Psychological, cardiovascular, and metabolic correlates of individual differences in cortisol stress recovery in young men

Transcription

1 Psychoneuroendocrinology 26 (2001) Psychological, cardiovascular, and metabolic correlates of individual differences in cortisol stress recovery in young men Mark P. Roy a, Clemens Kirschbaum b, Andrew Steptoe c,* a Department of Psychology, University of Central Lancashire, Preston PR1 2HE, UK b Institute of Physiological Psychology II, University of Düsseldorf, Düsseldorf, Germany c Department of Epidemiology and Public Health, University College London, 1 19 Torrington Place, London WC1E 6BT, UK Received 5 July 2000; received in revised form 6 September 2000; accepted 11 November 2000 Abstract The relationship of free salivary cortisol stress recovery and basal cortisol with psychological, cardiovascular and metabolic factors was investigated in 82 healthy young men. Blood pressure, heart rate, cortisol and mood were assessed during a single laboratory session involving mental arithmetic and speech tasks, and lipid profiles were analysed from a fasting blood sample. Participants were divided into high (n=31) and low (n=51) cortisol stress recovery groups on the basis of the magnitude of changes between the peak cortisol responses to tasks and the lowest levels recorded at the end of a 30 min post-stress rest period. The high recovery group showed consistent increases in cortisol following each of the tasks, while the low recovery group showed little change across the session. Cortisol levels in the two groups did not differ at the end of the post-stress recovery period. The groups were indistinguishable in age, body mass index, smoking and alcohol consumption, and did not differ in psychological characteristics including anxiety, depression and perceived social support. However, the high stress recovery group had elevated low density lipoprotein cholesterol and total cholesterol/high density lipoprotein ratios, suggesting raised cardiovascular disease risk. The high stress recovery group also reported greater psychological activation during tasks, and greater recent minor life stress, than did the low recovery group. There was no association between rate of cortisol recovery and cardiovascular responses to tasks. But resting cortisol was related to blood pressure stress reactivity, suggesting that cortisol played a permissive role in augmenting sympathetically-driven cardiovascular responses. The results suggest that * Corresponding author. Tel.: ; fax: address: a.steptoe@ucl.ac.uk (A. Steptoe) /01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved. PII: S (00)

2 376 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) the rate of cortisol stress recovery is associated with variations in metabolic risk, and with differences in psychological state but not trait characteristics Elsevier Science Ltd. All rights reserved. Keywords: Cortisol; Stress reactivity; Lipids; Blood pressure; Life stress; Stress recovery 1. Introduction Glucocorticoids play a central role in stress responses (Chrousos and Gold, 1992). The development of assays from saliva has greatly extended the possibilities for assessing cortisol responses in human psychobiological research (Kirschbaum and Hellhammer, 2000). The concentration of saliva free cortisol typically increases in response to acute psychological stressors such as problem-solving tasks and socially evaluative speech, with the rise being delayed by some minutes in comparison with electrodermal or cardiovascular responses. There are, however, substantial differences in cortisol reactivity, and a proportion of individuals demonstrate very little acute cortisol stress response. The magnitude of cortisol responses is influenced by situational characteristics such as task difficulty (Pruessner et al., 1999), lack of control (Peter et al., 1998), social support during stress (Kirschbaum et al., 1995a), and harassment during behavioural challenge (Earle et al., 1999). Individual factors also affect cortisol responses, including gender and age (Kirschbaum et al., 1992; Nicolson et al., 1997). Although the association between cortisol level and clinical depression is well established (Checkley, 1996), the relationship between individual differences in cortisol responses to standardised challenges and psychological factors in healthy populations have been inconsistent. No associations between cortisol reactivity and trait anxiety, depression, perceived stress or personality factors were found by van Eck et al. (1996) or Schommer et al. (1999). In a group of long-term unemployed men and women, Grossi et al. (1998) reported that participants with low basal cortisol and low responses to acute stress were more anxious and depressed. But Kirschbaum et al. (1995b) found that individuals who displayed persistent cortisol responses over 5 days of testing were more depressed and had lower self-esteem than less responsive individuals. In a previous analysis of the present data set, we found no association between cortisol responses and social support or recent aversive life events (Roy et al., 1998). There is also uncertainty about the relationship between cortisol reactivity and cardiovascular stress responses. Cardiovascular activation and cortisol release are both part of the acute stress response, and there is a substantial functional interrelationship between hypothalamic pituitary adrenocortical (HPA) and sympathoadrenal responses (Axelrod and Reisine, 1984). Studies to date have classified individuals on the basis of cardiovascular reactivity, then assessed cortisol responses. No association between cardiac reactivity and cortisol responses were observed by Manuck et al. (1991), while another study showed that heart rate reactivity predicted cortisol responses to an aversive but not an appetitive task, even though norepi-

3 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) nephrine responses were comparable to the two situations (Lovallo et al., 1990). Three relatively small scale experiments from one research group have all found evidence for positive associations between cardiovascular and cortisol responses. Sgoutas-Emch et al. (1994) pre-screened male students for heart rate reactivity to a stress test, then found that high cardiac reactors showed greater cortisol responses to a second task. A study of 24 female students showed positive associations between heart rate and blood pressure responses and cortisol reactions (Uchino et al., 1995). The third study found no association between heart rate reactivity and cortisol responses among older women; rather, cortisol responsivity was correlated with stress-induced blood pressure increases (Cacioppo et al., 1995). There is evidence to suggest that cortisol may act permissively to heightened stress-induced cardiovascular function. Glucocorticoids allow catecholamines to exert their full actions by promoting epinephrine synthesis and inhibiting catecholamine re-uptake (Munck and Naray-Fejes-Toth, 1994). It is therefore logical to assess the impact of cortisol reactivity on cardiovascular function rather than vice versa. In the present analysis, we classified individuals on the basis of cortisol change, and treated cardiovascular measures as dependent variables. In addition, we tested the hypothesis put forward by Sapolsky et al. (2000) that cortisol basal levels rather than stress-induced cortisol responses are likely to be more important as modulators of cardiovascular reactivity. The purpose of this study was to further understanding of individual differences in acute cortisol responsivity to psychological stress. Instead of classifying responsivity in terms of increases in cortisol from pre-stress levels, we evaluated individual differences in the magnitude of stress recovery. There is growing interest in evaluating the rate of stress recovery in psychobiological research, and failure to adapt after stress termination may lead to dysregulation of protective mechanisms (Linden et al., 1997). McEwen (1998) has argued that failure of stress recovery may be indicative of sustained allostatic load. A pattern of large cortisol stress responses followed by rapid recovery is observed in dominant as opposed to subordinate baboons (Sapolsky, 1995), while delayed HPA recovery may be associated with inadequate responses, leading to the mobilisation of compensatory mechanisms. In the present study, we defined stress recovery by calculating differences between peak cortisol responses to tasks and levels recorded at the end of a 30 min post-stress resting period. We hypothesised that individuals showing large recovery effects would also have higher cortisol levels in response to stressors. The study was limited to men, since there are consistent sex differences in acute cortisol stress responsivity that would have complicated the analysis (Kirschbaum et al., 1992). A final issue studied in this analysis was the association between cortisol stress responses and lipid levels. Recent evidence has implicated dysregulation of HPA function in cardiovascular disease risk, particularly insulin resistance, and lipid levels may be affected by glucocorticoid concentration in the circulation (Andrews and Walker, 1999). Phillips et al. (1998) showed in a sample of middle aged men that cortisol levels recorded at 9.00 a.m. were positively associated with blood pressure, insulin resistance and plasma triglyceride concentration, and negatively correlated with high density lipoprotein (HDL) cholesterol. Longitudinal cohort studies indicate

4 378 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) that plasma cortisol concentration in adult life is negatively associated with birth weight, itself a cardiovascular disease risk factor (Phillips et al., 2000). Elevated cortisol and high total/hdl-cholesterol ratios have also been linked in indices of allostatic load (Seeman et al., 1997). We therefore predicted that high cortisol stress recovery would be associated with unfavourable lipid profiles, including elevated low density lipoprotein (LDL) cholesterol, low HDL-cholesterol, and a high total/hdl cholesterol ratio. 2. Method 2.1. Participants Ninety-three male probationary firefighters (age range 19 32) from the London Fire & Civil Defence Authority took part in the study soon after completing basic training. Eleven firefighters were not included in the analyses due to incomplete data. Female probationary firefighters were not included in the study as they represented fewer than 1% of the graduates across the recruiting year Measures Cortisol Saliva samples were collected using cotton dental rolls (Salivettes, Sarstedt) held in the mouth until saturated, then centrifuged at 2400 rpm and 4 C for 5 min. Samples were stored at 20 C until cortisol determination using a biotin streptavidin immunoassay (Dressendörfer et al., 1992) Cardiovascular measures Beat-to-beat cardiovascular measures were computed from continuous finger blood pressure using a Finapres model 5 (TNO Biomedical Instrumentation, Amsterdam), and an EKG (V 6 ) using Na NaCl electrodes (for details, see Roy et al., 1994). A standard occlusion cuff was also used to measure blood pressure at the start of the experiment Lipid measures Fasting blood samples were assayed enzymatically (Sigma Diagnostics, Poole, Dorset) for total serum cholesterol, high density lipoprotein (HDL) cholesterol, and triglycerides. Low density lipoprotein (LDL) cholesterol was estimated by calculation. The total/hdl-cholesterol ratio was also calculated Mood Mood was assessed with the Activation Deactivation Adjective Check List (AD ACL, Thayer, 1989). Respondents used 4-point rating scales to indicate the degree to which each of 16 adjectives applied. The two dimensions of tension and psychological activation were computed, with ranges from zero (0) to 24 (high).

5 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) Psychosocial measures Anxiety was assessed with the trait anxiety scale (Spielberger, 1983), and neuroticism with the Eysenck Personality Questionnaire. Social support was measured with a shortened (12 items) version of the Social Support Questionnaire (Sarason et al., 1983). Participants were presented with 12 scenarios, and asked to list the people who would provide support under these circumstances, together with a rating of satisfaction. The satisfaction ratings were highly skewed so were not analysed. The mean number of people listed as potential sources of support was therefore used as the measure of social support. Depression was assessed with the Beck Depression Inventory (BDI, Beck, 1961), a standard measure that has been widely validated. The scale consists of 21 items, each of which is rated from 0 to 3, with higher scores indicating greater depression. The tendency to report social desirable responses was assessed using the Marlowe Crowne Social Desirability Scale (Crowne and Marlowe, 1960). The levels of psychological stress experienced by participants at the time of the laboratory session were assessed in two ways. Firstly, participants completed the Daily Stress Inventory (DSI) every day for the 7 days preceding the session. The DSI consisted of 35 items selected as being appropriate to the experience of this cohort, each of which was assessed on a 7-point scale from 1=occurred but was not stressful to 7=caused me to panic (Brantley and Jones, 1989). The average number of items endorsed per day, the mean impact per item, and total severity ( number impact) were analysed. Secondly, participants completed the Hassles scale described by Kanner et al. (1981). This consisted of 34 hassle statements, and participants were asked whether or not each hassle had occurred over the past month, and to rate intensity from 1=slight to 3=severe. The number of hassles, average intensity, and total severity (number intensity) were analysed Laboratory tasks The arithmetic task was a nonverbal task that has been used in a number of studies of stress and cardiovascular disease risk to elicit blood pressure and heart rate responses (e.g. Rüddel et al., 1988; Vogele and Steptoe, 1993). Participants were shown a slide consisting of rows of integers between 1 and 40, and asked to add them in a cumulative sequence. The matrix was displayed for 5 min. They were instructed that a financial incentive would be calculated on the basis of the percentage of the matrix completed and the accuracy achieved during the task. To increase the sense of pressure, a distracting audiotape, consisting of a random selection of sound effects was played at 70 db. The speech task was a social evaluative challenge, involving responses to two questions related to participants jobs. It was modelled on the procedure employed by Stoney et al. (1988), and produced sizeable cardiovascular responses in a previous study in our laboratory (Steptoe et al., 1993). Questions were chosen that addressed potentially stressful experiences which would have been encountered by probationary firefighters, and participants were asked to speak continuously for 2 min in response

6 380 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) to each question. They were told that their behaviour would be video recorded for later analysis Procedure Laboratory sessions were scheduled within a month of participants joining a fire station, being held on the first day of an 8-day shift cycle, with all sessions starting at a.m. Participants fasted from midnight the night before a session, and did not to brush their teeth or smoke in the 2 hours before the session. The session began with an explanation of the procedures, measures of blood pressure with a standard cuff, instrumentation with the Finapres, and a 60 min adaptation period, during which participants relaxed. Saliva samples were taken after 30 and 45 min of the adaptation phase. The main protocol consisted of seven 10 min trials: baseline, mental arithmetic task, post-arithmetic recovery period, speech task, post-speech recovery period, and two further recovery trials. Continuous cardiovascular monitoring was carried out with the Finapres during trials, and mood ratings were collected at the end of each trial. Saliva samples were obtained at the end of all trials except the baseline. At the end of the session, a fasting blood sample was taken from the antecubital vein using a butterfly cannula Data reduction Data were incomplete for the two adaptation saliva samples, so cortisol measures were averaged to produce a single adaptation measure. Cortisol data were analysed for the adaptation phase and the six further samples: arithmetic, post-arithmetic, speech, post-speech, and recovery trials 1 and 2. Systolic pressure, diastolic pressure, and heart rate were averaged over the 10 min of seven trials: baseline, arithmetic, post-arithmetic, speech, post-speech, and recovery trials 1 and 2. Some data were missing because of equipment problems, so these analyses are based on 77 of the 82 participants Classification of high and low cortisol stress recovery groups As has been observed several times before, salivary cortisol responses were delayed by several minutes. Thus the response to the arithmetic task was evidenced in the saliva sample collected 10 min after task completion, rather than that obtained immediately after the task. Similarly, responses to the speech task were indexed by the post-speech sample. Difference scores were therefore computed between recovery cortisol (defined as the lower of recovery trials 1 and 2) and highest value recorded from the post-arithmetic or post-speech sample. The mean difference was 1.57±1.9 nmol/l. The cohort was therefore divided around this value into a high cortisol stress recovery group (n=31) who showed a larger difference in salivary cortisol between task and recovery phases, and a low cortisol stress recovery group (n=51) who showed differences of 1.5 nmol/l.

7 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) Definition of high and low basal cortisol groups Since the study was carried out in the morning when cortisol levels are generally falling, pre-stress samples could not be used to define basal cortisol. However, cortisol values were low and relatively stable across recovery samples 1 and 2. The association between cardiovascular reactivity and basal cortisol was therefore tested by dividing participants on the basis of the lowest cortisol value recorded during the recovery phase. A binary division was made around the median of 2.70 nmol/l into low (n=42) and high (n=40) resting cortisol level groups. This division was entirely independent of the classification in terms of stress recovery: 38.1% (16/42) of the low resting cortisol group were in the high cortisol stress recovery category, compared with 37.5% (15/40) of the high resting cortisol group. 3. Results 3.1. Cortisol responses during the laboratory session The average salivary cortisol values recorded during the laboratory session from the high and low stress recovery groups are shown in Fig. 1. Repeated measures analysis of variance over the seven cortisol samples showed a main effect of recovery Fig. 1. Mean levels of saliva free cortisol (nmol/l) recorded in high and low cortisol stress recovery groups during the adaptation phase (Adap), and at the end of the arithmetic (MA), post-arithmetic (PostMA), speech (Speech), post-speech (PostSP), recovery 1 (Rec1) and recovery 2 (Rec2) trials.

8 382 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) group (F(1,80)=5.73, P=0.019), and sample (F(6,480)=19.7, P 0.001), together with a significant recovery group by sample interaction (F(6,480)=5.53, P 0.001). It is evident from Fig. 1 that both groups showed a decrease in cortisol between the adaptation phase and arithmetic sample. Subsequently, cortisol increased in the high stress recovery group in response to the two tasks, declining towards the end of the session. In contrast, cortisol values remained stable in the low stress recovery group. Separate analysis of the low stress recovery group showed that after the adaptation period there were no significant changes across sample (F(5,250)=1.49, P=0.19). It should not be assumed that the low stress recovery group were completely unresponsive to tasks, since in the absence of stimulation cortisol might otherwise have continued to decline. Comparisons between the groups revealed no significant differences during the adaptation phase (P=0.125) or recovery samples 1 and 2, but that cortisol was significantly elevated in the high stress recovery group during the arithmetic (P=0.012), post-arithmetic (P 0.001), speech (P 0.001) and post-speech ( P=0.007) trials. Thus it is apparent that high stress recovery was associated with heightened cortisol responsivity to tasks Demographic and psychological characteristics of stress recovery groups The background characteristics of the high and low cortisol stress recovery groups are summarised in Table 1. There were no significant differences in age, body weight, body mass, or in the prevalence of smoking or alcohol consumption. Nor did the groups differ in psychosocial measures, including trait anxiety, neuroticism, depression, social support or social desirability. Table 1 Characteristics of cortisol stress recovery groups a High cortisol stress Low cortisol stress Significance of group recovery recovery differences (n=31) (n=51) Age (yr) 24.7 (3.4) 25.4 (3.6) 0.39 Body weight (kg) 76.2 (9.4) 78.1 (8.2) 0.38 Body mass index 24.3 (2.0) 24.3 (2.5) 0.94 Cigarette smokers 34.5% 39.2% 0.81 Alcohol (frequency) 1.89 (0.83) 2.16 (1.0) 0.25 Trait anxiety 36.2 (5.6) 35.4 (6.6) 0.62 Neuroticism 9.42 (5.1) 7.55 (4.7) Beck depression scores 5.30 (0.60) 5.19 (0.90) 0.59 Social support number 3.74 (2.3) 3.55 (1.9) 0.69 Social desirability 17.6 (6.9) 15.8 (5.6) 0.28 a Means with standard deviations in parentheses.

9 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) Lipid levels and cortisol stress recovery The fasting lipid levels of the cortisol stress recovery groups are shown in Table 2. The high stress recovery group had significantly higher LDL-cholesterol ( F(1,77)=6.11, P=0.016), and total/hdl-cholesterol ratio (F(1,77)=9.49, P=0.003) than the low stress recovery group. There were also non-significant trends towards higher total cholesterol (P=0.063) and lower HDL-cholesterol (P=0.059) in the high than low stress recovery groups. The two groups did not differ in plasma triglyceride concentration. These data suggest that greater cortisol stress recovery is associated with a less healthy lipid profile Cardiovascular responses and cortisol stress recovery As reported previously, participants in this study showed substantial systolic pressure, diastolic pressure, and heart rate responses to the two behavioural tasks, with cardiovascular activity returning to low levels during the recovery phase (Roy et al., 1998). The increase in systolic pressure from rest to the mental arithmetic task averaged 23.6 mmhg, with increases of 13.4 mmhg in diastolic pressure and 11.4 bpm in heart rate. The corresponding increases in the speech task were 27.7 mmhg systolic pressure, 16.9 mmhg diastolic pressure and 16.2 bpm heart rate. The cortisol stress recovery groups did not differ in cardiovascular responses at any point in the session. Repeated measures analysis of variance of systolic pressure, diastolic pressure and heart rate revealed main effects of trial (F (6,450)=109.9, and respectively, P 0.001), with no effects of recovery group, and no stress recovery group by trial interaction. These results are illustrated in Fig. 2, where it is clear that the blood pressure response profiles of high and low cortisol stress recovery groups are indistinguishable. Table 2 Cardiovascular risk factors and cortisol stress recovery a High cortisol stress Low cortisol stress Significance of group recovery recovery differences (n=31) (n=50) Total cholesterol (mmol/l) 4.48 (1.4) 3.90 (1.3) HDL cholesterol (mmol/l/) 0.95 (0.26) 1.09 (0.36) LDL cholesterol (mmol/l/) 3.42 (1.4) 2.66 (1.3) Total/HDL ratio 5.13 (2.2) 3.81 (1.6) Triglyceride (mmol/l) 0.72 (0.37) 0.60 (0.30) 0.12 Systolic pressure (mmhg) (9.2) (9.4) 0.35 Diastolic pressure (mmhg) 63.0 (9.8) 62.3 (11.8) 0.77 a Means with standard deviations in parentheses.

10 384 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) Fig. 2. Mean systolic pressure (upper panel) and diastolic pressure (lower panel) in mmhg recorded in high and low cortisol stress recovery groups during baseline and subsequent trials (for abbreviations, see legend to Fig. 1).

11 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) Cardiovascular responses and cortisol resting level Blood pressure measured at the beginning of the session with a standard occlusion cuff did not differ between the cortisol resting level groups, averaging 114.6±8.1/62.5±11.3 mmhg in the high resting cortisol group, and 114.4±10.4/63.1±11.1 mmhg in the low resting cortisol group. Repeated measures analysis of variance was carried out on systolic and diastolic pressure, with resting cortisol level grouping as the between-subject factor. In the analysis of systolic pressure, there was a significant main effect of resting cortisol level (F(1,76)=5.56, P=0.021), together with a significant group by trial interaction (F(6,456)=3.10, P=0.006). These data are summarised in Fig. 3. It is apparent that the high resting cortisol group showed larger systolic pressure responses to the tasks, and that their systolic pressure remained elevated during the recovery trials. Posthoc analysis confirmed that the groups were significantly different during the speech task and recovery trials 1 and 2 (all P 0.05). In addition, systolic pressure reactions (indexed by change scores) to the arithmetic and speech tasks were significantly greater in the high than low resting cortisol groups (F(1,79)=5.55 and 13.5 respectively, P 0.05). The analysis of diastolic pressure recorded an effect of resting cortisol group that approached significance (F(1,76)=3.44, P=0.068), suggesting a similar though less marked pattern to that observed for systolic pressure. Fig. 3. Mean systolic pressure in mmhg recorded in high and low resting cortisol groups during baseline and subsequent trials (for abbreviations, see legend to Fig. 1).

12 386 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) Subjective responses during the laboratory session The psychological activation and tension scales were subjected to repeated measures analysis of variance, with cortisol stress recovery group as the between-subject factor. For both scales, there were significant effects of trial (F(6,474)=88.4 and 98.5 respectively, P 0.001), since participants were more psychologically activated and tense during task trials compared with baseline and recovery. In addition, the stress recovery group by trial interaction was significant in the analysis of psychological activation (F(6,474)=2.26, P=0.037). This effect is shown in Fig. 4. In post hoc analyses, we found that activation ratings were greater in the high than low cortisol stress recovery groups for the arithmetic (P=0.029), and speech (P 0.001) trials. The mean severity of daily stressors in the 7 days prior to the laboratory session averaged 13.0±9.1 in the high cortisol stress recovery group, and 9.26±6.7 in the low cortisol stress recovery group (F(1,75)=4.28, P=0.042). This suggests that the magnitude of cortisol stress recovery was positively associated with the experience of more stressors in the days preceding the laboratory session. This pattern was confirmed in the analyses of the hassles scale. The high cortisol stress recovery group reported an average of 13.6±11.2 out of the 34 hassles on the scale, compared with 7.78±5.5 in the low stress recovery group (F(1,79)=9.72, P=0.003). The average intensity rating of hassles did not differ between the groups, but the total severity was greater in the high (mean 20.4±17.6) than low (mean 11.1±8.5) stress recovery groups (F(1,79)=10.1, P=0.002). Fig. 4. Mean psychological activation ratings recorded in high and low cortisol stress recovery groups during baseline and subsequent trials (for abbreviations, see legend to Fig. 1).

13 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) Discussion This study was carried out with a sample of healthy young men newly recruited to the demanding occupation of firefighting. They had all undergone extensive evaluations of their capacity to carry out this work, and had successfully completed training. The results therefore represent findings in comparatively fit young men, and this may be relevant to their interpretation. Participants were classified into stress recovery groups on the basis of the magnitude of differences between cortisol responses to tasks and levels recorded at the end of the post-stress resting period. The cortisol responses to the tasks were somewhat smaller than those typically elicited with procedures such as the Trier Social Stress Test, possibly because of the nature of the tasks or the reassuring atmosphere in which stress tasks were administered. We chose to analyse data by dividing participants into groups rather than using regression methods, so as to take advantage of the repeated measures available. It is striking from Fig. 1 that the division of participants on this basis not only distinguished groups with large and small decreases in cortisol following tasks, but also identified groups of high and low cortisol stress responders. The high cortisol stress recovery group not only showed greater reductions to the end of the rest period, but greater responses to the tasks themselves. Linden et al. (1997) have pointed out that defining rate of post-stress recovery in this way to some extent confounds reactivity and recovery. However, the pattern we identified was not therefore a mathematical artefact, but a genuine phenomenon. It was theoretically possible, for example, that the low stress recovery group would have reacted to the stressors just as much, but then maintained high cortisol during the later trials of the session. Instead, the low stress recovery group did not show any rise in cortisol during the stress protocol. They should not be regarded as non-reactors, since cortisol might have been expected to continue a gradual reduction over the morning hours. Unfortunately, we have no cortisol data from a comparable period without behavioural tasks, so the normal rate of decline for these individuals cannot be estimated. The pattern of cortisol change showed by the high stress recovery group suggests a parallel with the large cortisol stress responses followed by rapid recovery in observed dominant wild baboons (Sapolsky, 1995). The result indicates close interrelationships between stress responsivity and recovery. We were unable to identify a group of participants showing high cortisol reactivity but no recovery, as might be predicted from the allostatic load model (McEwen, 1998). This may be because the participants were all young and relatively fit Cortisol stress recovery and lipids We found that the high cortisol stress recovery group showed an adverse lipid profile in comparison with the low stress recovery group. Although values were within the normal range, the high cortisol stress recovery group had elevated LDLcholesterol, reduced HDL-cholesterol, and a high total/hdl-cholesterol ratio (Table 2). This result is consistent with raised cardiovascular disease risk in the high cortisol

14 388 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) stress reactors. It was not associated with differences in age, smoking status or body mass. Cortisol mobilises lipids by inhibiting insulin secretion and increasing lipolysis in fat cells. A role for HPA stress reactivity in cardiovascular disease has been proposed by Bjorntorp (1998), though the situation is complicated by the possibility that chronic stress is associated in older adults with hypocortisolism (Heim et al., 2000; Rosmond and Bjorntorp, 1998). These results suggest that the relationships between resting plasma cortisol and cholesterol observed by previous investigators may be paralleled by dynamic associations with stress-induced saliva free cortisol responses Psychological characteristics We found no association between personality or trait measures of psychological state and cortisol stress recovery. This is consistent with previous research using conventional stress reactivity measures in single session experiments (Schommer et al., 1999; van Eck et al., 1996). However, the pattern may be due in part to the unreliability of estimates of cortisol responsivity based on a single session, since more robust associations may emerge when persistent high responders across several sessions are identified (Kirschbaum et al., 1995b). By contrast, associations were found between cortisol stress recovery, psychological activation during the stress session, and recent minor life stress. As shown in Fig. 4, the high cortisol stress recovery group showed greater psychological activation during tasks than low reactors, and also reported more recent life stress on two instruments, the daily stress inventory, and the hassle scale. van Eck et al. (1996) reported no association between cortisol reactivity and high or low perceived stress; however, their measure was based on a subjective estimate of current stress, rather than the endorsement of more objective day-to-day stressors as in the present study. Our results suggest that acute cortisol stress recovery may be associated with the recent experience of minor life stress, leading to enhanced subjective activation during acute challenge Cardiovascular responses The analysis of cardiovascular responses showed no association with cortisol stress recovery (Fig. 2). This is in agreement with Manuck et al. (1991) but is less consistent with other studies in the literature (Cacioppo et al., 1995; Lovallo et al., 1990; Sgoutas-Emch et al., 1994). In these latter experiments, cortisol was obtained by venepuncture, and it is possible that this procedure may disguise responses to behavioural tasks. However, it should also be recognised that cardiovascular stress responses occur much more rapidly than cortisol reactions, so are unlikely to be functionally dependent upon them. We therefore tested the hypothesis that basal glucocorticoid levels exert a permissive action on cardiovascular reactivity (Sapolsky et al., 2000). We indexed basal cortisol by the lowest value obtained during the recovery phase of the experiment. Though not ideal as a basal value, we reasoned that it would not be affected by anticipatory reactions, since it followed behavioural

15 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) tasks. In addition, the resting cortisol value was not significantly correlated with the size of decreases in cortisol from maximum levels during tasks (r=0.14), so does not appear to have been affected by the magnitude of cortisol stress responses. Systolic pressure stress responses were clearly associated with resting cortisol level. As shown in Fig. 3, responses to the two tasks were enhanced in the high resting cortisol group, and post-stress recovery was impaired. This result is consistent with the hypothesis that basal cortisol rather than cortisol stress reactivity plays a permissive role in augmenting sympathetically driven cardiovascular stress responses Study limitations There are important limitations to this study that should be recognised. As noted above, the study was carried out with healthy young men, and results may not generalise to other groups. It is possible that older adults who have experienced chronic stress would display a pattern of cortisol hyporeactivity, although such a pattern has not consistently been observed in either men or women (Kudielka et al. 1999, 2000). The study was carried out in a laboratory over a single session, and we do not have evidence concerning cortisol profiles over the day. Nonetheless, the data suggest that assessing post-stress cortisol recovery does identify characteristics that are associated with psychological and metabolic processes, so may augment the more conventional methods of assessing cortisol responsivity through changes from pre-stress reference levels. Acknowledgements This research was supported by the Medical Research Council, UK. The authors are grateful to the London Fire and Civil Defence Authority for their co-operation. References Andrews, R.C., Walker, B.R., Glucocorticoids and insulin resistance: old hormones, new targets. Clin. Sci. 96, Axelrod, J., Reisine, T.D., Stress hormones: their interaction and regulation. Science 224, Beck, A., An inventory for measuring depression. Archives of General Psychiatry 4, Bjorntorp, P., Etiology of the metabolic syndrome. In: Bray, G.A., Bouchard, C., James, W.P.T. (Eds.), Handbook of Obesity. Marcel Dekker, New York, pp Brantley, P.J., Jones, G.N., Daily Stress Inventory: Professional Manual. Psychological Assessment Resources, Inc, Odessa, FL. Cacioppo, J.T., Malarkey, W.B., Kiecolt-Glaser, J.K., Uchino, B.N., Sgoutas-Emch, S.A., Sheridan, J.F., Berntson, G.G., Glaser, R., Heterogeneity in neuroendocrine and immune responses to brief psychological stressors as a function of autonomic cardiac activation. Psychosom. Med. 57, Checkley, S., The neuroendocrinology of depression and chronic stress. Br. Med. Bull. 52, Chrousos, G.P., Gold, P.W., The concepts of stress and stress system disorders. Overview of physical and behavioral homeostasis. JAMA 267,

16 390 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) Crowne, D.P., Marlowe, D., A new scale of social desirability independent of psychopathology. J. Consult. Psychol. 24, Dressendörfer, R.A., Kirschbaum, C., Rohde, W., Stahl, F., Strasburger, C.J., Synthesis of a cortisol biotin conjugate and evaluation as a tracer in an immunoassay for salivary cortisol measurement. J. Steroid Biochem. Mol. Biol. 43, Earle, T.L., Linden, W., Weinberg, J., Differential effects of harassment on cardiovascular and salivary cortisol stress reactivity and recovery in women and men. J. Psychosom. Res. 46, Grossi, G., Ahs, A., Lundberg, U., Psychological correlates of salivary cortisol secretion among unemployed men and women. Integr. Physiol. Behav. Sci. 33, Heim, C., Ehlert, U., Hellhammer, D.H., The potential role of hypocortisolism in the pathophysiology of stress-related bodily disorders. Psychoneuroendocrinology 25, Kanner, A.D., Coyne, J.C., Schaefer, C., Lazarus, R.S., Comparisons of two modes of stress measurement: daily hassles and uplifts versus major life events. J. Behav. Med. 4, Kirschbaum, C., Hellhammer, D.H., Salivary cortisol. In: Fink, G. (Ed.), Encyclopedia of Stress, vol. 3. Academic Press, San Diego, pp Kirschbaum, C., Klauer, T., Filipp, S.-H., Hellhammer, D.H., 1995a. Sex-specific effects of social support on cortisol and subjective responses to acute pychological stress. Psychosom. Med. 57, Kirschbaum, C., Prussner, J.C., Stone, A.A., Federenko, I., Gaab, J., Lintz, D., Schommer, N., Hellhammer, D.H., 1995b. Persistent high cortisol responses to repeated psychological stress in a subpopulation of healthy men. Psychosom. Med. 57, Kirschbaum, C., Wust, S., Hellhammer, D., Consistent sex differences in cortisol responses to psychological stress. Psychosom. Med. 54, Kudielka, B.M., Schmidt-Reinwald, A.K., Hellhammer, D.H., Kirschbaum, C., Psychological and endocrine responses to psychological stress and dexamethasone/corticotropin-releasing hormone in healthy postmenopausal women and young controls: the impact of age and a two-week estradiol treatment. Neuroendocrinology 70, Kudielka, B.M., Schmidt-Reinwald, A.K., Hellhammer, D.H., Kirschbaum, C., Psychosocial stress and HPA functioning: no evidence for a reduced resilience in healthy elderly men. Stress 3, Linden, W., Earle, T.L., Gerin, W., Christenfield, N., Physiological stress reactivity and recovery: conceptual siblings separated at birth? J. Psychosom. Res. 42, Lovallo, W.R., Pincomb, G.A., Brackett, D.J., Wilson, M.F., Heart rate reactivity as a predictor of neuroendocrine responses to aversive and appetitive challenges. Psychosom. Med. 52, Manuck, S.B., Cohen, S., Rabin, B.S., Muldoon, M.F. et al., Individual differences in cellular immune response to stress. Psychol. Sci. 2, McEwen, B.S., Protective and damaging effects of stress mediators. New Engl. J. Med. 338, Munck, A., Naray-Fejes-Toth, A., Glucocorticoids and stress: permissive and suppressive actions. Ann. N.Y. Acad. Sci. 746, Nicolson, N., Storms, C., Ponds, R., Sulon, J., Salivary cortisol levels and stress reactivity in human aging. J. Gerontol. A, Biol. Sci. Med. Sci. 52, M Peter, R., Alfredsson, L., Hammar, N., Siegrist, J., Theorell, T., Westerholm, P., High effort, low control, and cardiovascular risk factors in employed Swedish men and women: baseline results from the WOLF study. J. Epidemiol. Comm. Health 52, Phillips, D.I., Barker, D.J., Fall, C.H., Seckl, J.R., Whorwood, C.B., Wood, P.J., Walker, B.R., Elevated plasma cortisol concentrations: a link between low birth weight and the insulin resistance syndrome? J. Clin. Endocrinol. Metab. 83, Phillips, D.I., Walker, B.R., Reynolds, R.M., Flanagan, D.E.H., Wood, P.J., Osmond, C., Barker, D.J.P., Whorwood, C.B., Low birth weight predicts elevated plasma cortisol concentrations in adults from 3 populations. Hypertension 35, Pruessner, J.C., Hellhammer, D.H., Kirschbaum, C., Low self-esteem, induced failure and the adrenocortical stress response. Pers. Ind. Diff. 27, Rosmond, R., Bjorntorp, P., Endocrine and metabolic aberrations in men with abdominal obesity in relation to anxio-depressive infirmity. Metabolism 47,

17 M.P. Roy et al. / Psychoneuroendocrinology 26 (2001) Roy, M., Steptoe, A., Kirschbaum, C., The association between smoking status and cardiovascular and cortisol stress responsivity in healthy young men. Int. J. Behav. Med. 1, Roy, M.P., Steptoe, A., Kirschbaum, C., Life events and social support as moderators of individual differences in cardiovascular and cortisol reactivity. J. Pers. Soc. Psychol. 75, Rüddel, H., Langewitz, W., Schächinger, A., Schmieder, R., Schulte, W., Hemodynamic response patterns to mental stress: Diagnostic and therapeutic implications. Am. Heart J. 116, Sapolsky, R.M., Social subordinance as a marker of hypercortisolism. Some unexpected subtleties. Ann. N.Y. Acad. Sci. 771, Sapolsky, R.M., Romero, L.M., Munck, A.U., How do glucocorticoids influence stress responses? Integrating permissive, suppressive, stimulatory, and preparative actions. Endocr. Rev. 21, Sarason, I.G., Levine, H.M., Basham, R.B., Sarason, B.R., Assessing social support: the Social Support Questionnaire. J. Pers. Soc. Psychol. 44, Schommer, N.C., Kudielka, B.M., Hellhammer, D.H., Kirschbaum, C., No evidence for a close relationship between personality traits and circadian cortisol rhythm or a single cortisol stress response. Psychol. Rep. 84, Seeman, T.E., Singer, B.H., Rowe, J.W., Horwitz, R.I., McEwen, B.S., Price of adaptation allostatic load and its health consquences. Arch. Intern. Med. 157, Sgoutas-Emch, S.A., Cacioppo, J.T., Uchino, B.N., Malarkey, W., Pearl, D., Kiecolt-Glaser, J.K., Glaser, R., The effects of an acute psychological stressor on cardiovascular, endocrine, and cellular immune response: a prospective study of individuals high and low in heart rate reactivity. Psychophysiology 31, Spielberger, C.D., Manual for the State Trait Anxiety Inventory (Form Y). Consulting Psychologists Press, Palo Alto. Steptoe, A., Kearsley, N., Walters, N., Cardiovascular activity during mental stress following vigorous exercise in sportsmen and inactive men. Psychophysiology 30, Stoney, C.M., Matthews, K.A., McDonald, R.H., Johnson, C.A., Sex differences in lipid, lipoprotein, cardiovascular, and neuroendocrine responses to acute stress. Psychophysiology 25, Thayer, R.E., The Psychobiology of Mood and Arousal. Oxford University Press, Oxford. Uchino, B.N., Cacioppo, J.T., Malarkey, W., Glaser, R., Individual differences in cardiac sympathetic control predict endocrine and immune responses to acute psychological stress. J. Pers. Soc. Psychol. 69, van Eck, M.M., Nicolson, N.A., Berkhof, H., Sulon, J., Individual differences in cortisol responses to a laboratory speech task and their relationship to responses to stressful daily events. Biol. Psychol. 43, Vogele, C., Steptoe, A., Anger inhibition and family history as modulators of cardiovascular responses to mental stress in adolescent boys. J. Psychosom. Res. 37,