.g. Koval et al., 2012, Kuppens et al., 2010). This suggests that an initial subtle change in affect to an increase in cortisol may eventually result in a relatively large net change over a longer time period, especially for depressed individuals. With regard to the response pattern of cortisol, evidence suggests that the cortisol response to stress is cyclic (i.e. fluctuating around a certain set point) (e.g. Schubert et al., 2006, Schubert et al., 2003, Schubert et al., 2012). This may relate to cortisol’s crucial role in the maintenance of the body’s homeostasis. To our knowledge, no study has examined the temporal aspects of the relationship between affect and cortisol in detail, that is, including the direction of the effects, potential feedback loops, and the time period over which the effects take place.

Recently, sophisticated regression techniques for the analyses of time-series data have become available in the field of psychiatry. These techniques are suitable for explaining variance within single individuals instead of variance in the population. With sufficient data points (T > 60), time-lagged associations between variables as well as the temporal ordering of the effects can be assessed in detail, at the level of the individual. Moreover, so-called ‘impulse-response functions’ can be used to predict how an increase in one variable can work through the system over time to result in dynamic changes in other variables. This approach can reveal the timing and direction of the effects, as well as possible feedback loops (Brandt and Williams, 2007).

Adopting an intensive time-series approach has another benefit. Because individuals are examined one by one, individual differences that would go unnoticed in group designs can be revealed. Depression can present very differently in different individuals, with regard to both symptoms and severity (van Loo et al., 2012). In addition, various studies suggest that only a subgroup of depressed individuals (e.g. individuals with melancholic depression) has increased cortisol levels (Lamers et al., 2012, Sachar et al., 1970), and that other subgroups (e.g. individuals with severe or chronic depression) may even show the opposite association (Booij et al., 2013, Burke et al., 2005, Oldehinkel et al., 2001). Because of such heterogeneity, associations at the group level may be small and seemingly irrelevant, whereas in certain individuals the association may be large and meaningful. The present study complements the available literature in a unique way by examining the temporal dynamics of cortisol and mood in daily life, and potential individual differences therein. For this purpose, we adopted an intensive time-series approach in which we monitored 30 pair-matched depressed and non-depressed participants three times a day for 30 days in their natural environment.

Because group-level evidence has been found both for an influence of affective states on cortisol and of cortisol on affective states, we did not adopt specific hypotheses about the direction of the relationship within individuals. Rather, the study was designed to explore whether the relationship between cortisol and affect is bidirectional for all individuals, or whether for some individuals the relationship is particularly evident in one direction. Furthermore, we examined whether individual differences in the sign and size of the relationship between cortisol and affect were related to individual differences in depression characteristics. Finally, with regard to the temporal dynamics of the effects, we hypothesized that the response of affect to an increase in cortisol would show a steady increase or decrease over time, particularly in depressed individuals, resulting in a large net change over several days. Regarding an increase in affect on cortisol, we hypothesized that it would result in cyclic response pattern of cortisol, resulting in a relatively small net effect over several days.

Section snippets

Participants

The sample was part of the ‘Mood and movement in daily life’ (MOOVD) study, which was set up to investigate the dynamic relationship between physical activity and mood in daily life, and the role of several biomarkers therein. Participants (age 20–50 years) were intensively monitored in their natural environments for 30 days, by means of electronic diaries, actigraphy, and saliva sampling, resulting in a total of 90 measurements per individual. Of the 62 participants who started the study, 4

Descriptive statistics

Sample characteristics are presented in Table 1. Two-thirds of the depressed sample had prior depressive episodes. One non-depressed person had a prior episode (>2 years ago). Of the depressed individuals, 11 had a current episode, and three individuals were in remission since 2–4 weeks. Mean cortisol levels over the entire study period were not significantly higher in the depressed than in the non-depressed group (mixed-model analysis, B = 0.15, 95%CI = −0.11 to 0.41). Positive affect was lower

Discussion

We adopted an intensive time-series approach to examine the temporal relationship between cortisol and affective states in depressed and non-depressed individuals, and to explore the timing of the effects. For the majority of the participants, a significant relationship was found between cortisol and positive or negative affect, but the direction, sign, and timing of the relationship varied greatly across individuals.

A significant temporal relationship between cortisol and affect was found in

Conclusions

To conclude, we adopted an intensive time-series approach to examine the temporal relationship between cortisol and affective states in daily life. This approach proves to be a valuable addition to traditional group designs, because our results showed that daily life fluctuations in cortisol can influence affective states, and vice versa, but not in all individuals and in varying ways.

Role of funding source

PdJ and EB are supported by a NWO/ZonMW VICI-Grant (no: 91812607). The funder was not involved in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.

Conflicts of interest

All authors declare that they have no conflicts of interest.

Author contributions

SB, EB, PdJ and AO designed the study. SB and EB collected the data. SB analyzed the data and wrote the first draft of the manuscript. EB, PdJ and AO critically revised the manuscript. All authors approved final content.

Acknowledgements

We acknowledge Mara Bouwmans for her important contributions to the design and completion of the MOOVD study. The authors also thank the participants who volunteered for this study and the students who participated in the data collection. In addition, they thank the therapists of the Center for Integrative Psychiatry and the University Center for Psychiatry, Groningen, The Netherlands, for facilitating this research. Finally, they acknowledge the iLab and the Research School of Behavioural and

References (56)

  • J. Smyth et al.

    Stressors and mood measured on a momentary basis are associated with salivary cortisol secretion

    Psychoneuroendocrinology

    (1998)
  • American Psychiatric Association

    Diagnostic and Statistical Manual-Text Revision (DSM-IV-TRim, 2000)

    (2000)
  • A. Beck et al.

    Manual for the BDI-II San Antonio

    (1996)
  • S.H. Booij et al.

    Cortisol and α-amylase secretion patterns between and within depressed and non-depressed individuals

    PLoS One

    (2015)
  • P.T. Brandt et al.

    Multiple Time Series Models

    (2007)
  • L.M. Bylsma et al.

    Emotional reactivity to daily events in major and minor depression

    J. Abnorm. Psychol.

    (2011)
  • G.P. Chrousos

    Stress and disorders of the stress system

    Nat. Rev. Endocrinol.

    (2009)
  • L.A. Clark et al.

    Tripartite model of anxiety and depression: psychometric evidence and taxonomic implications

    J. Abnorm. Psychol.

    (1991)
  • M. Csikszentmihalyi et al.

    Validity and reliability of the experience-sampling method

    J. Nerv. Ment. Dis.

    (1987)
  • S.S. Dickerson et al.

    Acute stressors and cortisol responses: a theoretical integration and synthesis of laboratory research

    Psychol. Bull.

    (2004)
  • P. Gilbert

    Depression and stress: a biopsychosocial exploration of evolved functions and mechanisms

    Stress

    (2001)
  • C.W. Granger

    Investigating causal relations by econometric models and cross-spectral methods

    Econometrica

    (1969)
  • E.L. Hamaker et al.

    Modeling affect dynamics: state-of-the-art and future challenges

    Emot. Rev.

    (2015)
  • E. Hamaker

    Why researchers should think within-person: a paradigmatic rationale

  • M.J. Henckens et al.

    Time-dependent effects of corticosteroids on human amygdala processing

    J. Neurosci.

    (2010)
  • G. Henriques

    Depression: disease or behavioral shutdown mechanism

    J. Sci. Health Policy

    (2000)
  • F. Holsboer et al.

    Stress hormone regulation: biological role and translation into therapy

    Annu. Rev. Psychol.

    (2010)
  • J. Honaker et al.

    Amelia II: a program for missing data

    J. Stat. Softw.

    (2011)
  • Cited by (18)

    • Momentary emotions and salivary cortisol: A systematic review and meta-analysis of ecological momentary assessment studies

      2021, Neuroscience and Biobehavioral Reviews
      Citation Excerpt :

      For example, Slatcher et al. (2010) found that neither concurrent nor prior negative emotions was associated with momentary cortisol levels. On the other hand, Booij et al. (2016) found indications of both concurrent and lagged associations, but only in some individuals. Hoyt et al. (2016b), on the other hand, found that sadness did not predict concurrent cortisol levels but trended towards predicting cortisol at the following time point.

    • Ambulatory measurement of cortisol: Where do we stand, and which way to follow?

      2019, Sensing and Bio-Sensing Research
      Citation Excerpt :

      Interestingly, novel analytical methods are being developed that are based on the timing of maximum corticosteroid reactivity, i.e. corticosteroid-peak locked as opposed to time or event locked (e.g. [38]). In addition to group-based approaches, high frequency-sampling techniques based on (multiple) N = 1 approaches provide ways to interpret the collected data on an individual level, focusing on intra-individual variance over time instead of inter-individual variance [8,47,48,57]. Cortisol assessment techniques allowing quick and repeatable assessment are likely to stimulate this N = 1 approach.

    • Temporal dynamics of the HPA axis linked to exploratory behavior in a wild European songbird (Parus major)

      2017, General and Comparative Endocrinology
      Citation Excerpt :

      The idea that fine-scale temporal dynamics in the HPA axis underlie individual variation in behavior has recently been demonstrated in humans (Booij et al., 2016) and rodents (Sarabdjitsingh et al., 2010) and we believe this line of research would benefit from experimental approaches using free-living animal models for personality.

    View all citing articles on Scopus
    View full text