State variation in the cortisol awakening response

Abstract

The cortisol awakening response (CAR) is a much studied but poorly understood aspect of the circadian pattern of cortisol secretion. A Scopus search of “cortisol” and “awakening” reveals 666 publications in this area since 1997 when it was first identified by Pruessner and colleagues as a “reliable biomarker of adrenocortical activity”. The primary focus of the majority of these studies is centered on its utility as a biomarker associated with a range of psychosocial, physical and mental health variables. Such studies typically examine differences in the CAR (studied on 1 or 2 days) between healthy participants and other comparator groups of interest. Fewer studies (25 in our estimation) have examined correlates of day-to-day variation in the CAR in healthy participants, informing its role and regulation within the healthy circadian pattern of cortisol secretion. This is the first review to examine these studies which, although limited in number, offer a relatively coherent emerging story about state factors that influence the CAR and the impact of the CAR on daily functioning. Greater understanding of these issues helps illuminate the utility of the CAR as a promising biomarker in psychophysiological and epidemiological research. The review also highlights areas that require greater clarification and points to potentially fruitful areas of further research.

• Awakening
• CAR
• cortisol
• saliva
• state variation
• trait variation

Introduction

The cortisol awakening response (CAR) is the rapid increase in cortisol concentrations within the first hour after awakening from sleep (Clow et al., 2004; Pruessner et al., 1997). Cortisol concentrations typically increase between 50% and 160% during this hour, with a peak in the CAR typically between 30- and 45-min post-awakening (Clow et al., 2004). Despite a large number of studies the role of the CAR remains poorly understood. Within the circadian rhythm of cortisol secretion the CAR is a relatively discrete aspect (Clow et al., 2004, 2010a), initiated in response to morning awakening (Wilhelm et al., 2007). It is characterized by marked inter-individual variability, and CAR research to date has primarily focused on its application as a biomarker in association with a range of psychosocial and physical variables including health outcomes (Chida & Steptoe, 2009). However, there is accumulating additional evidence describing marked intra-individual state variation in the CAR (Hellhammer et al., 2007). Understanding such variation within healthy populations should shed light on the potential role of the CAR within the healthy cortisol circadian cycle. In addition, study of the “healthy” CAR may illuminate how exposure to certain state factors at sensitive stages of the life course could affect the HPA axis and become embedded trait characteristics of the individual (Lupien et al., 2009). In addition, increased understanding of CAR state variation should be informative for interpretation of the broad range of results from studies using the CAR as a trait biomarker (Chida & Steptoe, 2009). It is the aim of this review to highlight and discuss the evidence from the 25 published studies that have explored intra-individual variation in the CAR within healthy participants. The evidence provided from these studies has the potential to lead toward a clearer understanding of the role of the CAR in healthy functioning and the consequences of dysregulation in a range of conditions.

The CAR can be measured in various ways: e.g. the “area under the curve with respect to increase” (AUCi: s2 + s3 + [(s4 − s1)/2] − 2s1) (see Edwards et al., 2001; Pruessner et al., 2003; Smyth et al., in press b) for a discussion of this calculation); the “mean increase” (MnInc: (s2 + s3 + s4)/3 − s1) (see Wuest et al., 2000) or a simple delta score, e.g. cortisol concentration at 30 min (or peak concentration) minus that on awakening (see Steptoe & Usher, 2006). Over the years, the CAR has also been assessed in relation to absolute levels of cortisol secretion post-awakening using measures such as the AUCg: s1 + s2 + s3 + [(s4 − s1)/2] (see Pruessner et al., 2003). Although this could be an interesting measure, interpretation is limited as it gives no indication of the dynamic increase post-awakening, which is the defining feature of the CAR. It has been recommended (Clow et al., 2010b) that the best practice is to report the first waking sample plus the dynamic increase, from which the total secretion can be deduced.

Background: use of the CAR as a trait biomarker

The first report of the CAR, by Pruessner et al. (1997), suggested relative intra-individual stability in a sample of both male and female participants across a broad age range (see Table 1 for a summary of the study details). In this initial study, it was reported that the CAR was not affected by participants’ time of awakening, menstrual cycle phase or alcohol consumption in any of the three groups. This finding was supported by several further demonstrations of intra-individual stability (e.g. Edwards et al., 2001; Wuest et al., 2000). These results proved to be particularly influential in recommending the use of the CAR (measured across 1 to 2 days only) as a reliable trait biomarker for HPA-axis status. Subsequently, a series of studies emerged in which contradictory outcomes were observed when using the CAR as a trait measure. At the same time, further within-subjects studies of the CAR demonstrated state differences across sampling days, bringing the reliability of the CAR as a trait biomarker into question. One of the earliest state associations to emerge was the weekend–weekday difference (Kunz-Ebrecht et al., 2004; Schlotz et al., 2004) (see Table 1 for details). These early studies also provided an indication that variation in the CAR may be associated with the individual’s psychological state and anticipation of the day ahead.

Table 1. Summary of studies that have examined state variation of the CAR in healthy participants.

Author(s)	Participants	HPA measurement	Findings
Pruessner et al. (1997)	Healthy. N = 152 (1:1 male–female), 11, 26 and 70 (3 groups).	Group 1 – 0, 10, 20, 30 on three consecutive days. Group 2 – 0, 15, 30, 60 on 3 days at 1-week intervals. Group 3 – 0, 15, 30 and 60 on two consecutive days	CAR in all three studies reported as stable within subjects. Sleep duration, time of awakening and alcohol also appeared unrelated to the CAR.
Scheer & Buijs (1999)	Healthy males. N = 14, ages 24–41 years (Mean 32.5 years)	0, 20, 40, 60 and 120 min	Bright light (800 lux) for the hour following awakening produced a significant increase in CAR magnitude at 20- and 40-min post-awakening.
Wuest et al. (2000)	Healthy. N = 509 (319 female, 190 male), ages 18–71 years (Mean 37 years)	0, 30, 45 and 60 min on two consecutive days	Observed a MnInc of ∼50% within the first 30 minutes. Intraindividual stability was shown to be high with correlations up to r = 0.63.
Edwards et al. (2001)	Healthy. N = 40 (31 female, 9 male), ages 23–53 years (Mean age 35 years)	0, 15, 30 and 45 min on two consecutive days	Moderate stability of the CAR across two sampling days, with early waking participants showing a larger CAR.
Federenko et al. (2004)	Healthy shift workers. N = 86, 55 female (24 nurses and 31 students). Nurses ages 29–52 years (Mean 40.3 years). Students ages 20–31 years (Mean 25 years)	0, 30, 45 and 60 min on two consecutive days	In the sample of nurses, the CAR was larger in the early shift than in the late and night shifts.
Kunz-Ebrecht et al. (2004)	Healthy. N = 196, ages 47–59 years	0 and 30 min, on 2 days	The cortisol awakening response (defined as the difference between waking and 30 min later) was greater on work than weekend days.
Schlotz et al. (2004)	Healthy. N = 219, ages 24–83 years (Mean 48.6 years)	0, 30, 45 and 60 min on six consecutive days starting on Saturday	Clear weekend–weekday difference in the CAR, and this difference was associated with chronic work overload and worry. This was independent of sex weekend–weekday differences in time of awakening and sleep duration.
Thorn et al. (2004)	Healthy. N = 12 (5 female, 7 male), ages 24–54 years (Mean age 39 years)	0, 15, 30 and 45 min, on four consecutive days	Increased AUCg magnitude over 45 min when participants were exposed to ∼250 lux light for 30 min on awakening.
Williams et al. (2005)	Healthy shift workers. N = 32 (17 male, 15 female) no age data	2, 30 and 60 min on 6 days, including two early shift days, 2-day shift days, and two control/leisure days	Observed a greater CAR on early-shift days. However, respondents were more stressed over the hour after waking and reported more sleep disturbance on early-shift days; when these factors were taken into account, the CAR-early shift relationship was no longer significant.
Adam et al. (2006)	Healthy older adults. N = 156, ages 50–68 years (mean 57 years)	0 min, 30 min and at bedtime each day for 3 days	Prior-day feelings of loneliness, sadness, threat and lack of control were associated with a higher cortisol awakening response the next day, but morning awakening responses did not predict experiences of these states later the same day.
Clow et al. (2006)	Healthy military sample. N = 20 ages 17–24 years	0, 15, 30 min on 5 days throughout a military training course	Reduction in CAR (AUC) during the middle of an intense training course, and a return to baseline CAR following completion.
Thorn et al. (2006)	Healthy students. N = 48 (4:1 female to male), ages 18–36 years (Mean 20 years)	0, 15, 30 and 45 min on two consecutive weekdays, and two consecutive weekend days	Steeper rise in the CAR on weekdays, not accounted for by differences in waking time, state stress or perceived stress over the previous month. AUC over 4 days also negatively correlated with the measure of longer term stress and awakening time. Mean increase also negatively correlated with waking time, not mediated by stress or vice versa, since both independently predicted cortisol.
Hellhammer et al. (2007)	Healthy. N = 193, ages 28–88 years	0, 30, 45 and 60 min on six consecutive days	Concluded that the CAR of a single day is largely determined by situational factors, and only a small proportion is determined by trait factors.
Wilhelm et al. (2007)	Healthy males. N = 16, ages 20–33 years (Mean 25 years)	Saliva at 0, 30, 45 and 60 min. Blood sampling every 15 min	Demonstrated that the CAR is a response to waking, and did not differ in lab conditions from that observed at home.
Dettenborn et al. (2007)	Healthy females. N = 13, ages 20–32 years (Mean 24 years)	0 and 15 min following three night awakenings, and again at 0 and 15 min following morning awakening on three nights. Also taken at 0 and 15 after three undisturbed nights	Significant difference between night time and morning CAR. No CAR after waking during the first half of the night, but some reactivity during the early hours of the morning, and pronounced CAR in the morning before getting out of bed.
Almeida et al. (2009)	Healthy. N = 1143, ages 33–84 years (Mean 57 years)	0 min, 30 min, before lunch and at bedtime for 4 days	Substantial day-to-day variability in the CAR. State factors reported to account for 78% of CAR variation.
Stalder et al. (2009)	Healthy male. N = 1 (Case study), age 27 years	0, 15, 30 and 45 min on 50 measurement days, at 3-day intervals	Considerable day-to-day variability in the CAR. Waking time also strongly related to first waking sample (stronger relationship on days when awakening time was earlier). Preliminary indication of inverse association between alcohol consumption.
Doane & Adam (2010)	Healthy. N = 108 (27 male), mean age 19 years	0 and 30 min	Prior day feelings of loneliness were associated with an increased CAR. Prior day feelings of nervousness/stress were associated with lower average wake-up levels of cortisol.
Doane et al. (2010)	Healthy males. N = 786, middle-aged, no specific age data given on full sample	0 and 30 min, then 10 am	Evidence that travelling across time zones is associated with diurnal cortisol regulation. Reduced peak of post-travel CAR following Eastward or Westward travel, and also a steeper decline after travelling Eastward.
Mikolajczak et al. (2010)	Healthy males. N = 42, ages 30–50 years (Mean 38 years)	0, 15, 30, 45 and 60 min on three consecutive measurement days (Sunday, Monday, Tuesday)	Protective factors (i.e. high happiness, low stress and low neuroticism) are associated with a flexible CAR (i.e. greater difference between Sunday and Monday CARs).
Stalder et al. (2010b)	Healthy male. N = 1 (Case study), age 27 years	0, 15, 30 and 45 min on 50 measurement days, at 3-day intervals	Relationships found between psychosocial state variables and the AUCI, including an inverse relationship with the level of prior-day happiness and a positive relationship with study-day anticipations of the level of obligations/no leisure.
Stalder et al. (2010a)	Healthy females. N = 12, ages: 22–41 years (Mean 29 years)	0, 15, 30 and 45 min, on 12 days at 3-day intervals	CAR variability (AUCI) was found to be inversely related to simultaneous variability in awakening time, and positively related to variability in adverse psychosocial states of stress and tension 45-min post-awakening.
Strahler et al. (2010)	Healthy (martial arts practitioners). N = 20, ages 16–30 years (Mean 23 years)	0, 15, 30, 45, 60 min on 3 days prior to competition	No change observed in the CAR in the build up to competition.
Thorn et al. (2011)	Self-reported SAD and control (healthy) sample. N = 52 (26 SAD sample, 26 Control), ages 26–75 years (mean 50 years)	0, 15, 30, 45 min on 2 days in summer, and 2 days in winter	In winter the CAR was significantly attenuated in SAD participants in comparison to the healthy control participants, but no difference in summer.
Wolfram et al. (2011)	Healthy females. N = 29, ages 20–34 years (Mean 26.3 years)	0, 30, 45 and 60 min on four measurement days, representing four phases across the complete menstrual cycle	The CAR is elevated during ovulation. But no differences in mood states across the four cycle phases.

It has since been established that the CAR is prone to significant state variation which is greater than trait variation (Almeida et al., 2009; Hellhammer et al., 2007; Stalder et al., 2009, 2010a) (see Table 1 for a summary of studies). Such studies have been generally consistent, with Hellhammer et al. (2007) reporting that state factors can account for between 61% and 82% of the variation in the CAR. Almeida et al. (2009) suggested that these factors can account for 78% of CAR variation, and Stalder et al. (2010a) suggested a figure of 64% for a smaller sample of female participants. The first longitudinal case study of the CAR in a healthy participant (Stalder et al., 2009) employed a researcher–participant design to avoid participant non-adherence to the protocol. Again, this approach provided clear evidence of state variation in the CAR, which ranged from 3.6 to 39.0 nmol/l on any one day. Indeed, recent research by Mikolajczak et al. (2010) (see Table 1 for details) found that factors known to be protective of health such as high happiness, low stress and low neuroticism, were found to be associated not with the size of the CAR but rather greater day-to-day variation in the CAR. This hinted that CAR flexibility (e.g. a greater difference between days) rather than CAR magnitude may be a better way to understand the relationship between the CAR and positive psychosocial status.

Intra-individual studies in healthy participants have provided evidence (sometimes from a single study) that the CAR can be associated with a range of situational and psychosocial state variables. These include time of awakening, ambient light, prior day experiences, anticipation of the day ahead, ovulation, jet lag and alcohol consumption (Adam et al., 2006; Doane et al., 2010; Edwards et al., 2001; Stalder et al., 2009, 2010b; Wolfram et al., 2011). These reports provide clues about the role of the CAR in healthy functioning which can hopefully can be replicated and extended.

Factors associated with state variation in the CAR

Light and the regulatory influence of the suprachiasmatic nucleus

The first report of a relationship between ambient light and state variation in the CAR was provided by Scheer & Buijs (1999), with the demonstration from a repeated measures study design that awakening in darkness (induced by blindfold) leads to an attenuated CAR AUCg compared to awakening in moderate ambient light (800 lux). These effects of light on the CAR AUCg have since been replicated with relatively low light levels (250 lux). In this repeated measures study exposure to dawn simulation in winter (i.e. increasing light intensity to a maximum of 250 lux in the 30 min immediately prior to awakening) resulted in both significantly higher cortisol levels for participants during the first 45-min post-awakening and an associated significant increase in self-reported levels of arousal (Thorn et al., 2004). Surprisingly, only these two studies have directly examined the impact of light on the CAR, with both studies suggesting an influence of ambient light on the CAR. In so doing they implicate a role for the hypothalamic suprachiasmatic nucleus (SCN) in regulation of the CAR, as previous studies in rodents have demonstrated the dependence of the SCN for the effects of light on basal cortisol levels (Buijs et al., 1999; see Clow et al., 2010a for a more detailed discussion of this issue).

In self-assessed seasonal affective disorder (SAD) participants have been found to show a significant decrease in the CAR in winter when waking before sunrise in comparison to both their CAR in the summer, and the CAR of non-SAD participants. Furthermore, a general dysphoria construct correlated inversely with the CAR in the winter, indicating that participants reporting greater depression, stress and anxiety and lower arousal exhibited lower CARs. It was noteworthy that participants who lacked marked seasonality in terms of affect did not show seasonal variation in the CAR (Thorn et al., 2011) (see Table 1 for study details). These data suggest that the effects of light on the CAR may differ between individuals. In separate studies it has been demonstrated that individuals who report symptoms of SAD show reduced retinal sensitivity to light during the winter months in comparison to healthy controls (Hebert et al., 2002; Terman & Terman, 1999). It is possible (although not yet tested) that these two factors (retinal sensitivity and the CAR) are linked via light detection-SCN-neuroendocrine pathways, an intriguing possibility that deserves direct investigation.

Contrary to these findings, it has also been reported (again in a repeated measures study design) that awakening in darkness in a sleep laboratory environment was associated with the same size of CAR as awakening in the domestic setting with typical ambient light (Wilhelm et al., 2007). However, as the authors acknowledge this study was not designed to study the effect of light on the CAR as there was no standardization of other potentially confounding state factors. For example, it is possible that awakening in the dark in a strange environment with experimenters present might impact on the CAR, since it has been demonstrated that anticipation of the day ahead can influence the magnitude of the CAR (Stalder et al., 2010b).

In summary, it is possible, and deserves further investigation, that there is not only variation in the CAR in response to ambient light but also individual differences in that variation, potentially dependent on retinal sensitivity. This added complexity suggests that although it is important to take ambient light into account when using the CAR as a biomarker it is not usually possible to make allowances for individual differences in sensitivity to that light. Notwithstanding this limitation, greater variation in pre- and post-awakening light exposure between participants increases the potential for a confounding influence when comparing CAR data across individuals. Indeed, it has been demonstrated that variation between light levels of 0–800 lux can account for a not insubstantial ±35% of state variation in the CAR (Scheer & Buijs, 1999). It is suggested here that more work be undertaken to replicate and extend the small number of studies in this area but that in the meantime it would be wise if ambient light levels were kept stable when comparing the CAR between participants (e.g. <50 lux difference between participants), or alternatively ambient light should be recorded and accounted for in the data analysis. Controlling for this factor is unfortunately not considered in much of the trait CAR research.

It is possible that attenuated CAR magnitude when awakening in the dark may prevent unnecessary sleep disturbance in response to accidental nocturnal awakening. If the CAR does assist in the re-establishment of cognitive function (as proposed by Fries et al., 2009) then in theory, light-responsive systems may allow for optimally enhanced cognitive function when awakening at dawn, in preparation for the day ahead. Indeed, preliminary support for this theory is provided by studies using functional magnetic resonance imaging techniques which indicate that exposure to light during the morning period results in increased activation of the subcortical regions associated with alertness and working memory (Vandewalle et al., 2006, 2010). To date, however, there have been no studies which have explored the effects of light on the CAR in combination with the activation of these brain regions. It would be interesting to undertake such studies in order to help establish the underlying mechanisms which may regulate the transition from consciousness to alertness upon awakening. For example, it is well established that the SCN utilizes a range of signaling methods, including the neuroendocrine system, to entrain the peripheral clocks in different regions of the body including the brain (Menet & Rosbash, 2011). Of particular relevance here is that signaling from the SCN is vital to the functional integrity of the hippocampus with impairments of hippocampus-dependent memory demonstrated in rats with SCN lesions or changes in the light/dark cycle (Devan et al., 2001; Ruby et al., 2008; Stephan & Kovacevic, 1978). Physical and functional integrity of the hippocampus is also related to circadian cortisol secretion and the CAR (Buchanan et al., 2004; Pruessner et al., 2007; Rimmele et al., 2010; Sapolsky, 2001; Wolf et al., 2005). Further research could investigate whether one of the functions of the CAR may be to signify the start of the awakening phase and maximize optimal capacity for appropriate daytime activity.

Sleep variables

Although not yet fully understood, there is evidence for between and within participant associations between the time of awakening and the magnitude of the CAR, first demonstrated between participants by Edwards et al. (2001) and later confirmed by several further studies within participants (e.g. Almeida et al., 2009; Dettenborn et al., 2007; Federenko et al., 2004; Stalder et al., 2009, 2010a, Zeiders et al., 2011). Later awakening has been associated with higher cortisol levels immediately upon awakening and in multiple regression analyses this factor was able to account for the reduced CAR (Stalder et al., 2009). However, not all studies have reported an association between time of awakening and CAR magnitude (Brooke-Wavell et al., 2002; Kunz-Ebrecht et al., 2004; Pruessner et al., 1997; Wuest et al., 2000) and this inconsistency may be attributable to a range of factors. For example, in studies where the range of awakening times is small there is less likely to be an association. There is also a potential confound in terms of participant age, as aging is associated with earlier waking and yet an attenuated CAR (Kudielka & Kirschbaum, 2003), and intra-individual variability of the CAR has also been shown to increase with age among men (Almeida et al., 2009). Another important factor in large epidemiological studies that have not found this relationship may be related to participant non-adherence (Smyth et al., in press a; Thorn et al., 2006). Many studies rely on self-report of awakening and sampling times, despite the fact that these have consistently been shown to be inaccurate (Broderick et al., 2004; Smyth et al., in press a). Notably, whilst earlier studies had suggested that delays of up to 15 min between awakening and collection of the first sample did not affect the accuracy of the CAR measurement (DeSantis et al., 2010; Dockray et al., 2008; Okun et al., 2010), more recent research has shown that delays of only 8 min do in fact lead to inaccurate assessment of the CAR (Smyth et al., in press a).

Interestingly, as observed for light, there may be other trait factors that interact with this state association. It has been reported that the relationship between time of awakening and the magnitude of the CAR is moderated by depressive symptoms, such that the relationship was reduced in mild to moderate depression (Stetler & Miller, 2005). In addition, in the case of night awakenings, the magnitude of the CAR upon forced awakening appears to increase as the night goes on, with the most pronounced CAR upon normal morning awakening (Dettenborn et al., 2007).

Evidence to demonstrate that the CAR adapts to changes in awakening times is provided by studies exploring the CAR during shift work and altered sleeping patterns (e.g. Griefahn & Robens, 2008, 2010; Harris et al., 2010; Kudielka et al., 2007; Williams et al., 2005). When individuals begin working night shifts (i.e. begin to sleep during the daylight hours) the CAR is initially blunted, but then gradually increases following consecutive night shifts until it returns to baseline magnitude (Griefahn & Robens, 2010; Harris et al., 2010; Kudielka et al., 2007). Additionally, there appears to be a sex difference in this return to baseline CAR magnitude in night shift workers, with the CAR typically reaching baseline levels after 3 days in men, and 4 days in women (Griefahn & Robens, 2010). Griefahn & Robens (2010) note that the adjustment of the CAR across consecutive night shift work occurs faster than the other aspects of the circadian system, suggesting that circadian functions are not the sole cause of the adjustment of the CAR. It has also been demonstrated that the process of cortisol circadian rhythm adaptation in night shift workers is facilitated by the presence of appropriate cues for waking. For instance, the extent to which the circadian rhythm of cortisol adapts following changes in shift work patterns is increased in individuals exposed to bright light during a night-shift and protected from light exposure during the day-sleep period (James et al., 2004). The evidence provided by these studies may therefore implicate a regulatory influence of the SCN that can be moderated by external zeitgeibers. It also provides some indication that although the CAR can adapt to waking during nocturnal hours, this adaptation may not be complete without facilitative alterations to the light/dark cycle.

The findings from shift-work studies may also illuminate the phenomenon of “jet lag” experienced by individuals travelling across several time zones and subsequently changing their sleep and waking times. Indeed, the re-adjustment of the body clock following travel is similar to that experienced in shift work, perhaps only differing in the natural external cues to waking (i.e. timing of the natural light cycle). Recent research by Doane et al. (2010) has confirmed that jet-lag when travelling across three or fewer time zones does result in an attenuated CAR, with both Eastward and Westward travel showing an association with a lower peak in the CAR on the following morning (see Table 1 for study details). This study reported an effect on the CAR (difference between awakening and 30-min samples) of nearly twice the magnitude for Eastward travel than for Westward travel. The extent to which these results can be generalized is limited as there has only been this one study to date. Additionally the authors recognized that there was an imbalance in the study conditions, with the majority of the eastward travelling sample (66%) crossing only one time zone, and the majority of the Westward travelling sample (80%) crossing two to three time zones (Doane et al., 2010). It is clear that further research is needed in order to properly explore the association between these two variables. Nonetheless, this study does provide some preliminary evidence of a relationship with travel across time zones and associated jet lag.

As previously discussed, an issue to consider in the majority of studies of altered sleep patterns and the CAR is that of participant adherence to the sampling protocol, which is known to be a significant problem in CAR research (Clow et al., 2004; Smyth et al., in press a; Thorn et al., 2006). Participants may be worse at completing the CAR sampling procedure when in a compromised situation (such as when suffering the effects of a change of sleep patterns) and the majority of research into the effects of changes to the sleep--wake cycle has relied solely upon participants’ self-reported wake times. Therefore, further research in this area should incorporate rigorous methods of measurement such as electronic recording of sleep, awakening and saliva sampling times.

Prior day experiences and anticipation of challenge

An increased CAR has been reported in response to subjective feelings of prior day threat, lack of control, loneliness and other negative feelings (Adam et al., 2006; Doane & Adam, 2010; Stalder et al., 2010a,b). This prior day association suggests that the CAR may play a preparatory (homeostatic) role in healthy functioning (Adam et al., 2006). This may be achieved by a pre-emptive increase in CAR magnitude in response to prior-day negative mood, in order to help to prepare the individual for a challenging day (Adam et al., 2006). Increased CARs have also been reported in response to anticipation of challenge or workload in the day ahead (Stalder et al., 2010a,b). Differences in anticipations of the day ahead have also been proposed as an explanation of reported weekend versus weekday differences in the CAR across the typical working week (Kunz-Ebrecht et al., 2004; Schlotz et al., 2004). Although it should be noted that Thorn et al. (2006) found the weekend/weekday difference in their data was no longer significant in students when suspected non-adherence was controlled for, suggesting that this particular association may be a product of participants failing to comply with the sampling protocol on weekends. Alternatively, it may reflect the nature of the student lifestyle in which weekday/weekend differences may not be as marked as in the typical working population.

In another study, a progressive decrease in CAR magnitude over several weeks has been demonstrated in a military sample participating in an intense training program (Clow et al., 2006). This finding is likely representative of a response to exhaustion, although this was not measured directly it would also be in agreement with trait studies of burnout and fatigue (see Adam et al., 2006; Fries et al., 2009 for further discussion). In another study, no change in the CAR was observed during the build-up to athletic competition (Strahler et al., 2010) (see Table 1 for study details). The authors noted that this was unexpected and attributed the lack of variation in the CAR to psychological and neuroendocrine habituation to competition. Nonetheless, these inconsistent findings emphasize the need for further research in this area, in order to clarify the complex relationship between the CAR, psychological anticipation and prior-day experience.

Evidence of a relationship with prior negative experience and same day anticipated challenge has led to the prominent theory that the CAR plays a preparatory (homeostatic) role in healthy functioning activating and preparing the individual for the challenges of the day ahead (Adam et al., 2006; Fries et al., 2009). The means by which the CAR may prepare the individual are unknown. However, the results of studies looking at between participant variation in the CAR show that an increased CAR is associated with lower fatigue, increased alertness and energy (Adam et al., 2006; Fries et al., 2009; Thorn et al., 2011). Whether an increased CAR also enhances cognitive or physical functioning in the day ahead are both strong possibilities. Evidence from trait studies does indicate that in older people a greater CAR magnitude is associated with enhanced overall cognitive performance (Evans et al., 2011), but it is yet to be explored whether state variation in the CAR results in state variation in cognitive function.

Menstrual cycle

Early work by Pruessner et al. (1997) indicated that the CAR is stable across the female menstrual cycle, and this was replicated by Kudielka & Kirschbaum (2003) in a study that compared females in the luteal and follicular phases of the menstrual cycle. A more recent study by Wolfram et al. (2011) again reported no difference in the CAR between the phases of the menstrual cycle, although there was significant elevation in the CAR during the 2 days of ovulation. The findings of this study showed not only an interaction between CAR magnitude and ovulation within participants, but also a later peak during ovulation (mean peak of CAR at 45 min during ovulation, as opposed to 30 min during all other phases). Although Wolfram et al. (2011) did measure mood states in this study (using the Profile of Mood States Questionnaire), no difference was observed in participants’ moods between the phases of the menstrual cycle.

Future research in this area could take into account the status of other hormones known to affect cortisol levels, such as estrogen, as well as disruption of sleep across the menstrual cycle (particularly pronounced during the luteal phase) (Shechter & Boivin, 2010). Consistent sleep disturbance may produce confounding variables in CAR studies not only by affecting participants’ mood states, but also by complicating the measurement of initial wake time. Another factor to examine in such studies is cognitive function as evidence suggests that visuo-spatial memory is enhanced during ovulation, whilst verbal fluency is decreased (Solís-Ortiz & Corsi-Cabrera, 2008). If the CAR plays a role in orientation, context recollection and general cognitive activation upon awakening, as proposed by Fries et al. (2009), then it may be that the CAR is related to this enhanced visuo-spatial memory activation during ovulation.

In summary, further examination of the CAR across the menstrual cycle, including ovulation, in relation to variables such as mood and cognitive function would be useful to inform the role of the CAR in healthy functioning as well as guide methodology of future CAR research in samples of female participants.

Alcohol consumption

A single study has reported a negative correlation between the CAR and the number of alcoholic drinks consumed during the previous evening (Stalder et al., 2009). This finding was observed in a case study of a healthy young male conducted over 50 study days (accounting for awakening time). This was interpreted as evidence of the acute effects of alcohol consumption in a healthy participant as opposed to the chronic effects of general alcohol consumption (associated with an elevated CAR in trait CAR research) (Adam et al., 2006; Badrick et al., 2008). However, the finding of an effect of alcohol on state variation in the CAR has only been confirmed in this single case study, and was observed using a very simple method of measurement which did not take account of units or volume of alcohol, and therefore may not necessarily be considered a reliable finding unless confirmed by further research.

Other potentially related factors

There are a number of further state factors which may be related to the CAR, but which have not yet been explored in any depth. For instance, exercise (both aerobic and anaerobic) causes acute increases in serum cortisol (Kindermann et al., 1982), and increased basal levels of serum cortisol during the waking day are demonstrated in athletes during periods of intense activity (Tsai et al., 1991). Notably, these elevated basal cortisol levels were recorded at 1- to 1.5-h post-awakening, and therefore temporally proximal to the CAR period. Whilst it is common practice to discourage participants from exercising during the CAR testing period, so as to avoid a potential confound in studies, there has not been any research to investigate the effects of acute exercise during the prior day or the first hour post-awakening on the CAR. Although Stalder et al. (2009) provided evidence that motility was not associated with the dynamic of the CAR within a single case study, this was only a measure of motility within a range of resting states as the participant was instructed to avoid exercising during the CAR testing period. Likewise, the mode of awakening (alarm or natural awakening) does not appear to impact on the CAR (Stalder et al., 2009).

Summary

In healthy individuals, the CAR shows a large degree of variability across days, supporting a regulatory role within the healthy circadian pattern of cortisol secretion. Although limited in number, the studies reviewed here offer a relatively coherent emerging story about state factors that influence the CAR and the impact of the CAR on daily functioning. Greater understanding of these issues helps illuminate the utility of the CAR as a promising biomarker in psychophysiological and epidemiological research. There is evidence that the magnitude of the CAR in healthy individuals is positively correlated with increased levels of ambient light in the morning, earlier morning waking times, anticipation of significant workload or challenge during the day ahead, negative experiences during the previous day, and also with the ovulatory period of the menstrual cycle in females. Although this literature is relatively small it appears generally more consistent than the literature on trait variation in the CAR. Indeed, since the degree of state variation is so large (e.g. Hellhammer et al., 2007; Stalder et al., 2009) this may well account for the lack of consistency in the trait variation literature. While the utility of the CAR as a trait biomarker is evident, it has been recommended that for a reliable assessment it is sampled over 6 consecutive days (Hellhammer et al., 2007) and it is notable that this recommendation is rarely implemented.

It may be that state-related increases in the magnitude of the CAR are a homeostatic response allowing for a physical or psychological “boost”, either in anticipation of the day ahead or as a compensatory increase following negative experiences of the previous day (Adam et al., 2006). Indeed, support for this theory is provided by the finding that increased intra-individual variation in the CAR is associated with protective psychological factors for human health (Mikolajczak et al., 2010). Whilst further evidence is required in order for this account of the CAR to be truly tested, the possibility that appropriate CAR flexibility in response to coincident state factors may be a feature of healthy HPA axis responding certainly intensifies the difficulty in interpreting trait CAR research. It is arguable that future research should seek to establish whether a lack of flexibility in the CAR may be a vulnerability factor which results in negative mood states, or is instead a product of repeated negative life events (see Lupien et al., 2009). In such a scenario, an interaction between state and trait factors would be apparent, with negative state factors at sensitive stages of the life course giving rise to sustained or “trait” characteristics in the CAR. It follows that longitudinal or comparative research exploring state variation in the CAR between individuals and across age groups may help to answer this question.

Evidence from shift work studies suggests that optimal adaptation of the CAR to changing sleep--wake patterns occurs when there are maximal external cues for awakening. This suggests that, due to its dependence on environmental cues, the CAR may be relatively easily adaptable to the effects of travel across time zones, but not so easily adaptable to awakening in the absence of such cues. Further research into the effects of shift work is warranted as well as research exploring state changes in the CAR during jet lag, potentially providing insight into the underlying factors determining the severity of jet lag symptoms, especially likely if it does prove to be the case that the CAR plays a role in preparing for the day ahead in terms of cognitive function or arousal.

The current review highlights the need for further research into state variation to help establish the role of the CAR within the cortisol circadian rhythm. Since cortisol and the CAR are associated with hippocampal and frontal lobe function (see Fries et al., 2009 for review), it has been suggested (Clow et al., 2010a) that a rational direction for future research would be the measurement of the relationship between the CAR and neuropsychological processes underlying awakening (e.g. changes in regional cerebral blood flow and the activation of the frontal cortex in particular). Furthermore, the function of these brain regions should also be measured later in the day, to establish whether the CAR in fact prepares the individual for a day of enhanced cognitive function, in order to respond to anticipated challenge. In this way, research could establish whether the flexibility of the CAR serves the purpose of anticipatory cognitive preparation for challenge during the waking day. Research exploring state changes in the CAR in relation to workload (e.g. weekend versus weekday differences) could perhaps help toward establishing whether a reduced CAR magnitude on days of rest is beneficial in terms of physical or psychological rest and recovery.

It is evident from the research discussed here that understanding state variation in the CAR may be key to understanding its role in healthy functioning, and in turn, shed light on the implications of abnormal CAR functioning. Greater understanding of these issues helps illuminate the utility of the CAR as a promising biomarker in psychophysiological and epidemiological research.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this article.

Notes

*s1 taken immediately on awakening, s2 collected at 15 min, s3 at 30 and s4 at 45-min post-awakening. These formulae all assume equal time intervals, arbitrarily denoted at unity, between all samples.