G Model
BIOPSY-6049; No of Pages 14
Biological Psychology xxx (2008) xxx–xxx
Contents lists available at ScienceDirect
Biological Psychology
journal homepage: www.elsevier.com/locate/biopsycho
Review
Cortisol awakening response and psychosocial factors:
A systematic review and meta-analysis
Yoichi Chida *, Andrew Steptoe
Psychobiology Group, Department of Epidemiology and Public Health, University College London, United Kingdom
ARTICLE INFO ABSTRACT
Article history: The magnitude of the cortisol awakening response, a relatively new indicator of hypothalamic–pituitary–
Received 14 May 2008 adrenocortical (HPA) axis activation, has been related to a number of psychosocial factors. But findings
Accepted 21 October 2008 have been inconsistent across studies. We systematically reviewed previous studies investigating the
Available online xxx association between the cortisol awakening response and psychosocial factors. 147 eligible studies from
62 articles were identified. Separate analyses were carried out on the increase in cortisol following
Keywords: waking (CARi), and the integrated volume of cortisol released over the waking period (CARauc). We found
Depression and anxiety that the CARi was positively associated with job stress and general life stress. It was negatively associated
HPA axis with fatigue, burnout, or exhaustion. There were less reliable negative associations between the CARi and
Job stress positive affects. The CARauc was positively related to general life stress and negatively related to
Positive well-being posttraumatic stress syndrome. This review concludes that different psychosocial factors are associated
Posttraumatic stress disorder with an enhanced or reduced cortisol awakening response.
Burnout and fatigue
ß 2008 Elsevier B.V. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
2.1. Data sources and searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
2.2. Study selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
2.3. Data extraction and quality assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
2.4. Data synthesis and analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
3.1. Summary of studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
3.2. Study characteristics and quality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
3.3. Meta-analysis of cortisol changes after waking (CARi). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
3.4. Meta-analysis of overall cortisol output (CARauc) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
4.1. Differences in CAR across psychosocial factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
4.2. Underlying mechanisms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
4.3. CARi and CARauc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
4.4. Issues from sensitivity analyses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
4.5. Limitations and guidelines for future studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
4.6. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 000
* Corresponding author at: Psychobiology Group, Department of Epidemiology and Public Health, University College London, 1-19 Torrington Place, London WC1E 6BT,
United Kingdom. Tel.: +44 20 7679 8265; fax: +44 20 7916 8542.
E-mail address: [email protected] (Y. Chida).
0301-0511/$ – see front matter ß 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.biopsycho.2008.10.004
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
2 Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx
1. Introduction questions. First, what psychosocial factors are associated with
larger and smaller CARs? Second, do associations with psychoso-
Cortisol levels are low in the night but rise in the early hours cial factors differ with study design factors such as control for
before waking. After waking up, most people show a further rise, possible confounders, the number of days or time-points of
the concentration peaking 20–45 min later. The cortisol awakening measurement, participant characteristics (age, gender, and health
response (CAR), first established by Pruessner et al. (1997) as a status), and types of CAR outcomes (CARi or CARauc)? In order to
useful index, can be defined as the change in cortisol concentration evaluate the consistency of CAR results in relation to psychosocial
that occurs during the first hour after waking from sleep (Clow factors, we divided the constructs tested into eight categories: (1)
et al., 2004; Steptoe, 2007). This endocrine marker has been job stress, including high demands/low control, effort/reward
analyzed using a number of different approaches, including imbalance, and overcommitment; (2) general (non-work) life
repeated measures analysis of samples taken at several time stress, measured with indicators such as perceived stress, daily
points, simple change scores between waking and 20–45 min later, hassles and poor marital quality; (3) depression, primarily
and area under the curve (AUC) indices (Clow et al., 2004; depressed mood within the normal range, and hopelessness; (4)
Pruessner et al., 2003; Fekedulegn et al., 2007). Broadly speaking, anxiety, neuroticism, and negative affect; (5) fatigue, burnout, and
two basic parameters have been assessed: the overall volume of exhaustion; (6) posttraumatic stress syndrome; (7) positive
cortisol released over the waking period, often measured by psychological states or traits. Some of these categories are closely
integrating the area under the curve imputed from repeated related; however, because of the interest in this literature in
samples (CARauc), and the change (typically increase) of cortisol relationships with several distinct sets of constructs, we adopted
from the level recorded on waking (CARi) (Pruessner et al., 2003). an exploratory approach in which multiple categories were
considered.
The CAR has been attracting growing attention from research-
ers for a number of reasons. First, the introduction of salivary 2. Methods
cortisol assessments has allowed response patterns to be
measured from salivas collected by research participants at home, 2.1. Data sources and searches
without the need for repeated blood sampling. Second, the CAR is
an easy parameter of HPA axis function to measure since it does not We developed a protocol using a widely recommended method for systematic
require laboratory conditions or administration of exogenous reviews (Stroup et al., 2000). Electronic searches were done in Medline (1966-
agents; rather awakening itself is a consistent, recurring, and September 2008); PsycINFO (1872-September 2008); Web of Science (1900-
strong stimulus for HPA activity (Wilhelm et al., 2007). Third, the September 2008); and PubMed (1950-September 2008). The main search strategy
CAR is under somewhat independent control from cortisol output was (‘‘cortisol awakening response’’) OR (‘‘awakening cortisol response’’). In addition,
during the remainder of the day, and associations between the CAR we manually searched the references lists of retrieved publications and reviews.
and cortisol sampled later in the day are low (Schmidt-Reinwald
et al., 1999; Edwards et al., 2001a). Fourth, twin studies have 2.2. Study selection
documented a genetic influence on the CAR that is distinct from the
heritability of daytime cortisol levels (Wu¨ st et al., 2000a; Bartels Criteria for inclusion were as follows: (1) English language full-length
et al., 2003; Kupper et al., 2005). Finally, the magnitude of the CAR publication in a peer-reviewed journal; (2) the article investigated associations
appears to be associated with psychosocial factors and health in between psychosocial factors and the CAR; (3) If more than one kind of psychosocial
potentially significant ways, suggesting that it may be a distinctive factor was assessed in one article, the associations with cortisol were included
indicator of HPA function and dysfunction. separately as separate studies; (4) we included studies evaluating certain mental
health problems such as depressive disorders, anxiety disorders, posttraumatic
Early psychological studies of the CAR indicated that the stress disorder, and chronic fatigue syndrome, since these conditions are thought to
response was heightened among individuals experiencing job be related to HPA axis dysregulation (McEwen, 2007). However, (5) we excluded
stress, overload, and low self-esteem (Schulz et al., 1998; Pruessner studies of psychotic illnesses and bipolar disorders, because disturbances of cortisol
et al., 1999; Wu¨ st et al., 2000a). It appeared that the CAR might be regulation may be secondary in these conditions; (6) studies using low socio-
an indicator of chronic psychosocial stress, and that larger CARs economic status (SES) as a psychosocial indicator were excluded, because many
were a marker of neuroendocrine activation as the individual studies included SES as a covariate, and low SES alone has many consequences apart
contemplated the challenges of the day ahead. However, negative from greater stress experience; (7) if a cohort overlapped across studies, the paper
associations between the CAR and other stress-related constructs with smaller sample size or poorer study quality was excluded.
emerged (de Vente et al., 2003; Roberts et al., 2004), suggesting
that a more nuanced interpretation was required. Steptoe (2007) 2.3. Data extraction and quality assessment
argued that enhanced CARs were generally related to work stress
(e.g., job strain, work overload, and overcommitment to work), A manual was prepared for coding the studies. The manual was revised during the
other types of life stress (e.g., financial strain, loneliness, poor coding to incorporate important aspects of the located studies. The final list of
marital quality), and depression and depressive symptoms. By variables included the following: first author, publication year; sample size with
contrast, a diminished CAR appeared to be present in people participant characteristics; psychosocial predictors (measurement method); study
suffering from posttraumatic stress syndrome and fatigue-related quality score (0–8); the number of days assessed and time points (min after
symptoms, and was also related to positive psychological affect wakening); CAR indicator; brief results and effect size (correlation coefficient, r) for
(e.g., happiness, well-being, optimism). Unfortunately, increasing CARi and CARauc. We extracted the following indicators of cortisol changes following
numbers of studies, particularly of depression and depressed waking or CARi: the area of cortisol increase under the curve (AUCi), the mean value of
mood, have shown inconsistent findings (Pruessner et al., 2003; cortisol values post-awakening minus wakening value (MINC), the absolute
Bhagwagar et al., 2005; Stetler and Miller, 2005). Several distinct difference score or increase of cortisol (AINC), or absolute cortisol values (ACOR)
study protocols and approaches to analysis have been employed, assessed by repeated measures analysis. The overall volume of cortisol released over
and a wide range of psychosocial variables has been investigated. the waking period (CARauc) was indexed by the total area under the curve (AUCt).
Given the variability of the findings, we believe this literature
would benefit from a quantitative meta-analytic approach. We assessed the quality of all manuscripts using a system for scoring
methodological quality, based on whether or not potential confounders were
The purpose of this systematic review is therefore to synthesize addressed, since these can contribute to biases associated with effect estimation.
existing data using meta-analytic techniques to address two main Based on existing evidence (Wu¨ st et al., 2000b; Clow et al., 2004; Hellhammer et al.,
2007), we considered a study to be of good quality if it considered or controlled age,
gender, smoking, participant adherence to salivary sampling (e.g., electronic
monitoring, self-reported time of sampling), medication with steroids, clear
instructions to participants about the sampling procedure (e.g., the subjects were
instructed not to brush their teeth, eat, or drink for at least 15 min before sampling),
waking time, and type of sampling day (weekday or weekend). We classified studies
arbitrarily into high or low quality categories by whether or not they fulfilled six or
more of these eight criteria.
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx 3
Study inclusion and data extractions were conducted by one author (Y.C.) and studies. Separate meta-analyses were carried out for each of the seven categories of
verified by the other (A.S.). Assessments of quality and validity were made psychosocial factor for the CARi and CARauc. If more than one psychosocial factor
independently, and 18 disputes were settled by consensus. was analyzed in a single article, they were included as separate studies. Provided
there was sufficient information, we aimed to perform sensitivity analyses
2.4. Data synthesis and analysis according to the characteristics of study population (gender and health status),
study quality or design, number of days of CAR assessment (!2 days), and studies
We followed meta-analytic procedures that have been described elsewhere that assessed cortisol !3 times over the waking period, so that more robust
(Chida et al., 2008a,b; Chida and Steptoe, 2008; Chida and Hamer, 2008). Briefly, an measures of the CAR were tested. We simultaneously employed the Q-test for
effect size was calculated from the difference in CAR levels between the control and homogeneity between studies, which tests whether the between-study variability
exposed groups. This was then transformed into r, which gives more weight to in effect sizes exceeds that expected from corresponding within study variability.
larger studies that tend to obtain more reliable estimates of the population effect Finally, to detect publication biases, we measured the degree of asymmetry by
size. If the raw data were not presented then an F-ratio for the main effect over time using Egger’s unweighted regression asymmetry test. This is a statistical method of
was used for conversion to r instead. If there were insufficient data reported to detecting bias by assessing the shape of the funnel plot of effect size against sample
calculate an r-value, we initially contacted the relevant authors to obtain more size (Egger et al., 1997). It is based on the notion that precision in estimation of
information. If no relevant convertible statistics were presented, other than a p- effect sizes increases with sample size. Results from smaller studies will scatter at
value, we calculated the t-statistic from the p-value and an r-sub (equivalent) the foot of the funnel, with the spread narrowing in larger studies. Funnel plot
(Rosenthal and Rubin, 2003). When a paper reported p < 0.05, <0.01, or ns, we asymmetry is thought to indicate publication bias since smaller negative studies are
computed r-sub (equivalent) with p-values of 0.025, 0.050, 0.50 (one-tailed), less likely to be published than smaller positive studies. All analyses were
respectively, which likely yielded a highly conservative estimate of the effect size. performed using the Meta-Analysis Program (Masui, 2003).
An inferential statistic or effect size was used to calculate a z-score for each study,
which was weighted by sample size to calculate an overall z-score and probability. 3. Results
Our meta-analytic procedure allowed more than one effect size per psychosocial 3.1. Summary of studies
category per study to be included. This method has been used in previous meta-
analyses (e.g., Myrtek, 1995; Pole, 2007). However, this violates conservative Fig. 1 shows details of the flow diagram for this systematic
principles of meta-analysis, namely that effects should be independent of one review. Table 1 and supplementary table detail the articles that
another. We therefore supplemented the analyses by computing aggregate effect were included (n = 62) and excluded (n = 81), respectively. Several
sizes for each psychological category by calculating the unweighted mean of all studies were excluded because although they measured cortisol
effect sizes in the same category in any particular study (e.g., mean effect size of all early in the day, they did not measure the CAR as defined here.
CARi or CARauc measures related to different job stress indicators). We performed
random effects rather than fixed effects modeling (Dersimonian and Laird, 1986) in
our analyses, since this takes into account the amount of variance caused by
differences between studies as well as differences among participants within
Fig. 1. Flow diagram of systematic review (QUOROM statement flow diagram).
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
4 Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx
Table 1
Summary of cortisol awakening response and psychosocial factors.
No. Authors (year) Sample Predictors (measurement) CAR confounder score (0–8) Assessed days: CARi CARauc
time points (min)
Effect (r) Effect (r)
size size
1 Bhattacharyya et al. (2008) 72mf Depression3 (BDI) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/15/30 AUCi 0.033 AUCt À0.058
Early life adversity2 (CTQ) Md = 1, In = 1, At = 1, W = 1) 2D: 0/30 ACOR 0.238 AUCt À0.260
(cd) Abuse chronicity2 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 1D: 0/30/45/60 AUCi À0.210 AUCt
Md = 0, In = 1, At = 1, W = 0) 1D: 0/30/45/60 AUCi 0.280 AUCt 0.080
2 Gonzalez et al. (2008) 61f 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 1D: 0/30/45/60 AUCi 0.340 AUCt 0.310
Md = 1, In = 1, At = 0, W = 0) 1D: 0/30 AINC 0.000 AUCt 0.000
(pp) 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 1D: 0/30 AINC 0.293 0.000
Md = 1, In = 1, At = 0, W = 0) 1D: 0/30/60 AUCi 0.000 AUCt
3a Johnson et al. (2008) 52f 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 1D: 0/30/60 AUCi 0.174 AUCt À0.093
Md = 1, In = 1, At = 0, W = 0) 1D: 0/30 AINC À0.027 AUCt À0.493
3b Johnson et al. (2008) 52f Depression3 (BDI) 3 (Ag = 1, Sx = 1, Sm = 0, Ad = 0, 2D: 0/30 or 45 AINC 0.220 AUCt
3c Johnson et al. (2008) 52f PTSD6 (CAPS) Md = 0, In = 0, At = 0, W = 1) 2D: 0/30 or 45 AINC À0.400 0.000
3 (Ag = 1, Sx = 1, Sm = 0, Ad = 0, 2D: 0/30 or 45 AINC 0.210 AUCt 0.131
4a Kallen et al. (2008) 53m Anxiety4 (MASC) Md = 0, In = 0, At = 0, W = 1) 1D: 0/30/45/60 AUCi 0.135
4b Kallen et al. (2008) (yg) 6 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 1D: 0/30 AINC 0.306 AUCt 0.497
5a Nater et al. (2008) 46f Anxiety4 (MASC) Md = 1, In = 1, At = 0, W = 1) 2D: 0/30/60 ACOR 0.000 AUCt
5b Nater et al. (2008) (yg) 6 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 2D: 0/30/60 ACOR 0.505 AUCt 0.060
6 O’Donnell et al. (2008) 45m Chronic fatigue syndrome5 Md = 1, In = 1, At = 0, W = 1) 1D: 0/15/30/45/60 À0.270
(1994 CFS case definition) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30/60 AUCi À0.370 À0.356
140f Chronic fatigue syndrome5 Md = 1, In = 1, At = 1, W = 0) 1D: 0/30
(1994 CFS case definition) 6 (Ag = 0, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30 AUCi À0.107
542mf Depression3 (CES-D) Md = 1, In = 1, At = 0, W = 1) 2D: 0/30/45 AUCi 0.170
6 (Ag = 0, Sx = 1, Sm = 1, Ad = 1, 2D: 0/30/45 AUCi 0.267
7a Quirin et al. (2008) 48f Social stress2 Md = 1, In = 1, At = 0, W = 1) 2D: 0/30/45 AUCi
7b Quirin et al. (2008) (Trier inventory) 6 (Ag = 0, Sx = 1, Sm = 1, Ad = 1, 2D: 0/30/45 AUCi À0.317
7c Quirin et al. (2008) Md = 1, In = 1, At = 0, W = 1) 2D: 0/30/45 AUCi À0.193
48f Attachment anxiety4 (ECRS) 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30/45/60 AINC
Md = 1, In = 1, At = 0, W = 0) 2D: 0/30 AINC 0.000
48f Self-esteem7 (RES) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 0, 2D: 0/30 AUCi À0.350
Md = 1, In = 1, At = 1, W = 1) 2D: 0/15/30/45 AUCi À0.240
8 Tops et al. (2008) 194f Depressive mood3 (POMS) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 2D: 0/15/30/45 AUCi
Md = 1, In = 1, At = 1, W = 1) 2D: 0/15/30/45 AUCi 0.000
9 Wahbeh et al. (2008) 30mf Caregiver stress2 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 2D: 0/15/30 AUCi 0.000
10a Weekes et al. (2008) 31m Examination stress2 Md = 1, In = 1, At = 1, W = 1) 2D: 0/15/30 AUCi 0.000
3 (Ag = 1, Sx = 1, Sm = 0, Ad = 0, 2D: 0/30/60 À0.130
10b Weekes et al. (2008) 35f Examination stress2 Md = 0, In = 0, At = 1, W = 0) À0.370
11 Chan et al. (2007) 65mf Neuroticism4 (EPI) 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 0, À0.200
Md = 1, In = 1, At = 1, W = 0)
12 de Kloet et al. (2007) 47m PTSD6 (CAPS) 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
Md = 1, In = 1, At = 0, W = 0)
13a Greaves-Lord et al. (2007) 376mf Current anxiety4 (RCADS) 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
13b Greaves-Lord et al. (2007) (yg) Persistent anxiety4 (RCADS) Md = 1, In = 1, At = 0, W = 0)
14a Harris et al. (2007) 354mf Job stress1 (ERI model) 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
(yg) Md = 0, In = 0, At = 0, W = 1)
39f 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
Md = 0, In = 0, At = 0, W = 1)
14b Harris et al. (2007) 44f Job stress1 (DC model) 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
Md = 0, In = 0, At = 0, W = 1)
14c Harris et al. (2007) 42f Poor social 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
14d Harris et al. (2007) functioning2 (SF-36) Md = 0, In = 0, At = 0, W = 1)
14e Harris et al. (2007) 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
42f Vitality7 (SF-36) Md = 0, In = 0, At = 0, W = 1)
5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1,
42f Well-being7 (SF-36) Md = 1, In = 1, At = 1, W = 0)
6 (Ag = 1, Sx = 1, Sm = 0, Ad = 1,
15 Izawa et al. (2007) 12mf Writing a graduation thesis2 Md = 1, In = 1, At = 1, W = 0)
6 (Ag = 1, Sx = 1, Sm = 0, Ad = 1,
16a Shea et al. (2007) 66f (pg) Early life trauma2 (CTQ) Md = 1, In = 1, At = 1, W = 0)
16b Shea et al. (2007) 66f (pg) Depression3 (EPDS) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
Md = 1, In = 0, At = 1, W = 1)
17a Smeets et al. (2007) 15f Continuous CSA memory2 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
Md = 1, In = 0, At = 1, W = 1)
17b Smeets et al. (2007) 16f Recovered CSA memory2 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
17c Smeets et al. (2007) 17f Repressed CSA memory2 Md = 1, In = 0, At = 1, W = 1)
3 (Ag = 0, Sx = 0, Sm = 0, Ad = 1,
18a Sonnenschein et al. (2007) 42mf General exhaustion5 (MBI) Md = 0, In = 1, At = 0, W = 1)
18b Sonnenschein et al. (2007) (bo) State exhaustion5 (ESM) 3 (Ag = 0, Sx = 0, Sm = 0, Ad = 1,
19 Steptoe et al. (2007) 42mf State happiness7 (EMA) Md = 0, In = 1, At = 0, W = 1)
(bo) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
73m Md = 0, In = 1, At = 1, W = 1)
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx 5
Table 1 (Continued ) Sample Predictors (measurement) CAR confounder score (0–8) Assessed days: CARi CARauc
No. Authors (year) time points (min)
Effect (r) Effect (r)
20a Therrien et al. (2007a) size size
20b Therrien et al. (2007a)
20c Therrien et al. (2007a) 50m Depression3 (BDI) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 3D: 0/30 AINC À0.170
20d Therrien et al. (2007a) Trait anxiety4 (STAI) Md = 1, In = 1, At = 0, W = 1) 3D: 0/30
21 ter Wolbeek et al. (2007) 50m Depression3 (BDI) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 3D: 0/30 AINC À0.220
22a Whitehead et al. (2007) Trait anxiety4 (STAI) Md = 1, In = 1, At = 0, W = 1) 3D: 0/30
22b Whitehead et al. (2007) 28f Fatigue5 (CIS) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/15/30 AINC À0.220
23 Wichers et al. (2007) Depression3 (BDI) Md = 1, In = 1, At = 0, W = 1) 1D: 0/15,30
24a Adam (2006) 28f Fatigue5 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/15,30 AINC À0.420
24b Adam (2006) IFN-induced depression3 Md = 1, In = 1, At = 0, W = 1) 1D: 0/30
24c Adam (2006) 132f Depression3 (CES-D) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 2D: 0/40 AINC À0.000 AUCt À0.003
25a Adam et al. (2006) (yg) Trait anxiety4 (Taylor’s) Md = 1, In = 1, At = 0, W = 1) 2D: 0/40
25b Adam et al. (2006) 72mf Trait anger4 (Taylor’s) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 2D: 0/40 AINC 0.000 AUCt 0.211
25c Adam et al. (2006) (ht) Loss of control2 (Diary) Md = 1, In = 1, At = 1, W = 1) 3D: 0/30
25d Adam et al. (2006) 72mf Threat4 (Diary) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 3D: 0/30 AINC À0.270 AUCt 0.010
25e Adam et al. (2006) (ht) Fatigue5 (Diary) Md = 1, In = 1, At = 1, W = 1) 3D: 0/30
25f Adam et al. (2006) 12mf Lively/energetic7 (Diary) 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 3D: 0/30 AINC 0.000
25h Adam et al. (2006) (hc) Sad4 (Diary) Md = 0, In = 1, At = 0, W = 0) 3D: 0/30
26a Alderling et al. (2006) 52mf Tense/angry4 (Diary) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 3D: 0/30 AINC 0.005
26b Alderling et al. (2006) (yg) Loneliness2 (Diary) Md = 1, In = 1, At = 1, W = 1) 3D: 0/30
27 Eek et al. (2006) 52mf Job stress1 (DC model) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30 AINC À0.095
28a Ellenbogen et al. (2006) (yg) Job stress1 (DC model) Md = 1, In = 1, At = 1, W = 1) 1D: 0/30
28b Ellenbogen et al. (2006) 52mf Perceived stress2 (SEI) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 3D: 0/30 AINC 0.214
28c Ellenbogen et al. (2006) (yg) Daily hassles2 Md = 1, In = 1, At = 1, W = 1) 2D: 0/30/60
28d Ellenbogen et al. (2006) 156mf Major life events2 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 2D: 0/30/60 AINC 0.165
28e Ellenbogen et al. (2006) (ol) Social problems2 (CBCL) Md = 1, In = 1, At = 1, W = 1) 2D: 0/30/60
28f Ellenbogen et al. (2006) 156mf Depression3 (CDI) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 2D: 0/30/60 AINC 0.186
28g Ellenbogen et al. (2006) (ol) Internalizing4 (CBCL) Md = 1, In = 1, At = 1, W = 1) 2D: 0/30/60
28h Ellenbogen et al. (2006) 156mf State anxiety4 (Penn’s) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 2D: 0/30/60 AINC À0.062
29a Eller et al. (2006) (ol) Negative affect4 (PANAS) Md = 1, In = 1, At = 1, W = 1) 2D: 0/30/60
29b Eller et al. (2006) 156mf Positive affect7 (PANAS) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 2D: 0/30/60 AINC À0.004
29c Eller et al. (2006) (ol) Job stress1 (Low control) Md = 1, In = 1, At = 1, W = 1) 1D: 0/30
29d Eller et al. (2006) 156mf Job stress1 (High demand) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30 AINC 0.125
29e Eller et al. (2006) (ol) Job stress1 (ERI model) Md = 1, In = 1, At = 1, W = 1) 1D: 0/30
29f Eller et al. (2006) 156mf Job stress1 (Overcommit) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30 AINC 0.005
29g Eller et al. (2006) (ol) Time pressure1 Md = 1, In = 1, At = 1, W = 1) 1D: 0/30
156mf Job stress1 (Low control) 8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30 AINC 0.141
(ol) Job stress1 (High demand) Md = 1, In = 1, At = 1, W = 1) 1D: 0/30
87m 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, AINC 0.000
Md = 1, In = 0, At = 0, W = 1)
169f 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, AINC 0.000
Md = 1, In = 0, At = 0, W = 1)
381mf 3 (Ag = 0, Sx = 0, Sm = 0, Ad = 0, AINC 0.000
Md = 0, In = 1, At = 1, W = 1)
57mf 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, AINC 0.017 AUCt 0.140
(yg) Md = 1, In = 1, At = 0, W = 0)
57mf 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, AINC À0.085 AUCt À0.110
(yg) Md = 1, In = 1, At = 0, W = 0)
57mf 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, AINC 0.156 AUCt 0.070
(yg) Md = 1, In = 1, At = 0, W = 0)
57mf 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, AINC À0.023 AUCt 0.090
(yg) Md = 1, In = 1, At = 0, W = 0)
57mf 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, AINC 0.082 AUCt 0.050
(yg) Md = 1, In = 1, At = 0, W = 0)
57mf 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, AINC 0.021 AUCt À0.070
(yg) Md = 1, In = 1, At = 0, W = 0)
57mf 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, AINC 0.067 AUCt 0.100
(yg) Md = 1, In = 1, At = 0, W = 0)
57mf 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, AINC 0.053 AUCt 0.170
(yg) Md = 1, In = 1, At = 0, W = 0)
28m 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, AINC 0.000
Md = 0, In = 1, At = 1, W = 0)
28m 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, AINC À0.020
Md = 0, In = 1, At = 1, W = 0)
28m 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, AINC 0.000
Md = 0, In = 1, At = 1, W = 0)
28m 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, AINC 0.025
Md = 0, In = 1, At = 1, W = 0)
28m 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, AINC 0.000
Md = 0, In = 1, At = 1, W = 0)
47f 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, AINC 0.008
Md = 0, In = 1, At = 1, W = 0)
52f 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, AINC 0.000
Md = 0, In = 1, At = 1, W = 0)
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
6 Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx
Table 1 (Continued ) Sample Predictors (measurement) CAR confounder score (0–8) Assessed days: CARi CARauc
No. Authors (year) time points (min)
53f Effect (r) Effect (r)
29h Eller et al. (2006) size size
50f
29i Eller et al. (2006) Job stress1 (ERI model) 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30 AINC 0.148 AUCt À0.099
55f Job stress1 (Overcommit) Md = 0, In = 1, At = 1, W = 0) 1D: 0/30 AINC 0.000 AUCt À0.093
29j Eller et al. (2006) Time pressure1 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30 AINC 0.313 AUCt À0.140
89mf Depression3 (SCID) Md = 0, In = 1, At = 1, W = 0) 2D: 0/30 AINC À0.272 AUCt
30 Giese-Davis et al. (2006) (ca) Witnessed violence2 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/60 ACOR 0.000 AUCt 0.095
45m Witnessed violence2 Md = 0, In = 1, At = 1, W = 0) 1D: 0/60 ACOR À0.093 AUCt À0.019
31a Kliewer (2006) (yg) Peer victimization2 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 1D: 0/60 ACOR À0.100 À0.065
33f Major life events2 Md = 0, In = 1, At = 1, W = 0) 1D: 0/60 ACOR 0.051 AUCt
31b Kliewer (2006) (yg) Internalizing4 (CBCL) 3 (Ag = 1, Sx = 1, Sm = 0, Ad = 0, 1D: 0/60 ACOR 0.000 AUCt À0.382
78mf Work engage1 (UWES) Md = 0, In = 1, At = 0, W = 0) 3D: 0/30 AUCi 0.000 AUCt 0.230
31c Kliewer (2006) (yg) Burnout5 (MBI) 3 (Ag = 1, Sx = 1, Sm = 0, Ad = 0, 3D: 0/30 AUCi 0.000 AUCt 0.284
78mf Depression3 (CES-D) Md = 0, In = 1, At = 0, W = 0) 2D: 0/15/30 AUCi 0.039 AUCt 0.000
31d Kliewer (2006) (yg) Exhaustion5 (MBI) 3 (Ag = 1, Sx = 1, Sm = 0, Ad = 0, 2D: 0/15/30 AUCi À0.141 AUCt 0.262
78mf Neuroticism4 (SCL-90) Md = 0, In = 1, At = 0, W = 0) 2D: 0/15/30 AUCi 0.035 AUCt 0.000
31e Kliewer (2006) (yg) Depression3 (CES-D) 3 (Ag = 1, Sx = 1, Sm = 0, Ad = 0, 2D: 0/15/30 AUCi 0.099 0.000
51m Exhaustion5 (MBI) Md = 0, In = 1, At = 0, W = 0) 2D: 0/15/30 AUCi À0.047
32a Langelaan et al. (2006) Neuroticism4 (SCL-90) 3 (Ag = 1, Sx = 1, Sm = 0, Ad = 0, 2D: 0/15/30 AUCi À0.123
45m Job stress1 (High demand) Md = 0, In = 1, At = 0, W = 0) 3D: 0/30 AINC À0.060
32b Langelaan Poor social support2 (Unde´ n’s) 7 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 3D: 0/30 AINC À0.040
et al. (2006) 34mf Hopelessness3 (Everson’s) Md = 1, In = 1, At = 1, W = 1) 3D: 0/30 AINC À0.050
Depression3 (MDIS) 7 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 3D: 0/30 AINC À0.060
33a Mommersteeg 34mf Exhaustion5 (MQ) Md = 1, In = 1, At = 1, W = 1) 3D: 0/30 AINC À0.010
et al. (2006) Self-esteem7 (Pearlin’s) 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 3D: 0/30 AINC 0.080
34mf Well-being7 Md = 0, In = 1, At = 0, W = 1) 3D: 0/30 AINC 0.060
33b Mommersteeg Childhood sexual abuse2 (CTQ) 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 2D: 0/45 AINC 0.270
et al. (2006) 73mf Major life events2 (PERI) Md = 0, In = 1, At = 0, W = 1) 2D: 0/45 AINC 0.179
(bo) Perceived stress2 (PSS) 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 2D: 0/45 AINC 0.132
33c Mommersteeg 73mf Depression3 (BDI) Md = 0, In = 1, At = 0, W = 1) 2D: 0/45 AINC À0.026
et al. (2006) (bo) PTSD6 (SCID) 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 1D: 0/30/45/60
72mf Poor marital QOL2 (MPRQ) Md = 0, In = 1, At = 0, W = 1) 1D: 0/30 AINC 0.237
33d Mommersteeg (bo) Depression3 (SCID) 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 1D: 0/15/30/45/60
et al. (2006) 257mf Caregiver stress2 Md = 0, In = 1, At = 0, W = 1) 1D: 0/30 AINC À0.213
Burnout5 (SMBQ) 5 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 1D: 0/15,30,60 MINC 0.000
33e Mommersteeg 257mf Burnout5 (SMBQ) Md = 0, In = 1, At = 0, W = 1) 1D: 0/15,30,60 MINC 0.000
et al. (2006) Negative affect4 (CAS) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 2D, 0/20/40
257mf Positive affect7 (CAS) Md = 1, In = 0, At = 1, W = 1) 2D, 0/20/40
33f Mommersteeg 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
et al. (2006) 257mf Md = 1, In = 0, At = 1, W = 1)
7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
34a Sjo¨ gren et al. (2006) 257mf Md = 1, In = 0, At = 1, W = 1)
7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
34b Sjo¨ gren et al. (2006) 257mf Md = 1, In = 0, At = 1, W = 1)
7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
34c Sjo¨ gren et al. (2006) 257mf Md = 1, In = 0, At = 1, W = 1)
7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
34d Sjo¨ gren et al. (2006) 85f Md = 1, In = 0, At = 1, W = 1)
(fm) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
34e Sjo¨ gren et al. (2006) 85f Md = 1, In = 0, At = 1, W = 1)
(fm) 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
34f Sjo¨ gren et al. (2006) 85f Md = 1, In = 0, At = 0, W = 0)
(fm) 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
34g Sjo¨ gren et al. (2006) 85f Md = 1, In = 0, At = 0, W = 0)
(fm) 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
35a Weissbecker 63mf Md = 1, In = 0, At = 0, W = 0)
et al. (2006) 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
75mf Md = 1, In = 0, At = 0, W = 0)
35b Weissbecker 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
et al. (2006) 60mf Md = 0, In = 1, At = 1, W = 0)
8 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
35c Weissbecker 98mf Md = 1, In = 1, At = 1, W = 1)
et al. (2006) 4 (Ag = 1, Sx = 1, Sm = 0, Ad = 1,
29m Md = 0, In = 0, At = 1, W = 0)
35d Weissbecker 6 (Ag = 1, Sx = 1, Sm = 0, Ad = 1,
et al. (2006) 35f Md = 1, In = 1, At = 1, W = 0)
4 (Ag = 1, Sx = 1, Sm = 0, Ad = 0,
36 Wessa et al. (2006) 80 mf Md = 0, In = 0, At = 1, W = 1)
4 (Ag = 1, Sx = 1, Sm = 0, Ad = 0,
37 Barnett et al. (2005) 80 mf Md = 0, In = 0, At = 1, W = 1)
6 (Ag = 1, Sx = 1, Sm = 0, Ad = 1,
38 Bhagwagar Md = 1, In = 0, At = 1, W = 1)
et al. (2005) 6 (Ag = 1, Sx = 1, Sm = 0, Ad = 1,
Md = 1, In = 0, At = 1, W = 1)
39 de Vugt et al. (2005)
40a Grossi et al. (2005)
40b Grossi et al. (2005)
41a Lai et al. (2005)
41b Lai et al. (2005)
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx 7
Table 1 (Continued ) Sample Predictors (measurement) CAR confounder score (0–8) Assessed days: CARi CARauc
No. Authors (year) time points (min)
Effect (r) Effect (r)
size size
41c Lai et al. (2005) 80 mf Optimism7 (LOT-R) 6 (Ag = 1, Sx = 1, Sm = 0, Ad = 1, 2D, 0/20/40 AUCt À0.314
Md = 1, In = 0, At = 1, W = 1) 1D: 0/30
42 Meinlschmidt 95mf Early loss event2 (CTES) 5 (Ag = 1, Sx = 1, Sm = 1, Ad = 0, 1D: 0/30/45/60 AINC À0.285 AUCt À0.510
and Heim (2005) (yg) PTSD6 (SCID) Md = 0, In = 1, At = 0, W = 1) 1D: 0/30,60
30mf 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30,60 AINC À0.040 AUCt 0.270
43 Neylan et al. (2005) Md = 0, In = 1, At = 1, W = 1) 1D: 0/30,60 AINC 0.265 AUCt 0.278
6 (Ag = ,1 Sx = 1, Sm = 1, Ad = 0, 1D: 0/30,60 AINC AUCt À0.471
44a Polk et al. (2005) 301mf State negative affect4 Md = 1, In = 0, At = 1, W = 1) 1D: 0/30,60 AINC À0.081 AUCt 0.127
6 (Ag = ,1 Sx = 1, Sm = 1, Ad = 0, 1D: 0/15/3/45/60 AINC À0.020 AUCt 0.000
44b Polk et al. (2005) 143m Trait negative affect4 Md = 1, In = 0, At = 1, W = 1) 1D: 0/30 AUCi À0.080 AUCt 0.000
44c Polk et al. (2005) 158f Trait negative affect4 6 (Ag = ,1 Sx = 1, Sm = 1, Ad = 0, 1D: 0/30 AINC AUCt 0.000
Md = 1, In = 0, At = 1, W = 1) 3D: 0/30/60 AINC 0.377 AUCt 0.000
44d Polk et al. (2005) 298mf State positive affect7 6 (Ag = ,1 Sx = 1, Sm = 1, Ad = 0, 1D: 0/30 0.179 AUCt 0.000
Md = 1, In = 0, At = 1, W = 1) 1D: 0/10/20/30/60 AINC 0.000
44e Polk et al. (2005) 298mf Trait positive affect7 6 (Ag = ,1 Sx = 1, Sm = 1, Ad = 0, 1D: 0/10/20/30/60 AUCt À0.584
Md = 1, In = 0, At = 1, W = 1) 1D: 0/10/20/30/60 AUCi 0.174
45 Portella et al. (2005) 30mf Neuroticism4 (EPI) 4 (Ag = 1, Sx = 1, Sm = 0, Ad = 0, 1D: 0/10/20/30/60 MINC AUCt 0.135
Md = 1, In = 0, At = 1, W = 0) 1D: 0/10/20/30/60 MINC À0.202 AUCt 0.119
46a Steptoe et al. (2005) 88m Chronic economic stress2 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/10/20/30/60 MINC À0.400 AUCt 0.045
Md = 0, In = 1, At = 1, W = 1) 2D: 0/30/45/60 AINC AUCt 0.316
46b Steptoe et al. (2005) 72f Chronic economic stress2 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 4D: 0/30,45,60 AINC 0.160 AUCt 0.373
Md = 0, In = 1, At = 1, W = 1) 4D: 0/30,45,60 AINC 0.155 AUCt 0.300
47 Stetler and Miller (2005) 69f Depression3 (DISH) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30 0.132 AUCt 0.310
Md = 1, In = 1, At = 1, W = 0) 1D: 0/30 MINC 0.226
48 Wright and Steptoe (2005) 76mf High financial strain2 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1, 1D: 0/30 0.124
Md = 1, In = 1, At = 1, W = 0) 1D: 0/15/30/45/60 ACOR
49a Ebrecht et al. (2004) 24m Perceived stress2 (PSS) 4 (Ag = 1, Sx = 1, Sm = 1, Ad = 0, 2D: 0/15,30,45 AINC 0.000
Md = 1, In = 0, At = 0, W = 0) 4D: 0/30/60 ACOR
49b Ebrecht et al. (2004) 24m Loneliness2 (UCLA-LC) 4 (Ag = 1, Sx = 1, Sm = 1, Ad = 0, 4D: 0/30/60 ACOR À0.337
Md = 1, In = 0, At = 0, W = 0) 1D, 0/30/60 ACOR 0.106
49c Ebrecht et al. (2004) 24m Optimism7 (LOT) 4 (Ag = 1, Sx = 1, Sm = 1, Ad = 0, 1D: 0/20 ACOR 0.170
49d Ebrecht et al. (2004) 24m Self-esteem7 (RSE) Md = 1, In = 0, At = 0, W = 0) 2D: 0/30/45/60 ACOR 0.214
4 (Ag = 1, Sx = 1, Sm = 1, Ad = 0, 2D: 0/30/45/60 0.215
49e Ebrecht et al. (2004) 24m Poor social support2 (SSS) Md = 1, In = 0, At = 0, W = 0) 2D: 0/30/45/60 0.000
4 (Ag = 1, Sx = 1, Sm = 1, Ad = 0, 2D: 0/30/45/60 0.000
50 Roberts et al. (2004) 91mf Chronic fatigue Md = 1, In = 0, At = 0, W = 0) 2D: 0/30/45/60
51 Rohleder et al. (2004) 25mf syndrome5 (SCAN) 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 0,
PTSD6 (DSM-IV) Md = 1, In = 1, At = 0, W = 1)
4 (Ag = 1, Sx = 1, Sm = 1, Ad = 0,
52a Schlotz et al. (2004) 219mf Job stress1 (Overload, TICS) Md = 0, In = 0, At = 1, W = 0)
6 (Ag = 1, Sx = 1, Sm = 0, Ad = 0,
52b Schlotz et al. (2004) 219mf Chronic worry4 (TICS) Md = 1, In = 1, At = 1, W = 1)
53 Steptoe et al. (2004b) 163mf Loneliness2 (UCLA-LC) 6 (Ag = 1, Sx = 1, Sm = 0, Ad = 0,
Md = 1, In = 1, At = 1, W = 1)
54a Steptoe et al. (2004c) 83m Job stress1 (Overcommit) 7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
54b Steptoe et al. (2004c) 81f Job stress1 (Overcommit) Md = 0, In = 1, At = 1, W = 1)
7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
55 Bhagwagar et al. (2003) 62mf Past depression3 (SCID) Md = 0, In = 1, At = 1, W = 1)
7 (Ag = 1, Sx = 1, Sm = 1, Ad = 1,
56 Edwards et al. (2003) 26mf Perceived stress2 (PSS) Md = 0, In = 1, At = 1, W = 1)
57a Pruessner et al. (2003) 39m Depressive symp3 (HDI) 3 (Ag = 1, Sx = 1, Sm = 0, Ad = 0,
Md = 0, In = 0, At = 1, W = 0)
57b Pruessner et al. (2003) 39m Chronic stress2 (TICS) 3 (Ag = 0, Sx = 0, Sm = 0, Ad = 1,
Md = 1, In = 1, At = 0, W = 0)
58 de Vente et al. (2003) 45mf Burnout5 (CIDI) 5 (Ag = 0, Sx = 1, Sm = 1, Ad = 0,
Exhaustion5 (MQ) Md = 1, In = 1, At = 0, W = 1)
59 Nicolson and Diest (2000) 59m 5 (Ag = 0, Sx = 1, Sm = 1, Ad = 0,
Md = 1, In = 1, At = 0, W = 1)
60a Wu¨ st et al. (2000a) 102mf Chronic worry4 (TICS) 6 (Ag = 1, Sx = 1, Sm = 1, Ad = 0,
Md = 1, In = 1, At = 0, W = 1)
60b Wu¨ st et al. (2000a) 102mf Social stress2 (TICS) 4 (Ag = 1, Sx = 1, Sm = 1, Ad = 0,
Md = 0, In = 0, At = 1, W = 0)
60c Wu¨ st et al. (2000a) 102mf Lack of social 4 (Ag = 0, Sx = 0, Sm = 1, Ad = 0,
60d Wu¨ st et al. (2000a) 102mf recognition2 (TICS) Md = 1, In = 1, At = 0, W = 1)
60e Wu¨ st et al. (2000a) 102mf Job stress1 (Overload, TICS) 4 (Ag = 0, Sx = 0, Sm = 1, Ad = 0,
Md = 1, In = 1, At = 0, W = 1)
Job stress1 4 (Ag = 0, Sx = 0, Sm = 1, Ad = 0,
(Discontent, TICS) Md = 1, In = 1, At = 0, W = 1)
4 (Ag = 0, Sx = 0, Sm = 1, Ad = 0,
Md = 1, In = 1, At = 0, W = 1)
4 (Ag = 0, Sx = 0, Sm = 1, Ad = 0,
Md = 1, In = 1, At = 0, W = 1)
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
8 Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx
Table 1 (Continued )
No. Authors (year) Sample Predictors (measurement) CAR confounder score (0–8) Assessed days: CARi CARauc
time points (min)
Effect (r) Effect (r)
size size
60g Wu¨ st et al. (2000a) 104mf Self-esteem7 (RSE) 4 (Ag = 0, Sx = 0, Sm = 1, Ad = 0, 2D: 0/30/45/60 MINC À0.160 AUCt À0.010
60h Wu¨ st et al. (2000a) 104mf Self-efficacy7 (GS) Md = 1, In = 1, At = 0, W = 1) 2D: 0/30/45/60 MINC À0.190 AUCt À0.060
61 Pruessner et al. (1999) 66mf Perceived stress2 (PSS) 4 (Ag = 0, Sx = 0, Sm = 1, Ad = 0, 2D: 0/15/30/60 ACOR
62 Schulz et al. (1998) 85mf Job stress1 (Overload, TICS) Md = 1, In = 1, At = 0, W = 1) 3D: 0/15/30/45/60 ACOR 0.066
6 (Ag = 1, Sx = 1, Sm = 0, Ad = 0, 0.293
Md = 1, In = 1, At = 1, W = 1)
1 (Ag = 0, Sx = 1, Sm = 0, Ad = 0,
Md = 0, In = 0, At = 0, W = 0)
Abbreviation: BDI, Beck depression inventory; BPD, borderline personality disorder CAPS, clinical administered PTSD; CAR, cortisol awaking response, CAS, Chinese affect
scale; CBCL, child behavior checklist; CDI, child depression inventory; CES-D, center of epidemiologic studies depression scale; CIDI, composite international diagnostic
interview; CIS, checklist individual strength; CMHS, Cook-Medley hostility scale; CSA, childhood sexual abuse; CTES, childhood traumatic events survey; CTQ, childhood
trauma questionnaire; DC, demand/control; DISH, depression interview and structured Hamilton; DS-16, type-D scale-16; DSM-IV, diagnostic and statistical manual-IV;
ECRS, the experiences in close relationships scale; EDI-2, eating disorder inventory-2; EMA, ecological momentary assessment; EPDS, Edinburgh postnatal depression scale;
EPI, Eysenck personality inventory; ERI, effort/reward imbalance; ESM, experience sampling method; GS, generalized self-efficacy; HDI, Hamilton depression inventory; IFN,
interferon; KSD, Karolinska sleep diary; LOT-R, life orientation test-revised; MASC, multidimensional anxiety scale; MBI, Maslach burnout inventory; MDIS, major depression
inventory scale; MPRQ, marital/partner role quality scales; MPS, Frost multi-dimensional perfectionism scale; MQ, Maastricht questionnaire; PERI, psychiatric epidemiology
research interview; PSQI, Pittsburgh sleep quality index; PSS, perceived stress scale; PTSD, posttraumatic stress disorder; QOL, quality of life; RCADS, revised child anxiety and
depression scale; RSE, Rosenberg self-esteem scale; SCAN, schedules for clinical assessment in neuropsychiatry; SCID, structured clinical interview for DSM-IV (TR); SCL-90,
symptom checklist; SEI, stress and energy inventory; SES, socioeconomic status; SMBQ, Shirom-Melamed burnout questionnaire; SSS, short-form social support scale; STAI,
state-trait anxiety inventory; TICS, Trier inventory for the assessment of chronic stress; UCLA-LC, UCLA loneliness scale; UWES, Utrecht work engagement scale. Code number
(No.): The studies with the same number share a corresponding cohort sample. Sample: bo, clinical burnout patients; ca, cancer patients: cd, coronary artery disease; fm,
fibromyalgia patients; hc, hepatitis C patients; ht, coronary heart disease patients; ol, old subjects (>60 years old); pg, pregnant women; pp, postpartum; yg, young subjects
( 18 years old); m/f refers sex of samples. Predictors: (1) job stress; (2) general life stress; (3) depression or hopelessness; (4) anxiety, neuroticism, and negative affect; (5)
fatigue, burnout, or exhaustion; (6) posttraumatic stress syndrome; (7) positive psychological states or traits. CAR confounder score: 1, accounted for; 0, not accounted for; Ag,
age; At, awakening time; Ad, participant adherence; In, clear introduction to participants about the sampling procedure; Md, medication of any steroids; Sm, smoking; Sx,
sex; W, weekday/weekend. Assessed days: time points (min): D, the number of days studied; time points after post-awakening. Outcome: D, the number of days studied; CARi:
AUCi, area of cortisol increase under the curve, AINC, absolute increase of cortisol (max. value of ‘15, 20, 30, 45, or 60 min post-awakening’ minus wakening value), MINC,
mean value of ‘15, 20, 30, 45 or 60 min post-awakening’ minus wakening value), ACOR, absolute values post-awakening evaluated by repeated analysis of variables; CARauc:
AUCt, total area under the curve. Effect size: r, correlation coefficient.
Table 2 summarizes the detailed characteristics of 145 studies in Table 2
the review. Characteristics of the studies included in the meta-analyses.
3.2. Study characteristics and quality Characteristics Enrolled studies
The sample size of the enrolled studies averaged 91.8 Number %
participants (range: 12–542). These studies evaluated the follow-
ing psychosocial factors, listed in order of frequency: general (non- Total study number 147 100
work) life stress (41 studies, including major life events, caregiver
stress, poor marital quality, poor social functioning, loneliness, Sample size (average number Æ S.D.) 91.8 Æ 85.5 15.0
witnessed violence, early loss event, childhood sexual abuse, etc.); 27.9
anxiety, neuroticism, or negative affect (26 studies, including Predictors 22 14.3
negative affect, anxiety, chronic worry, angry, threat, sad, (1) Job stress 41 17.7
neuroticism, internalizing symptoms, etc.); job stress (22 studies (2) General life stress 21 10.9
of the demand-control model, effort-reward imbalance model, (3) Depression or hopelessness 26
overcommitment, overload, discontent with work, and time (4) Anxiety, neuroticism, or negative affect 16 3.4
pressure); depression and hopelessness (21 studies); positive (5) Fatigue, burnout, or exhaustion 10.9
psychological states or traits (16 studies of well-being, happiness, (6) Posttraumatic stress syndrome 5
optimism, vitality, lively/energetic feelings, self-esteem, and self- (7) Positive psychological states or traits 16 85.7
efficacy); fatigue, burnout, and exhaustion (16 studies); and 92.5
posttraumatic stress syndrome (5 studies). The methodological Methodological quality score (ÆSD) 5.62 Æ 1.50 64.6
quality scores (0–8) averaged 5.6, with 55.1% (81 studies) scoring 6 (1) Age details provided 126 70.1
or more. 54.4% of the studies measured the CAR for 2 days or more. (2) Sex details provided 136 64.6
More studies assessed the CARi (129 studies, 87.8%) than the (3) Smoking assessed or controlled 68.7
CARauc (59 studies, 40.1%) (Table 2). (4) Participant adherence assessed 95 52.4
(5) Steroid medication controlled 103 60.5
3.3. Meta-analysis of cortisol changes after waking (CARi) (6) Clear instructions to participants 55.1
(7) Satisfactory measures of awakening time 95
The meta-analysis of CARi studies demonstrated positive (8) Weekday/weekend taken into account 101 45.6
associations between the magnitude of the CARi and job stress Quality score 6 or more 36.1
and general life stress (r = 0.061, 95% CI 0.012–0.110, p = 0.015 and 77 15.6
r = 0.069 [0.012–0.126], p = 0.017, respectively). By contrast, there 89
tended to be a negative relationship between the CARi and fatigue, 81 2.7
Number of days assessed 67 87.8
1 day 53 40.1
2 days 23
3 days
4 days or more 4
Outcomes 129
CAR cortisol changes following waking (CARi) 59
CAR total volume of cortisol released (CARauc)
CAR, cortisol awakening response; S.D., standard deviation.
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx 9
burnout, or exhaustion (r = À0.065 [À0.138–0.010], p = 0.089). the CARi was inversely related to posttraumatic stress syndrome in
Depression and hopelessness, anxiety, neuroticism, or negative methodologically stronger studies. By contrast, the association
affect, posttraumatic stress syndrome, and positive psychological with general life stress was no longer significant.
states or traits were not associated with the CARi in the overall
meta-analyses. These findings were not accompanied by publica- We carried out subgroup meta-analyses separating the four
tion bias. The analyses on aggregate effect sizes confirmed the different types of CARi indicator. These showed a positive
positive association between CARi and job stress (Fig. 2). association with general life stress for ACOR, and inverse
associations with positive psychological factors for both AUCi
When we limited the meta-analysis to studies with high and MINC. Depression or hopelessness was negatively associated
methodological quality scores (!6), the positive associations with CARi in the analyses of AINC and AUCi, and so was fatigue,
between CARi and job stress remained significant. Additionally, burnout, or exhaustion in AUCi. In studies that assessed the CAR for
Fig. 2. CARi: results of meta-analyses, subgrouping, and sensitivity analyses r, correlation coefficient; AINC, absolute increase of cortisol; AUCi, area of cortisol increase under
the curve; MINC, mean value of ‘15, 20, 30, 45 or 60 min post-awakening’ minus wakening value; ACOR, absolute values post-awakening evaluated by repeated analysis of
variables; bold numbers indicate p < 0.05; CI, confidence interval. *Indicates significant publication bias by Egger’s method (p < 0.10).
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
10 Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx
2 days or more, there were positive associations between CARi and assessed the CARauc for 3 time-points or more were significant
general life stress and negative associations with depression or (Fig. 3).
hopelessness. The studies that assessed the CAR over 3 or more
time-points showed that the CARi was significantly associated 4. Discussion
with general life stress, fatigue, burnout, or exhaustion, and
positive psychological states or traits. Since the general life stress To our knowledge, this is the first systematic review using
category was heterogenous, we analyzed separately different types meta-analytic methods to confirm that psychosocial factors are
of negative life experience, including childhood and recent life differentially associated with an enhanced or reduced CAR. These
events, low social support and loneliness. The effects were not associations differed somewhat across study design factors,
significant in these analyses, but the number of studies contribut- including study methodology quality score and the number of
ing to each category was small (data not shown). days or time-points assessed, and between the two types of CAR
outcome.
3.4. Meta-analysis of overall cortisol output (CARauc)
4.1. Differences in CAR across psychosocial factors
There were fewer studies of the overall level of cortisol release
over the waking period compared with the CARi. In particular, only The most robust findings from these meta-analyses indicate
two studies of job stress could be identified. Nevertheless, the that job stress and general life stress are associated with an
magnitude of the CARauc was positively related to general life increased CAR, whereas fatigue, burnout, or exhaustion are
stress (r = 0.132 [0.044–0.218], p = 0.003). These overall findings characterized by a reduced CAR. It is interesting that there is
were not accompanied by publication bias. The analysis on some evidence for positive psychological traits being related to a
aggregate effect sizes also confirmed a positive association reduced CAR, and this is consistent with positive characteristics
between CARauc and general life stress. The subgroup meta- having biological correlates that are the opposite of negative
analyses revealed a significant association between general life affective responses (Steptoe et al., 2007). A relationship between
stress and the CARauc in studies of men. Other factors such as age posttraumatic stress syndrome and a reduced CAR was also
and the health status of the sample were not related to the CARauc observed, but only in the methodologically better studies. Several
in studies of general life stress. When analysis was limited to of the factors associated with a reduced CAR differ from the
studies with high methodological quality scores, significant experience of general and work stress in eliciting a conservation/
associations were found not only with general life stress, but also withdrawal response rather than active engagement with the
with posttraumatic stress syndrome and with positive psycholo- environment. The lack of overall associations between depression
gical states or traits. None of the analyses restricted to studies that and the CAR was not because of an absence of effects, but rather
Fig. 3. CARauc: results of meta-analyses, subgrouping, and sensitivity analyses r, correlation coefficient; bold numbers indicate p < 0.05; CI, confidence interval. *Indicates
significant publication bias by Egger’s method (p < 0.10).
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx 11
because depression has been related both to increased (Bhagwagar related to chronic stress (Heim et al., 2000; McEwen, 2007; Wirtz
et al., 2003; Pruessner et al., 2003) and reduced (Stetler and Miller, et al., 2007). A second possibility is that in conditions such as
2005; Ellenbogen et al., 2006) CARs. These discrepancies may burnout and fatigue, sleep patterns are seriously disrupted
relate to the intensity of depression, with some studies investigat- (Ekstedt et al., 2006). The CAR is thought to be due to 2–4
ing major depressive disorder while others concerned elevated secretory bursts over the period after waking. If sleep patterns are
depressed mood in the normal range. Additionally, some studies disturbed, perhaps people become aroused to levels near
managed less than complete control for factors associated with consciousness before they define themselves as awake. The
depression. It is interesting that studies involving two or more days secretory bursts may then be distributed over a longer time
of CAR assessment showed an inverse association with depression, period, so the CAR itself is smaller. Coupling cortisol assessments
whereas those involving three or more time points of saliva with objective indicators of sleep and waking (polysomnography,
sampling showed a positive relationship. However, it is striking heart rate monitoring and actigraphy) would help resolve this
that when analyses were limited to studies with high methodo- issue (Kupper et al., 2005; Wilhelm et al., 2007; Dockray et al.,
logical quality scores, the positive association between the CARi 2008).
and job stress and between the CARauc and general life stress
remained significant. We focused on the psychosocial correlates of the CAR. The
degree to which these effects are related to health states remains to
It should be noted that nearly all the studies reviewed were be determined. However, there have been a number of interesting
cross-sectional, so the causal significance of associations is unclear. observations suggesting that the CAR changes may be linked with
However, one recent study assessed changes in financial strain physical health risk. For example, the CAR has been positively
over 3-year period, and demonstrated that a reduction in financial associated with waist/hip ratio in men, suggesting it may be a
strain was accompanied by a decrease in the CAR, indicating a marker of HPA dysfunction in abdominal adiposity (Wallerius
parallel between changes in chronic stress and the magnitude of et al., 2003; Steptoe et al., 2004a). Another study assessed
the CAR (Steptoe et al., 2005). More acutely, Adam et al. (2006) relationships with the rate of healing of experimentally adminis-
showed that greater loneliness, sadness or threat 1 day predicted tered wounds on healthy volunteers (Ebrecht et al., 2004). Wound
an increased CAR on the following morning, while a larger CAR was healing was slower in people with a larger CAR. Eller et al. (2005)
in turn associated with lower fatigue over the subsequent day. reported that the progression of subclinical atherosclerosis,
indexed by carotid intima–media thickness, was accelerated in
4.2. Underlying mechanisms women with a greater CAR, while a recent comparison of patients
with and without documented coronary artery disease showed
The association between psychosocial factors and the CAR could greater CARauc values in the disease group (Bhattacharyya et al.,
be mediated through behavioral pathways. For example, psycho- 2008). Therefore, it is conceivable that variations in the CAR
social factors are thought to influence behavioral processes related stimulated by psychosocial factors may affect later health states.
to the CAR including smoking (Steptoe and Ussher, 2006; Badrick
et al., 2007), awakening time (Edwards et al., 2001b; Federenko 4.3. CARi and CARauc
et al., 2004; Kudielka et al., 2006), sleep duration and quality (Wu¨ st
et al., 2000b), and obesity (Steptoe et al., 2004a; Therrien et al., The current review confirmed that associations with psycho-
2007b). However, significant associations between psychosocial social factors varied across the two types of CAR outcome. The
factors and the CAR persisted even after controlling for factors such change in cortisol relative to waking values has been the primary
as smoking and waking time, suggesting that other pathways may measure of the CAR, with fewer studies of the overall level of
also be involved. Although we cannot rule out other unmeasured cortisol release (CARauc). The latter of course depends on the
factors that could potentially have confounded the associations cortisol level present on waking; if the waking cortisol level is high,
observed, direct physiological pathways probably also contribute then the CARauc will be large even if the increase following waking
to associations between psychosocial factors and the CAR. As is minimal. One recent study carried out under controlled
Wilhelm et al. (2007) have noted, waking up is associated with a laboratory conditions demonstrated that the cortisol increase
profound and global activation of neocortical networks mediated after waking is a response to awakening, and is distinct from the
mainly by the brain stem arousing systems (Steriade and McCarley, circadian rise in HPA activity in the morning hours (Wilhelm et al.,
2005). On a psychological level, this arousal is presumably linked 2007). Furthermore, the CARauc but not the CARi is correlated with
to the reactivation of memory representations residing in 12-h diurnal mean cortisol concentration (Edwards et al., 2001a).
neocortical circuits. There might be a preferential reactivation of The CARi may therefore be a more appropriate measure for
representations that are deeply rooted and serve trait-like assessing HPA activation following waking in relation to psycho-
functions in the organization of personality, identity and the self, social factors.
as well as of representations that have remained preactivated from
more acute experiences, thereby increasing CAR. Indeed, the CAR is 4.4. Issues from sensitivity analyses
abolished in patients with memory disorders due to damage in the
frontal lobes or the hippocampal region (Buchanan et al., 2004; The subgroup meta-analyses by four different CARi indicators
Wolf et al., 2005). revealed a positive association with general life stress for ACOR,
and inverse associations with positive psychological factors in
These analyses indicate that psychosocial factors such as studies assessing both AUCi and MINC but not AINC. Depression or
fatigue, burnout, or exhaustion and probably posttraumatic stress hopelessness was negatively related to the CARi as measured with
syndrome were found to be negatively associated with CAR. One AINC or AUCi, while fatigue, burnout, or exhaustion were
possible explanation is that the mechanisms underlying the CAR negatively associated when only the AUCi was recorded. It is,
become exhausted, as in other cases of hypocortisolism. It is however, difficult to conclude that there are specific associations
notable, for example, that the phenomenon of hypocortisolism has between particular measures of the AUCi and psychosocial factors,
not only been reported for patients with PTSD, but was also since there were insufficient studies to evaluate relationships
observed in patients with several bodily disorders including between each indicator and psychosocial categories. Additionally,
chronic fatigue syndrome, fibromyalgia, hypertension, rheumatoid it should be noted that the different measures of change after
arthritis, and asthma, and many of these disorders have been
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
12 Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx
waking are highly intercorrelated. If, for example, only two cortisol It should also be noted that the method of scaling psychosocial
samples are taken – one on waking and a second after 30 or 45 min factors was inconsistent across studies, with some using binary
– then the MINC and AINC measures are identical. divisions, others tertiles or quartiles, established cut-off scores, or
continuous scores. The mood variables assessed in these meta-
In the analyses of studies involving measurement for at least 2 analyses included both traits and states, but there were too few
days, the positive association between the CARi and general life studies available usefully to compare these categories. None of the
stress, and the negative associations with depression or hope- existing measures of the CAR is perfect, since they do not all
lessness were significant, while the relationship with job stress capture elements such as latency, exact peak levels, and rate of
shown in the overall analyses was no longer reliable. Unfortu- decline following the peak. Taken together, given the small number
nately, it is difficult to decide on the basis of this review whether of studies analyzed and the small combined effect sizes in some
these discrepancies are due to the greater reliability of having analyses, the current findings showing associations between
assessments on more days, or the number of studies analysed. For psychosocial factors and the CAR should be interpreted with
example, the nonsignificant relationship between job stress and caution and additional research is needed. The publication of
the AUCi with 2 or more days monitoring was based on only eight negative or null results would be beneficial, and improvement in
studies, whereas the association with general life stress that study quality is needed, with control for confounders, better
remained significant involved 20 studies. Moreover, CAR con- identification of delays in taking measures, more days of
founders might mask associations, since more than half of the assessment, and reporting of both the CARi and CARauc. Moreover,
studies involving at least 2 days measurement which showed no the use of prospective designs to investigate the effects of
psychosocial associations had rather poor methodological quality psychosocial factors on the CAR in relation to health outcomes,
scores ( 5). It has recently been argued that at least 6 days of and the measurement of possible behavioral and biological
measurement are needed to assess the CARi reliably, and 2 days for mediators such as sleep duration and quality, physical activity,
the CARauc (Hellhammer et al., 2007). However, there is also obesity, and genetic polymorphisms, would advance this field.
important between-day covariation between the CAR and psy-
chosocial experience (Adam et al., 2006). It is also possible that 4.6. Conclusions
more strict control of adherence to measurement protocols and
timing will increase measurement reliability without the need for The current meta-analysis showed that different psychosocial
multiple days of assessment (Broderick et al., 2004; Thorn et al., factors are associated both with an enhanced and reduced CAR.
2006). Although the CAR has only begun to be studied intensively over
recent years in stress research, it is already providing valuable
Another method of assessing the robustness of effects was to information about psychosocial factors and health.
limit the analyses to studies that measured the CAR over at least 3
time points. The results from two time points (such as waking and Conflict of interest
30 min later) may be misleading if peak CARs occur over a wide
time window (15–45 min). It is notable that in the analyses of None declared.
studies with !3 measurement points, the associations between the
CARi and general life stress, depression or hopelessness, fatigue, Acknowledgements
burnout or exhaustion, and positive psychological states or traits,
were confirmed. We are grateful to colleagues in many research centres for
providing the additional data required for meta-analysis.
Some of the overall meta-analyses demonstrated significant
heterogeneities as indexed by the Q-test. This implies that there was Contributors: Both authors commented critically on the manu-
heterogeneity within categories of psychosocial factor, resulting script and agreed the final version. Y.C. was responsible for concept
either from differences in population characteristics (e.g., gender and design, retrieved and screened papers against inclusion
and health status) or from differences in study methodology (e.g., criteria, abstracted data from papers, appraised the quality of
study quality and design). In an attempt to evaluate which variables papers, and performed the analyses. A.S. verified study inclusion
affected these heterogeneities, we carried out sensitivity analyses by and data extractions, appraised quality of papers, commented on
population characteristics and study methodologies. Interestingly, the analysis and interpretation, and contributed to writing up.
the overall heterogeneities were not consistently reduced by
sensitivity analyses; for instance, the heterogeneity in the overall Role of the funding source: This research was funded by the
association between CARi and general life stress was no longer Kanae Foundation for the Promotion of Medical Science (Japan,
significant in the sensitivity analysis of studies assessing cortisol at 3 Y.C.), the Sumitomo Life Social Welfare Services Foundation (Japan,
or more time-points, while it remained significant in studies Y.C.), the Medical Research Council (U.K., Y.C.), and the British
assessing the CAR for 2 or more days. Therefore, these overall Heart Foundation (U.K., A.S.). The funders of this research had no
heterogeneity effects are probably due to a complex combination of role in the study design, analysis and interpretation of data, writing
two or more study population characteristics and methodologies. the report, or in the decision to submit the paper for publication.
We cannot completely delineate the causes for these heterogene-
ities, because the number of studies was too small for carrying out Appendix A. Supplementary data
more refined sensitivity analyses.
Supplementary data associated with this article can be found in
4.5. Limitations and guidelines for future studies the online version, at doi:10.1016/j.biopsycho.2008.10.004.
Our review has several limitations. It was restricted to the References
evaluation of results in published papers. We found evidence of
publication biases a few of the subgroup meta-analyses (e.g., Adam, E.K., 2006. Transactions among adolescent trait and state emotion and
studies of anxiety, neuroticism, or negative affect involving !2 diurnal and momentary cortisol activity in naturalistic settings. Psychoneur-
days), though not in the main meta-analyses. This may imply a oendocrinology 31, 664–679.
positive publication bias, if authors are more likely to submit, or
editors accept, positive than null (negative or inconclusive) results.
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004
G Model
BIOPSY-6049; No of Pages 14
Y. Chida, A. Steptoe / Biological Psychology xxx (2008) xxx–xxx 13
Adam, E.K., Hawkley, L.C., Kudielka, B.M., Cacioppo, J.T., 2006. Day-to-day Eller, N.H., Netterstrøm, B., Hansen, A˚ .M., 2006. Psychosocial factors at home and at
dynamics of experience-cortisol associations in a population-based sample work and levels of salivary cortisol. Biological Psychology 73, 280–287.
of older adults. Proceedings of National Academy of Science USA 103, 17058–
17063. Eller, N.H., Netterstrøm, B., Allerup, P., 2005. Progression in intima media thick-
ness—the significance of hormonal biomarkers of chronic stress. Psychoneur-
Alderling, M., Theorell, T., de la Torre, B., Lundberg, I., 2006. The demand control oendocrinology 30, 715–723.
model and circadian saliva cortisol variations in a Swedish population based
sample (the PART study). BMC Public Health 6, 288. Federenko, I., Wu¨ st, S., Hellhammer, D.H., Dechoux, R., Kumsta, R., Kirschbaum, C.,
2004. Free cortisol awakening responses are influenced by awakening time.
Badrick, E., Kirschbaum, C., Kumari, M., 2007. The relationship between smoking Psychoneuroendocrinology 29, 174–184.
status and cortisol secretion. Journal of Clinical Endocrinology and Metabolism
92, 819–824. Fekedulegn, D.B., Andrew, M.E., Burchfiel, C.M., Violanti, J.M., Hartley, T.A., Charles,
L.E., Miller, D.B., 2007. Area under the curve and other summary indicators of
Barnett, R.C., Steptoe, A., Gareis, K.C., 2005. Marital-role quality and stress-related repeated waking cortisol measurements. Psychosomatic Medicine 69, 651–665.
psychobiological indicators. Annals of Behavioral Medicine 30, 36–43.
Giese-Davis, J., Wilhelm, F.H., Conrad, A., Abercrombie, H.C., Sephton, S., Yutsis, M.,
Bartels, M., de Geus, E.J.C., Kirschbaum, C., Sluyter, F., Boomsma, D.I., 2003. Herit- Neri, E., Taylor, B., Kraemer, H.C., Spiegel, D., 2006. Depression and stress
ability of daytime cortisol levels in children. Behavior Genetics 33, 421–433. reactivity in metastatic breast cancer. Psychosomatic Medicine 68, 675–683.
Bhagwagar, Z., Hafizi, S., Cowen, P.J., 2005. Increased salivary cortisol after waking in Gonzalez, A., Jenkins, J.M., Steiner, M., Fleming, A.S., in press. The relation between
depression. Psychopharmacology 182, 54–57. early life adversity, cortisol awakening response and diurnal salivary cortisol
levels in postpartum women. Psychoneuroendocrinology.
Bhagwagar, Z., Hafizi, S., Cowen, P.J., 2003. Increase in concentration of waking
salivary cortisol in recovered patients with depression. American Journal Greaves-Lord, K., Ferdinand, R.F., Oldehinkel, A.J., Sondeijker, F.E.P.L., Ormel, J.,
Psychiatry 160, 1890–1891. Verhulst, F.C., 2007. Higher cortisol awakening response in young adolescents
with persistent anxiety problems. Acta Psychiatrica Scandinavica 116, 137–
Bhattacharyya, M.R., Molloy, G.J., Steptoe, A., 2008. Depression is associated with 144.
flatter cortisol rhythms in patients with coronary artery disease. Journal of
Psychosomatic Research 65, 107–113. Grossi, G., Perski, A., Ekstedt, M., Johansson, T., Lindstro¨ m, M., Holm, K., 2005. The
morning salivary cortisol response in burnout. Journal of Psychosomatic
Broderick, J.E., Arnold, D., Kudielka, B.M., Kirschbaum, C., 2004. Salivary cortisol Research 59, 103–111.
sampling compliance: comparison of patients and healthy volunteers. Psycho-
neuroendorinology 29, 636–650. Harris, A., Ursin, H., Murison, R., Erksen, H.R., 2007. Coffee, stress and cortisol in
nursing staff. Psychoneuroendorinology 32, 322–330.
Buchanan, T.W., Kern, S., Allen, J.S., Tranel, D., Kirschbaum, C., 2004. Circadian
regulation of cortisol after hippocampal damage in humans. Biological Psy- Heim, C., Ehlert, U., Hellhammeer, D.H., 2000. The potential role of hypocortisolism
chiatry 56, 651–656. in the pathophysiology of stress-related bodily disorders. Psychoneuroendo-
crinology 25, 1–35.
Chan, S.W.Y., Goodwin, G.M., Harmer, C.J., 2007. Highly neurotic never-depressed
students have negative biases in information processing. Psychological Med- Hellhammer, J., Fries, E., Schweisthal, O.W., Schlotz, W., Stone, A.A., Hagemann, D.,
icine 37, 1281–1291. 2007. Several daily measurements are necessary to reliably assess the cortisol
rise after awakening: state and trait components. Psychoneuroendocrinology
Chida, Y., Hamer, H., 2008. Chronic psychosocial factors and acute physiological 32, 80–86.
responses to laboratory induced stress in the healthy populations: a quantita-
tive review of 30 years of investigations. Psychological Bulletin 134, 829–885. Izawa, S., Sugaya, N., Ogawa, N., Nagano, Y., Nakano, M., Nakase, E., Shirotsuki, K.,
Yamada, K.C., Machida, K., Kodama, M., Nomura, S., 2007. Episodic stress
Chida, Y., Hamer, M., Steptoe, A., 2008a. A bidirectional relationship between associated with writing a graduation thesis and free cortisol secresion after
psychosocial factors and atopic disorders: a systematic review and meta- awakening. International Journal of Psychophysiology 64, 141–145.
analysis. Psychosomatic Medicine 70, 102–116.
Johnson, D.M., Delahanty, D.L., Pinna, K., 2008. The cortisol awakening response as a
Chida, Y., Hamer, M., Wardle, J., Steptoe, A., 2008b. Do stress-related psychosocial function of PTSD severity and abuse chronicity in sheltered battered women.
factors contribute to cancer incidence and survival? Nature Clinical Practice Journal of Anxiety Disorders 22, 793–800.
Oncology 5, 466–475.
Kallen, V.L., Tulen, J.H.M., Utens, E.M.W.J., Treffers, P.D.A., de Jong, F.H., Ferdinand,
Chida, Y., Steptoe, A., 2008. Positive psychological well-being and mortality: a R.F., 2008. Associations between HPA axis functioning and level of anxiety in
quantitative review of prospective observational studies. Psychosomatic Med- children and adolescents with an anxiety disorder. Depression and Anxiety 25,
icine 70, 741–756. 131–141.
Clow, A., Thorn, L., Evans, P., Hucklebridge, F., 2004. The awakening cortisol Kliewer, W., 2006. Violence exposure and cortisol responses in urban youth.
response: methodological issues and significance. Stress 7, 29–37. International Journal of Behavioral Medicine 13, 109–120.
de Kloet, C.S., Vermeten, E., Heijnen, C.J., Geuze, E., Lentjes, E.G.W.M., Westenberg, Kudielka, B.M., Federenko, I.S., Hellhammer, D.H., Wu¨ st, S., 2006. Morningness and
H.G.M., 2007. Enhanced cortisol suppression in response to dexamethasone eveningness: the free cortisol rise after awakening in ‘‘early birds’’ and ‘‘night
administration in traumatized veterans with and without posttraumatic stress owls’’. Biological Psychology 72, 141–146.
syndrome. Psychoneuroendocrinology 32, 215–226.
Kupper, N., de Geus, E.J.C., van den Berg, M., Kirschbaum, C., Boomsma, D.I., Will-
Dersimonian, R., Laird, N., 1986. Metaanalysis in clinical-trials. Controlled Clinical emsen, G., 2005. Familial influences on basal salivary cortisol in an adult
Trials 7, 177–188. population. Psychoneuroendocrinology 30, 857–868.
de Vente, W., van Amsterdam, J.G.C., Kamphuis, J.H., Emmelkamp, P.M.G., 2003. Lai, J.C.L., Evans, P.D., Ng, S.H., Chong, A.M.L., Siu, O.T., Chan, C.L.W., Ho, S.M.Y., Ho,
Physiological differences between burnout patients and healthy controls: blood R.T.H., Chan, P., Chan, C.C., 2005. Optimism, positive affectivity, and salivary
pressure, heart rate, and cortisol responses. Occupational and Environmental cortisol. British Journal of Health Psychology 10, 467–484.
Medicine 60, i54–i61.
Langelaan, S., Bakker, A.B., Schaufeli, W.B., van Rhenen, W., van Doornen, L.J.P., 2006.
de Vugt, M.E., Nicolson, N.A., Lousberg, R., Jolle, J., Verhey, F.R.J., 2005. Behavioral Scandinavian Journal of Work Environment & Health 32, 339–348.
problems in dementia patients and salivary cortisol patterns in caregivers.
Journal of Neuropsychiatry and Clinical Neuroscience 17, 201–207. Nater, M.N., Maloney, E., Boneva, R.S., Gurbaxani, B.M., Jin-Mann, L., Jones, J.F.,
Reeves, W.C., Hein, C., 2008. Attenuated morning salivary cortisol concentra-
Dockray, S., Bhattacharyya, M.R., Molloy, G.J., Steptoe, A., 2008. The cortisol awa- tions in a population-based study of persons with chronic fatigue syndrome and
kening response in relation to objective and subjective measures of waking in well controls. Journal of Clinical Endocrinology and Metabolism 93, 703–709.
the morning. Psychoneuroendocrinology 33, 77–82.
Masui, K., 2003. In: Masui, K. (Ed.), Meta-analysis software User’s Guide (written in
Ebrecht, M., Hextall, J., Kirtley, L.-G., Taylor, A., Dyson, M., Weinman, J., 2004. Japanese) Kokokara-hajimeru Meta-analysis. Shinko-Koueki Medical Press,
Perceived stress and cortisol levels predict speed of wound healing in healthy Tokyo, pp. 91–121., http://www2.ocn.ne.jp/$shinkoh/.
male adults. Psychoneuroendocrinology 29, 798–809.
McEwen, B.S., 2007. Physiology and neurobiology of stress and adaptation: central
Edwards, S., Hucklebridge, F., Clow, A., Evans, P., 2003. Components of the diurnal role of the brain. Physiological Review 87, 873–904.
cortisol cycle in relation to upper respiratory symptoms and perceived stress.
Psychosomatic Medicine 65, 320–327. Meinlschmidt, G., Heim, C., 2005. Decreased cortisol awakening response after early
loss experience. Psychoneuroendocrinology 30, 568–576.
Edwards, S., Clow, A., Evans, P., Hucklebridge, F., 2001a. Exploration of the awaken-
ing cortisol response in relation to diurnal cortisol secretary activity. Life Mommersteeg, P.M.C., Heijnen, C.J., Verbraak, M.J.P.M., van Doornen, L.J.P., 2006.
Science 68, 2093–2103. Clinical burnout is not reflected in the cortisol awakening response, the day-
curve or the response to a low-dose dexamethasone suppression test. Psycho-
Edwards, S., Evans, P., Hucklebridge, F., Clow, A., 2001b. Association between time of neuroendocrinology 31, 216–225.
awakening and diurnal cortisol secretory activity. Psychoneuroendocrinology
26, 613–622. Myrtek, M., 1995. Type-A behavior pattern, personality-factors, disease, and phy-
siological reactivity—a metaanalytic update. Personality and Individual Differ-
Eek, F.C., Garde, A.H., Hansen, A˚ .M., Persson, R., Ørbæk, P., Karlson, B., 2006. The ences 18, 491–502.
cortisol awakening response—an exploration of intraindividual stability and
negative responses. Scandinavian Journal of Work Environment & Health Neylan, T.C., Brunet, A., Pole, N., Best, S.R., Metzler, T.J., Yehuda, R., Marmar, C.R.,
Supplement 2, 15–21. 2005. PTSD symptoms predict waking salivary cortisol levels in police officers.
Psychoneuroendocrinology 30, 373–381.
Egger, M., Smith, G.D., Schneider, M., Minder, C., 1997. Bias in meta-analysis
detected by a simple, graphical test. BMJ 315, 629–634. Nicolson, N.A., Diest, R.V., 2000. Salivary cortisol patterns in vital exhaustion.
Journal of Psychosomatic Research 49, 335–342.
Ekstedt, M., Soderstrom, M., Akerstedt, T., Nilsson, J., Sondergaard, H.P., Aleksander,
P., 2006. Disturbed sleep and fatigue in occupational burnout. Scandinavian O’Donnell, K., Badrick, E., Kumari, M., Steptoe, A., 2008. Psychological coping styles
Journal of Work Environment & Health 32, 121–131. and cortisol over the day in healthy older adults. Psychoneuroendocrinology 33,
601–611.
Ellenbogen, M.A., Hodgins, S., Walker, C.-D., Couture, S., Adam, S., 2006. Daytime
cortisol and stress reactivity in the offspring of parents with bipolar disorder. Pole, N., 2007. The psychophysiology of posttraumatic stress disorder: a meta-
Psychoneuroendocrinology 31, 1164–1180. analysis. Psychological Bulletin 133, 725–746.
Please cite this article in press as: Chida, Y., Steptoe, A., Cortisol awakening response and psychosocial factors: A systematic review and
meta-analysis. Biol. Psychol. (2008), doi:10.1016/j.biopsycho.2008.10.004