ReviewDiurnal cortisol slopes and mental and physical health outcomes: A systematic review and meta-analysis
Introduction
The glucocorticoid hormone cortisol is a primary product of the hypothalamic-pituitary-adrenal (HPA) axis, a key biological stress response system. Cortisol is one of the most frequently employed biomarkers in psychobiological research for several reasons. First, cortisol levels are responsive to social and psychological stress (Dickerson and Kemeny, 2004, Gunnar et al., 2009a). Cortisol levels respond to both acute stress (e.g., acute loneliness or negative social evaluation) and chronic stress (e.g., the stress of poverty or ongoing family conflict) (Adam, 2012). Second, the development and adult functioning of the HPA axis is profoundly influenced by prior developmental experience (Lupien et al., 2009). Third, cortisol has pervasive effects throughout the body and brain, and is thought to play important roles in daily cognitive and behavioral functioning (Lupien et al., 2009). Fourth, cortisol has also been implicated in the etiology of a wide range of mental and physical health outcomes (Chrousos and Gold, 1992). As a result, researchers have suggested that stress-related alterations in cortisol regulation may play a role in mediating associations between stress exposure and later developmental and health outcomes (Lupien et al., 2009, Davis and Sandman, 2010), including both the onset and progression of mental and physical health disorders (Heim et al., 2008).
Past research on cortisol and health has focused on cortisol reactivity to acute stress (Granger et al., 1996, Heim et al., 2008) as well as variations in average basal cortisol levels (Chrousos and Gold, 1992). More recently, researchers have appreciated the importance of circadian variability in cortisol levels, by examining influences on, and consequences of, individual differences in the diurnal (daytime) cortisol rhythm. The current meta-analysis examined associations between one aspect of the diurnal cortisol rhythm – the diurnal cortisol slope (DCS) – and mental and physical health outcomes.
Cortisol levels typically follow a strong diurnal rhythm: levels are high on waking, surge an average of 50–60% in the 30–40 min after waking, drop rapidly in subsequent few hours after the awakening surge and then drop more slowly until reaching a nadir around bedtime (Pruessner et al., 1997, Adam and Kumari, 2009). Variation in cortisol levels as a function of time of day is substantial. In one study, time of day accounted for 72% of the variance in salivary cortisol levels (Adam, 2006). Early research often considered this time-of-day variation to be “nuisance” variation. Over the past 15 years, however, individual differences in the diurnal cortisol rhythm have emerged as a construct of interest (Adam et al., 2008). Researchers have examined the genetic, developmental, and psychosocial determinants of individual differences in the diurnal cortisol rhythm (Adam, 2012), as well as the potential health consequences of variation in the diurnal cortisol rhythm (Sephton et al., 2000).
The diurnal cortisol rhythm has been divided into several key components which provide complementary information. Most often examined are: the average level of cortisol across the day (daily average cortisol or DAC); the size of the post-awakening surge, called the cortisol awakening response (CAR); and the diurnal cortisol slope (DCS), the degree of change in cortisol from morning to evening over the waking day (Adam and Kumari, 2009).
Early research on diurnal cortisol rhythms generally focused on DAC (Yehuda et al., 1990, Gunnar et al., 2001, Nicolson, 2004). Since its discovery in the late 90′s (Pruessner et al., 1997), the CAR has also received extensive research attention, with reviews and meta-analyses examining the determinants and consequences of the CAR (Clow et al., 2004; Chida and Steptoe, 2009, Fries et al., 2009, Clow et al., 2010).
An accumulating body of research focusing on the DCS suggests that it is sensitive to emotional and psychosocial stress (Adam and Gunnar, 2001, Adam et al., 2006, Doane and Adam, 2010) and related to health outcomes (Sephton et al., 2000, Matthews et al., 2006, Kumari et al., 2009, Doane et al., 2013), with both adverse experience and worse health being associated with a flatter DCS across the waking day. It has therefore been proposed that a flattened DCS may be one mechanism by which stress influences negative health outcomes (Sephton and Spiegel, 2003, Adam and Kumari, 2009).
Cortisol has important regulatory effects throughout the body and brain, impacting arousal, energy and metabolic processes, immune and inflammatory system functioning, and mood and sexual behavior (Sapolsky et al., 2000). Cortisol’s diurnal variation may be an important element of its regulatory actions; indeed, cortisol is one pathway by which central circadian rhythms are signaled to multiple peripheral biological systems (Bass and Lazar, 2016, Man et al., 2016). We argue here that disruption of cortisol’s circadian pattern and signaling may affect the functioning of a diverse set of central and peripheral systems, with these effects cascading over time to contribute to a wide variety of negative health outcomes.
For example, prior studies have found associations between flatter cortisol slopes and depression (Doane et al., 2013), fatigue (Bower et al., 2005, Kumari et al., 2009), cardiovascular disease (Matthews et al., 2006), and mortality among both breast cancer patients and in community samples (Sephton et al., 2000, Kumari et al., 2011). Findings have, however, been inconsistent, and researchers have not systematically summarized the existing research, or fully explicated the meaning of the DCS or the potential mechanisms by which it may be related to mental and physical health outcomes. Since the early 2000′s (Gunnar and Vazquez, 2001), no systematic reviews on the DCS have been conducted. Moreover, no meta-analyses have been conducted either on the effects of psychosocial experience on the DCS or on its associations with health outcomes. The current manuscript addresses the latter question, with an eye to better understanding: a) what is the association between DCS and health (in particular, the average magnitude and direction of the association as well as its consistency across studies), b) whether the DCS relates to certain types of health outcomes more strongly than to others, c) the meaning of the DCS and the mechanisms by which it may relate to health outcomes, and d) how methodological variations in study design and DCS measurement may contribute to variations in study effect sizes.
Researchers have referred to the DCS in a wide variety of ways, including diurnal cortisol slopes (Adam and Kumari, 2009), diurnal cortisol declines (Cohen et al., 2006), diurnal cortisol variability (Sannes et al., 2013), diurnal cortisol rhythms (Bower et al., 2005), and the amplitude of the circadian cortisol rhythm (Goel et al., 2009). Likewise, researchers have quantified the DCS in different ways, which vary in the number and timing of the cortisol samples across the waking day, and in approaches to calculating slope measures from those samples. For the purposes of this review, any measure that provides an indication of the magnitude of the difference between morning and evening cortisol values is considered a measure of the DCS.
Common types of slopes include: 1) wake-to-bed slopes, which examine the absolute change or rate of change in cortisol from immediately upon waking to late evening or bedtime (e.g., Adam et al., 2010, Turner-Cobb et al., 2011); 2) peak-to-bed slopes, which examine the absolute change or rate of change in cortisol from the peak of the CAR to late evening or bedtime, (e.g., Hsiao et al., 2010, Vammen et al., 2014); 3) short daytime slopes, which measure slopes over a shorter portion of the waking day, typically from several hours after waking to evening or bedtime (late decline measures are one example of this; see Hajat et al., 2013); 4) fixed time point slopes (e.g., Bosch et al., 2007, den Hartog et al., 2003), in which samples are gathered at fixed clock times across the day (e.g., 0800 h and 2000 h), rather than in relation to time of waking; and 5) amplitude measures, which estimate the peak-to-trough difference of the diurnal cortisol rhythm from intensive repeated measures of cortisol values across the day (e.g., Bao et al., 2004; Fidan et al., 2013a, Fidan et al., 2013b).
For the first three cortisol slope types (i.e., wake-to-bed, peak-to-bed, and short daytime slopes), samples are timed relative to each individual’s sleep-wake schedule, or more specifically, relative to person- and day-specific time of waking. These slopes typically are quantified in one of three ways: a) taking a simple difference between the morning measure and the evening measure; b) taking a simple difference divided by the total time between the two samples; or c) using regression or multilevel growth curve modeling to predict cortisol levels across the day from time of day of measurement for each individual, with the slope obtained from the size of the person-specific beta coefficient for the effect of time of day on cortisol (e.g., Adam et al., 2006, Doane et al., 2013). Fixed-time point slopes are typically calculated using either a simple difference score from morning to evening cortisol levels or with a repeated-measures ANOVA examining within-person changes in cortisol from one sampling point to the next. Amplitudes are measured using the cosinor method, which fits a cosine curve to the repeated measures cortisol data and then calculates the characteristics of the curve, including its amplitude.
Although these different types of slope measures allow researchers to measure DCS across a variety of study designs, this heterogeneity has the potential to obscure associations between DCS and health outcomes. One important debate is whether DCS should be calculated from the peak of the CAR value to evening values or from waking values to evening values, excluding the CAR. Researchers have argued for excluding the CAR from DCS measures because the CAR is influenced by different biological mechanisms than the rest of the diurnal cortisol rhythm (Clow et al., 2010, Adam et al., 2015). Although current recommendations suggest that measures should be gathered relative to individual wake times (e.g., Adam and Kumari, 2009), rather than at fixed clock time points, studies have not examined the implications of this choice. The current meta-analysis, through examining type of slope measure as a moderator of the associations between DCS and health outcomes, provides important insights into these and other measurement debates.
Beyond type of slope, other study design factors that may have implications for the size of the association found between DCS and health include the number of samples across the day used to define the slope (Hoyt et al., 2016), the number of days of salivary cortisol data collection (Adam and Kumari, 2009), whether key health behavior confounds are covaried (Adam and Kumari, 2009), and the presence of objective measurement of compliance with sampling times (Kudielka et al., 2003). In addition, given developmental changes in the HPA axis (Gunnar et al., 2009b), the age or developmental stage of participants is another factor that should be considered as a moderator of associations between DCS and health.
The primary goal of the current study was to provide a meta-analysis of the literature (up until January 31, 2015) assessing the associations between diurnal cortisol slopes and health outcomes. Specifically, we examined the associations between DCS and 12 subtypes of mental and physical health outcomes, namely: 1) anxiety symptoms or disorders; 2) depression symptoms or disorders (excluding bipolar depression); 3) internalizing disorders (symptom scales reflecting a mixture of anxiety and depression symptoms); 4) externalizing symptoms or disorders (a spectrum of behaviors involving anger expression, aggression and delinquency); 5) fatigue symptoms or disorders; 6) immune or inflammatory markers or disorders; 7) obesity (including measures of body mass index or BMI, obesity, and adiposity); 8) cardiovascular disease symptoms and diagnoses; 9) cancer disease status or progression; 10) other mental health outcomes (mental health symptoms or disorders not classified as one of the above disease subtypes); 11) other physical health outcomes (physical health symptoms or disorders not classified as one of the above subtypes); and 12) mortality (death from any cause). While it is challenging to capture the existing literature on DCS and these varied health outcomes in a single meta-analysis, this comprehensive approach allows comparisons of relative effect sizes across different types of health outcomes. Through shedding light on the types of health symptoms and disorders most strongly associated with flatter cortisol slopes, this analysis may provide insights into the key biological pathways linking flattened cortisol slopes to multiple indices of poor health.
A secondary goal was to test whether the size of associations between DCS and health outcomes would be moderated by the following factors: age of participants, type of slope measure, number of cortisol samples measured per day, number of days of data collection, and a study quality index based on the number of relevant confounds measured and accounted for in design and/or analysis.
We hypothesized that a flatter DCS would be associated with worse health, across a range of health outcomes. We did not have strong hypotheses regarding which specific types of health outcomes would be most strongly associated with flatter cortisol slopes, although we expected immune and inflammatory outcomes to show robust associations, given the key role played by glucocorticoids in regulating inflammation (Silverman and Sternberg, 2012). We expected that wake-to-bed cortisol slopes would show stronger associations with health than other types of slope measures, that studies with more samples per day and more days of measurement would show stronger associations, and that studies utilizing objective monitoring of sampling compliance would reveal stronger associations than studies not utilizing objective monitoring. We expected to see DCS-health associations across multiple age groups, with effects potentially being larger in older age groups due to longer histories of stress exposure or more advanced disease processes.
Section snippets
Data sources and searches
Under the direction and guidance of the first author, doctoral students and postdoctoral fellows conducted electronic searches between April 2013 and January 2015. Electronic searches were done in Medline and Web of Science (both via Endnote X4 (2010) program search tool), PubMed, Psych Info, and Social Science Abstracts (via website or EbscoHost). Search terms for each database were: “cortisol rhythm”, “cortisol rhythms”, “cortisol slope”, “cortisol slopes”, “cortisol diurnal slope”, “cortisol
Study characteristics and quality
Fig. 1 shows the details of the PRISMA flow diagram for this systematic review and meta-analysis (Moher et al., 2009). Table 1 details the studies and findings from each study included in this analysis. It also includes the coded characteristics, sample size, and effect size of each included finding.
A total of 36,823 participants (26,167 unique individuals, when overlap in samples across studies is considered) from 80 studies were included in this meta-analysis. From the 80 studies, 179
Overview of primary results
Results of this systematic review and meta-analysis provide evidence to support prior assertions that flatter diurnal cortisol rhythms across the day are associated with poorer mental and physical health outcomes (Adam and Kumari, 2009). Notably, effects were both significant and in the predicted direction (flatter slopes associated with worse health outcomes) for 10 out of the 12 physical and mental health outcomes assessed. The significant outcomes included depression, internalizing
Conflicts of interest
The authors have no conflicts of interest to report. The authors’ time on this project was supported, in part, by Faculty and Graduate Fellowships from the Institute for Policy Research at Northwestern University and T32 MH100019-03. The funders played no role in the conduct of this research, the analyses, interpretations or conclusions.
References1 (157)
- et al.
Flattened cortisol rhythms in metastatic breast cancer patients
Psychoneuroendocrinology
(2004) - et al.
Relationship functioning and home and work demands predict individual differences in diurnal cortisol patterns in women
Psychoneuroendocrinology
(2001) - et al.
Assessing salivary cortisol in large-scale, epidemiological research
Psychoneuroendocrinology
(2009) - et al.
Prospective prediction of major depressive disorder from cortisol awakening responses in adolescence
Psychoneuroendocrinology
(2010) - et al.
Developmental histories of perceived racial discrimination and diurnal cortisol profiles in adulthood: a 20-year prospective study
Psychoneuroendocrinology
(2015) Transactions among adolescent trait and state emotion and diurnal and momentary cortisol activity in naturalistic settings
Psychoneuroendocrinology
(2006)- et al.
Immunity’s fourth dimension: approaching the circadian–immune connection
Trends Immunol.
(2012) - et al.
Diurnal rhythms of free estradiol and cortisol during the normal menstrual cycle in women with major depression
Horm. Behav.
(2004) - et al.
Depression is associated with flatter cortisol rhythms in patients with coronary artery disease
J. Psychosom. Res.
(2008) - et al.
Diurnal cortisol rhythm and fatigue in breast cancer survivors
Psychoneuroendocrinology
(2005)
Associations of ghrelin with eating behaviors, stress, metabolic factors, and telomere length among overweight and obese women: preliminary evidence of attenuated ghrelin effects in obesity?
Appetite
Cortisol awakening response and psychosocial factors: a systematic review and meta-analysis
Biol. Psychol.
Self-reported trauma, cortisol levels, and aggression in psychopathic and non-psychopathic prison inmates
Biol. Psychol.
The cortisol awakening response: more than a measure of HPA axis function
Neurosci. Biobehav. Rev.
24-Hour cortisol measures in adolescents with major depression: a controlled study
Biol. Psychiatry
Salivary cortisol patterns and cognitive speed in major depression: a comparison with allergic rhinitis and healthy control subjects
Biol. Psychol.
Cortisol secretion in depressed, and at-risk adults
Psychoneuroendocrinology
A day-centered approach to modeling cortisol: diurnal cortisol profiles and their associations among U. S. adults
Psychoneuroendocrinology
Loneliness and cortisol: momentary, day-to-day, and trait associations
Psychoneuroendocrinology
Negative emotionality, depressive symptoms and cortisol diurnal rhythms: analysis of a community sample of middle-aged males
Horm. Behav.
Sputum interleukin-6, tumor necrosis factor-alpha and salivary cortisol as new biomarkers of depression in lung cancer patients
Prog Neuro-Psychoph
No PTSD-related differences in diurnal cortisol profiles of genocide survivors
Psychoneuroendocrinology
Stress may add bite to appetite in women: a laboratory study of stress-induced cortisol and eating behavior
Psychoneuroendocrinology
Health anxiety moderates the daytime cortisol slope
J. Psychosom. Res.
Variance of melatonin and cortisol rhythm in patients with allergic rhinitis
Am. J. Otolaryngol.
Melatonin and cortisol rhythm in patients with extensive nasal polyposis
Am. J. Otolaryngol.
The cortisol awakening response (CAR): facts and future directions
Int. J. Psychophysiol.
Ethnicity, stress, and cortisol function in hispanic and non-hispanic white women: a preliminary study of family dementia caregivers and noncaregivers
Am. J. Geriatr. Psychiatry
Stressor paradigms in developmental studies: what does and does not work to produce mean increases in salivary cortisol
Psychoneuroendocrinology
Examining the association between salivary cortisol levels and subclinical measures of atherosclerosis: the multi-ethnic study of atherosclerosis
Psychoneuroendocrinology
The link between childhood trauma and depression: insights from HPA axis studies in humans
Psychoneuroendocrinology
The associations between diurnal cortisol patterns, self-perceived social support, and sleep behavior in Chinese breast cancer patients
Psychoneuroendocrinology
The self-perceived symptom distress and health-related conditions associated with morning to evening diurnal cortisol patterns in outpatients with major depressive disorder
Psychoneuroendocrinology
Co-occurring manic symptomatology influences HPA axis alterations in depression
J. Psychiatr. Res.
Dysregulated diurnal cortisol pattern is associated with glucocorticoid resistance in women with major depressive disorder
Biol. Psychol.
Cortisol secretion and fatigue: associations in a community based cohort
Psychoneuroendocrinology
Sleep deprivation and circadian disruption: stress, allostasis, and allostatic load
Sleep Med. Clin.
Interactions of the circadian CLOCK system and the HPA axis
ABBV Trends Endocrinol. Metab.
Day-to-day dynamics of experience-cortisol associations in a population-based sample of older adults
Proc. Natl. Acad. Sci.
Social regulation of the adrenocortical response to stress in infants, children and adolescents: implications for psychopathology and education
Incorporating hypothalamic–pituitary–adrenal axis measures into preventive interventions for adolescent depression: are we there yet?
Dev. Psychopathol.
Emotion-cortisol transactions occur over multiple time scales in development: implications for research on emotion and the development of emotional disorders
Adult attention-deficit hyperactivity disorder is associated with alterations in circadian rhythms at the behavioural, endocrine and molecular levels
Mol Psychiatr
Circadian time signatures of fitness and disease
Science
Subgroup analyses
Introduction to Meta-analysis
Comprehensive Meta-Analysis Version 3
Depressive symptoms predict mucosal wound healing
Psychosom. Med.
Cortisol awakening response and subsequent depression: prospective longitudinal study
Br. J. Psychiatry
TNF-α suppresses the expression of clock genes by interfering with E-box-mediated transcription
Proc. Natl. Acad. Sci.
Crosstalk between the circadian clock circuitry and the immune system
Chronobiol. Int.
Cited by (534)
Infant colic and HPA axis development across childhood
2024, PsychoneuroendocrinologyRelationship of sleep with diurnal cortisol rhythm considering sleep measurement and cortisol sampling schemes
2024, PsychoneuroendocrinologyHigher morning cortisol is associated with lower intuitive eating in midlife women
2024, PsychoneuroendocrinologyCortisol's diurnal rhythm indexes the neurobiological impact of child adversity in adolescence
2024, Biological Psychology
- 1
“*” in references denotes—studies included in the meta-analysis are indicated with a star symbol.