ARTICLE IN PRESS Psychoneuroendocrinology (2007) 32, 322–330 Available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/psyneuen Coffee, stress and cortisol in nursing staff Anette Harrisa,b,c, Holger Ursina,b, Robert Murisond, Hege R. Eriksena,b, a Unifob health, Christiesgt 13, N-5015 Bergen, Norway Department of Education and Health Promotion, 13, N-5015 Bergen, Norway c Faculty of Health and Social Sciences, Bergen University College, Haukelandsbakken 45, N-5009 Bergen, Norway d Department of Biological and Medical Psychology, University of Bergen, Jonas Liesvei 91, N-5009 Bergen, Norway b Received 4 May 2006; received in revised form 7 January 2007; accepted 9 January 2007 KEYWORDS Salivary cortisol; Job stress; Work load; Subjective health complaints; Health-related quality of life; Coffee Summary According to cognitive activation theory, long-lasting work demands without rest or lack of coping may lead to sustained activation and pathology. Cortisol is one of the most important stress hormones in humans and increased basal levels of cortisol are considered a valid marker for sustained activation. In order to investigate this association further, we investigate the relationships between salivary cortisol profiles, job stress, work load (effort/reward, demand/control) and health (subjective health complaints and healthrelated quality of life) in a population of health care workers. Forty-four women filled in a questionnaire and collected five saliva samples on two consecutive working days (1: wakeup time, 2: wake-up time+30 min, 3: wake-up time+45 min, 4: 1500 h and 5: 2200 h). There was no relationship between psychosocial factors at work and cortisol levels in the morning (cortisol level at wake-up time and awakening cortisol response (ACR)). Only the confounding variable tobacco reached a significant level in the hierarchical regressions analyses. Our significant findings are limited to the afternoon decline and the evening values. The decrease during the day relates to decision authority, physical functioning, general health, and vitality in the single, unadjusted regression analyses. The decrease also relates to coffee intake, which we included originally as a confounding variable. In the final hierarchical regression of the evening values, only decision authority and coffee were significantly related to cortisol levels in the evening. & 2007 Published by Elsevier Ltd. 1. Introduction The cognitive activation theory of stress (CATS) states that the stress response is synonymous with activation and should Corresponding author. Department of Education and Health Promotion, University of Bergen, 13, N-5015 Bergen, Norway. Tel.: +47 55586234/55589975; fax: +47 55589872. E-mail address: hege.eriksen@psych.uib.no (H.R. Eriksen). 0306-4530/$ - see front matter & 2007 Published by Elsevier Ltd. doi:10.1016/j.psyneuen.2007.01.003 be regarded as a general alarm system (Ursin and Eriksen, 2004). Activation (the stress response) occurs whenever there is a discrepancy between what the organism is expecting, and what really exists (Ursin, 1988; Levine and Ursin, 1991; Ursin and Eriksen, 2004). The stress response affects the biochemistry of the brain, the endocrine, vegetative, and immune systems (Levine and Ursin, 1991). The main components of the stress response are the hypothalamic–pituitary–adrenal axis (the HPA axis) and the ARTICLE IN PRESS Coffee, stress and cortisol in nursing staff sympathetic part of the vegetative nervous system. It is an essential element in the total adaptive system of the body. Hormones associated with the stress response protect the body in the short term and promote adaptation, but in the long run may cause changes in the body that lead to disease (McEwen and Seeman, 1999; Ursin, 2000). According to CATS, long-lasting work demands without rest, or lack of coping, may lead to sustained activation and pathology (Levine and Ursin, 1991; Ursin and Eriksen, 2004). Cortisol is one of the most important stress hormones in humans and increased basal levels of cortisol are considered to be a valid marker for sustained activation (Pruessner et al., 1997; Wüst et al., 2000). Cortisol exhibits a marked circadian rhythm, with peak levels usually found in the early morning hours with decreasing concentrations thereafter (Weitzman et al., 1971). The awakening cortisol response (ACR) can be defined as the period of cortisol secretory activity in the first 45–60 min immediately post-awakening (Clow et al., 2004). Pruessner et al. (1997) have shown that ACR is a reliable biological marker of HPA activity with high intra-individual stability and therefore it can be used to measure the level of psychological strain in daily life. Previous studies have shown relationships between cortisol and job stress (for literature review, see Hjortskov et al., 2004), workload (Steptoe et al., 2000; Lundberg and Hellstrøm, 2002; Steptoe et al., 2004) and subjective health and well being (Lindfors and Lundberg, 2002). A high ACR has been related to depressive symptoms (Pruessner et al., 2003a), however and attenuated ACR have been found to be related to ‘‘burnout’’ (Pruessner et al., 1999), A recent study has shown that high levels of time pressure, effort and effort/reward imbalance were significantly associated with higher levels of cortisol in women (Eller et al., 2006). Many factors influence cortisol levels (Wüst et al., 2000; Clow et al., 2004; Hjortskov et al., 2004). Characteristics of individuals like age and gender are important, as well as the method used for determining the cortisol levels. Some studies have found a relationship between age and cortisol levels in the morning (Seeman et al., 2001; Kudielka and Kirschbaum, 2003) while others did not find this association (Pruessner et al., 1997; Wüst et al., 2000). Researchers have found a relationship between gender and cortisol (Pruessner et al., 1997; Steptoe et al., 2000; Wüst et al., 2000; Clow et al., 2004). This is complicated further by the use of oral contraceptives which may attenuate ACR, while there does not seem to be any association between phase of the menstrual cycle and cortisol (Kudielka and Kirschbaum, 2003). Factors like coffee, tobacco and alcohol may also influence cortisol levels (Pruessner et al., 1997; Clow et al., 2004; Hjortskov et al., 2004; Kunz-Ebrecht et al., 2004; Steptoe and Ussher, 2006). Further clarification requires homogenous materials and control of confounding factors. We have therefore, studied a homogenous sample of female health care workers. Previous studies have been concerned mainly with the cortisol response in the morning or total 24 h cortisol secretion as measured in urine. The cortisol decline during the day and the cortisol levels in the evening appear to be related to strain in daily life and health. Sluiter et al. (1998, 1999, 2003) investigated how working conditions may lead to health complaints. Their studies showed that sufficient restitution was an important factor for health. In these 323 studies, recovery from work was measured as the return to base line levels of cortisol after work (Sluiter et al., 1998). A recent study has shown that in a population of women working within the public health care sector, self-rated recovery from work stress was related to allostatic load (von Thiele et al., 2006). We suggest that high cortisol levels in the evening indicate lack of restitution (recovery) and may be an indicator for sustained activation (Eriksen and Ursin, 2004). Studies have shown that women’s stress seems to be more reflected in elevated arousal not only during, but also after work (Lundberg and Frankenhauser, 1999). In the present study we investigated the relationship between salivary cortisol profile, self-reported job stress, work load and health in a sample of female health care workers. Their working situation is known to be taxing (Engstrom et al., 2006). We collected five saliva samples on two consecutive working days. Work load was measured with the demand/ control model (Karasek and Theorell, 1990) and the effort/ reward imbalance model (Siegrist, 1996). Job stress was measured with the Cooper Stress Index (CSI) (Cooper, 1981) and health was measured with subjective health complaints (SHC) inventory (Eriksen et al., 1999) and health-related quality of life (SF-36) (Ware et al., 1994; Loge et al., 1998). Based on the concept of need for recovery after work (Sluiter et al., 1998, 1999, 2003), we hypothesised that there would be a relationship between high scores on workload and/or job stress and a flat cortisol profile (high cortisol at wake-up time, attenuated response to stress stimuli and high cortisol in the evening). We also expected to find that participants with high scores on SHC and/or low health-related quality of life would have high cortisol levels in the evening. 2. Methods 2.1. Subjects The participants were forty-four women mean aged 48 years (SD ¼ 8.16) working in three nursing homes for the elderly in Bergen. The data were collected as part of an integrated health programme that combined exercise and health information. Two hundred and twenty-seven employees initially were invited to the program and 135 (59.5%) accepted the invitation to take part in the intervention. Only those that worked daytime shifts when they collected saliva were included in this study (n ¼ 54). Out of these 54 employees, nine subjects were excluded because of problems with sample collection, and one participant was excluded because the questionnaires were completed inadequately. The project was approved by the ethics committee and all participants gave written informed consent. 2.2. Study protocol Diurnal cortisol was measured from saliva using Salivette collection tubes (DPC Norway, Brakerøya, Drammen), which consists of a small cotton swab inside a standard centrifugation tube. Five saliva samples were collected on two consecutive working days: at wake-up time, 30 and 45 min after wake-up time, at 1500 and 2200 h. All participants at ARTICLE IN PRESS 324 the workplace were instructed in how to use the salivette. They were instructed to chew gently on the cotton swab for 1 min to obtain the desired amount of saliva. Participants were asked to avoid meals, drinks and nicotine for 15 min before sample collection and to take the first sample while still lying in bed. They were instructed to keep the samples frozen until they were returned to the laboratory for further analyses. In the laboratory they were subsequently stored in a freezer until they were thawed for the biochemical analysis. Participants were asked to note their time of awakening and the exact time they collected the saliva samples. Compliance with the waking sampling protocol was judged by computing the difference between the time participants stated they had woken up and the time they stated that the first saliva sample had been taken. Participants showing a difference of more than 10 min were excluded. 2.3. Instruments All participants filled in a comprehensive questionnaire covering a broad range of factors including demographic variables, number of cigarettes, cups of coffee and units of alcohol (one unit alcohol ¼ a half bottle of beer or one glass of wine or one drink), use of oral contraceptives, weight and height were measured to calculate BMI, health status/ problems, physical and psychological variables at work, factors related to work organisation and leadership, social support and family relationships, and individual coping resources. 2.3.1. Quality of life Health-related quality of life was measured by the generic health status measure SF-36 for health situations during the last 4 weeks (Loge et al., 1998; Ware et al., 1994, 2000). Eight subscales were computed; physical functioning (10 items), role limitations due to physical problems (4 items), bodily pain (2 items), general health (5 items), vitality (4 items), social functioning (2 items), role limitations due to emotional problems (3 items) and mental health (5 items). The scoring of the items varied from dichotomous scales (yes/no) to six-point ordinal scales. The questionnaire has been tested and has satisfactory reliability and validity (Loge et al., 1998). 2.3.2. Subjective health complaints SHC were measured by 29 items on subjective somatic and psychological complaints experienced during the last 30 days, using the SHC inventory (Eriksen et al., 1999). Severity was scored on a four-point scale, from 0 (no complaints) to 3 (severe complaints). Five sub-scales and a total health score (SHC-total) were computed; musculoskeletal pain (8 items: headache, neck pain, upper back pain, low back pain, arm pain, shoulder pain, migraine and leg pain), pseudoneurology (7 items: palpitation, heat flushes, sleep problems, tiredness, dizziness, anxiety and sadness), gastrointestinal problems (7 items: heartburn, epigastric discomfort, ulcer/ non-ulcer dyspepsia, stomach pain, gas discomfort, diarrhoea and constipation), allergy (5 items: asthma, breathing difficulties, eczema, allergy, and chest pain) and flu (2 items: cold flu and coughing). The questionnaire has A. Harris et al. been tested and has satisfactory reliability and validity (Eriksen et al., 1999). 2.3.3. Job stress The job stress factor was measured by the CSI (Cooper, 1981, Endresen et al., 1991). Questions were scored on a six-point scale ranging from 0 to 5 (0 ¼ no stress and 5 ¼ considerable stress). Four sub-scales were computed; communication (8 items) is related to lack of communication and influence, conflicts with management, co-workers, and different groups of employees. Leadership (4 items) is related to the employee’s relationship to management and subordinates, a feeling of being undervalued, and pay. Relocation (4 items) is related to stress caused by promotion prospects, relocation, managing people, and taking work home. Workload (3 items) as a source of job stress is related to workload, time pressure and deadlines, and workrelated demands influencing private life. The questionnaire has satisfactory validity and reliability (Endresen et al., 1991). 2.3.4. Coping Coping was measured by the instrumental mastery oriented coping (IMOC) factor from the CODE (Eriksen et al., 1997), based on the short version of the Utrecht Coping List (UCL) (Schreurs et al., 1988, 1993). IMOC implies an instrumental, active, goal-oriented coping style. To get a high score on IMOC, the score on active problem solving from UCL must be high, and the score on avoidance and passive expectancy, and depressive reaction pattern from the UCL must be low (Eriksen et al., 1997). Scores are based on active problem solving (7 items), avoidance and passive expectancy (8 items), depressive reactions (7 items). The questionnaire has satisfactory reliability and validity (Eriksen et al., 1997). 2.3.5. Effort/reward imbalance The effort/reward imbalance at work questionnaire was used to measure workload, effort and reward (Siegrist, 1996). The questionnaire consists of 17 questions rated on a four-point scale. The effort scale (6 items) measures an intrinsic personal, coping-related component of stressful experience at work, while the reward scale (11 items) measures an extrinsic perceived work situation component. The total workload score (effort/reward) is a measure of the imbalance between these two components. A high score indicates high workload, effort or reward. The questionnaire has satisfactory reliability and validity (Siegrist, 1996). 2.3.6. Job demands Job demands were measured by five questions from the demand dimension in the short Swedish version (Theorell et al., 1993) of the demand/control model (Karasek and Theorell, 1990). The job demands subscale is the sum of five items inquiring about excessive work, conflicting demands, insufficient time to work, fast pace, and working hard (5 items). Questions were scored on a four-point scale ranging from strongly disagreeing to strongly agreeing. The questionnaire has satisfactory validity and reliability (Theorell et al., 1993). ARTICLE IN PRESS Coffee, stress and cortisol in nursing staff 2.3.7. Job control Job control (decisions latitude) was measured by six questions from the decision latitude dimension in the short version of the demand/control model (Theorell et al., 1993). The job control (decision latitude) scale (6 items) is the sum of two subscales; skill discretion (4 items) and decision authority (2 items). Questions were scored on a four-point scale ranging from strongly disagreeing to strongly agreeing and were related to creativity, skills, task, variety, learning new things, and the amount of repetitive work. The questionnaire has satisfactory validity and reliability (Theorell et al., 1993). 2.3.8. Social support Social support was measured by six questions from the Swedish version of The Job Content Questionnaire (6 items) (Theorell et al., 1993). The social support scales include support from co-workers and supervisors. For each item the respondents could choose from one of four responses ranging from strongly disagreeing to strongly agreeing. 2.4. Cortisol analysis Salivary cortisol was assayed with Coat-a-Count RIA kit from Diagnostic Products Corporation (DPC, Los Angeles, CA). Intra-assay variability was 3.0–5.1 and inter-assay variability was 4.0–6.4. The distribution of cortisol level at each time point was positively skewed, so a logarithmic transformation (log 10) was conducted to ensure that scores approximated a normal distribution (Tabachnick and Fidell, 1996). 2.5. Statistics SPSS 13.0 for Windows was used for the statistical analyses. Outcome variables were cortisol level at wake-up time, ACR, cortisol decline from morning to evening and cortisol level in the evening (2200 h). The ACR was measured using the integrated area under the curve (AUC). This was calculated by first subtracting the 0-min awakening value from the 30 and 45 values, to obtain baseline-corrected values, and then using the trapezoidal method (Pruessner et al., 2003b). Cortisol decline from morning to evening was measured by calculating changes between levels obtained (awake+0045 h.) and the evening sample (2200 h). Pearson correlation coefficients were used to calculate the relationship between the two collecting days, and mean cortisol levels over the two collection days were used for the analyses. Pearson product–moment correlations were also used to examine the relationship between health-related factors, psychosocial factors, coping, potential covariates and measures of cortisol. Due to low sample size, only those variables which correlated significantly with the different cortisol measures were taken into further analyses. Multiple linear regression analyses were performed to examine the contributions of each of the significant variables to the different cortisol measures. The regression analyses were carried out manually as follows. Single regression analyses were executed for each variable that showed a significant correlation with that particularly cortisol measure. Then one variable at a time was entered into the hierarchical regressions model, starting with the potential covariates, followed by health-related factors and psychoso- 325 cial factors as independent variables and the different cortisol measures as dependent variables. Tolerances were checked to control for multicollinearity. P values of less than .05 were considered statistically significant. 3. Results 3.1. Simple statistics Twenty-eight participants collected all five saliva samples, 36 participants collected the first sample at wake-up time, 28 participants collected the three samples we used for calculating the ACR and 40 participants collected the evening sample. The relationship between the first and second sampling day was investigated using Pearson product–moment correlation coefficient. There was a strong positive correlation between the cortisol samples on the two collecting days: at wake-up time (r ¼ .67, n ¼ 20, po.01), 30 min after wakeup time (r ¼ .71, n ¼ 24, po.01), 45 min after wake-up time (r ¼ .69, n ¼ 23, po.01), at 15.00 (r ¼ .40, n ¼ 30, po.05) and at 2200 h (r ¼ .54, n ¼ 25, po.01). For descriptive statistics and correlations between potential covariates, health-related factors, psychosocial factors and measures of cortisol, see Table 1. There was a positive correlation between physical functioning (SF-36), general health (SF-36), vitality (SF-36) and cortisol decline and a negative correlation between vitality (SF-36) and cortisol levels in the evening. There were no significant correlations between the other factors in the SF-36 scale or the SHC inventory and any of the cortisol outcome variables. The population reported good health-related quality of life (mean score on SF-36 ¼ 77.8, SD ¼ 20.8) compared to the Norwegian population where the mean score is 50 (Loge et al., 1998). The factors from the demand control model showed a significant correlation with the different cortisol measurement in this population (Table 1). Decision latitude was positively correlated to cortisol levels at wake-up time and decision authority was positively correlated to cortisol levels at wake-up time and cortisol decline during the day but negatively correlated to ACR and cortisol levels in the evening. Social support was positively correlated to ACR. Effort/reward imbalance and self-reported job stress had no significant influence on cortisol in this population. Furthermore, there was a positive correlation between coping and cortisol decline during the day and a negative correlation between coping and cortisol level in the evening. When examining the association between potential confounding variables and the different cortisol measures, we found a negative correlation between coffee and cortisol level at wake-up time and cortisol decline during the day and a positive correlation between coffee and cortisol levels in the evening. Tobacco was negatively correlated to cortisol level at awakening and positively correlated to the ACR. There was a negative correlation between BMI and ACR. 3.2. Regression analyses of the different cortisol measurements 3.2.1. Cortisol levels at wake-up time and ACR Table 2 shows the result of the single unadjusted regression analyses and the hierarchical multiple regression analyses ARTICLE IN PRESS 326 A. Harris et al. Table 1 Descriptive statistics and Pearson product–moment correlations between potential covariates, health related factors, psychosocial factors and measures of cortisol. 1.Age 2. Coffee 3. Tobacco 4. Alcohol 5. BMI 6. Subjective health complaints total (SHC) 7. Musculoskeletal complaints (SHC) 8. Pseudoenurological complaints (SHC) 9. Gastrointestinal complaints (SHC) 10. Allergy (SHC) 11. Flu (SHC) 12. Physical functioning (SF-36) 13. Role physical (SF-36) 14. Bodily pain (SF-36) 15. General health (SF-36) 16. Vitality (SF-36) 17. Social functioning (SF-36) 18. Role emotional (SF-36) 19. Mental health (SF-36) 20. Job stress total (CSI) 21. Communication (CSI) 22. Leadership (CSI) 23. Workload (CSI) 24. Relocation (CSI) 25. Effort/reward imbalance (ERI) 26. Effort 27. Reward 28. Workload (demand/control) 29. Physical demands 30. Decision latitude (control) 31. Skill discretion 32. Decision authority 33. Social support 34. Coping (IMOC) N Mean (SD) Awakening ACR Decline 44 44 41 40 44 44 44 44 44 44 44 42 43 44 43 42 44 42 42 43 41 43 43 42 39 39 39 44 44 44 44 44 44 43 48.8 2.3 3.7 1.5 24.7 10.8 4.7 2.9 1.8 .8 .6 87.7 81.8 71.3 75.6 56.8 81.8 87.7 79.7 24.8 1.2 1.1 2.2 .7 .1 16.3 46.0 .8 13.8 17.7 11.2 17.7 19.5 3.1 .004 .359* .485** .118 .237 .082 .189 .176 .077 .135 .073 .078 .181 .119 .090 .302 .310 .125 .160 .145 .062 .204 .257 .002 .061 .343 .050 .224 .039 .338* .192 .365* .294 .076 .141 .278 .531** .060 .406* .001 .154 .130 .201 .228 .088 .209 .180 .155 .189 .317 .267 .123 .193 .234 .149 .340 .348 .010 .107 .173 .213 .170 .000 .286 .119 .387* .379* .025 .118 .376* .072 .013 .161 .266 .187 .276 .097 .262 .276 .363* .139 .267 .337* .420* .171 .101 .105 .068 .135 .052 .103 .034 .252 .127 .116 .224 .238 .061 .195 .390* .130 .389* (8.2) (.9) (6.4) (1.7) (3.9) (8.9) (4.5) (2.5) (2.5) (1.4) (1.1) (11.4) (31.1) (24.6) (16.1) (19.1) (21.8) (27.8) (14.3) (17.6) (1.1) (1.1) (1.4) (.8) (.3) (5.0) (6.8) (.2) (2.5) (2.5) (1.9) (2.5) (2.7) (.2) Cortisol 2200 h .140 .426** .000 .048 .074 .156 .149 .204 .023 .056 .185 .181 .170 .223 .299 .331* .064 .004 .090 .079 .005 .228 .193 .033 .246 .145 .103 .232 .224 .085 .216 .459** .020 .316* *Po.05 (Pearsons correlations coefficient). **Po.01 (Pearsons correlations coefficient). predicting the cortisol levels at wake-up time. In an unadjusted model coffee, tobacco, decision latitude and decision authority was significantly associated with the cortisol level at wake-up time, but none of the variables reached a significant level in the hierarchical analyses. Table 3 shows the result of the single unadjusted regression analyses and the hierarchical multiple regression analyse predicting the ACR. In an unadjusted model tobacco, BMI, decision authority, and social support were significantly associated with the ACR. Among the potential confounding variables, tobacco was the only variable that reached significant levels in the hierarchical analyse (see Table 3). 3.2.2. Cortisol decline during the day and cortisol levels in the evening (2200 h) Table 4 shows the result of the single unadjusted regression analyses and the hierarchical multiple regression analyses predicting the cortisol decline during the day. In an unadjusted model coffee, physical functioning (SF-36), general health (SF-36), vitality (SF-36), decision authority and coping was significantly associated with cortisol level at wake-up time, but only coffee reached a significant level in the hierarchical model. Coffee alone accounted for 8.4% of the variance in the cortisol decline during the day (b ¼ .34, po.05). Table 5 shows the result of the single unadjusted regression analyses and the hierarchical multiple regression analyses predicting the cortisol levels in the evening. Unadjusted coffee, vitality (SF-36), decision authority and coping were significantly associated with the cortisol level in the evening (2200 h). Coffee reached a significant level at step one and two but not in step three. Only decision authorities reached a significant level in step three (see Fig. 1). Coffee alone accounted for 15.2% of the variance in cortisol levels in the evening (b ¼ .42, po.01), while ARTICLE IN PRESS Coffee, stress and cortisol in nursing staff 327 Table 2 Single and hierarchical multiple linear regression analyses of age, coffee, tobacco, decision latitude (control) and decision authority with cortisol level at wake-up time as dependent variable. Not adjusted 1. 2. 3. 4. 5. Age Coffee Tobacco Decision latitude (control) Decision authority Step one Step two b P b P .004 .359 .357 .338 .365 .982 .032 .032 .044 .029 .270 .267 .114 .117 Adjusted R R change Step three b P .211 .241 .226 .220 .154 .175 .143 .192 b P .192 .227 .129 .157 .274 .183 .546 .471 .167 .046 .154 .013 Table 3 Single and hierarchical multiple linear regression analyses of age, tobacco, BMI, decision authority and social support with awakening cortisol response (ACR) as dependent variable. Not adjusted 1. 2. 3. 4. 5. Age Tobacco BMI Decision authority Social support Step one b P .141 .531 .406 .387 .379 .475 .004 .032 .042 .047 Adjusted R R change Step two Step three b P b P b P .446 .250 .018 .170 .364 .255 .155 .091 .166 .440 .317 .203 .200 .296 .126 .251 .302 .088 .282 .337 .270 .017 .333 .082 Table 4 Single and hierarchical multiple linear regression analyses of age, coffee, physical functioning (SF-36), general health (SF-36), vitality (SF-36), decision authority and coping with cortisol decrease as dependent variable. Not adjusted 1. 2. 3. 4. 5. 6. 7. Age Coffee Physical functioning General health Vitality Decision authority Coping (IMOC) Adjusted R R change b P .118 .376 .363 .337 .420 .390 .389 .492 .024 .032 .047 .012 .019 .019 Step one Step two b P .335 .053 .084 .112 coffee, vitality, decision authority and coping together accounted for 22.2% of the variance in cortisol levels in the evening. The tolerance figures were between .63 and .89 for all the independent variables. 4. Discussion There was no relationship between psychosocial factors at work and cortisol levels in the morning (cortisol level at b P .258 .297 .004 .228 .196 .127 .985 .238 .188 .174 Step three Step four b b P .149 .234 .039 .197 .067 .181 .506 .256 .858 .318 .727 .342 .207 .266 .005 .219 .117 .171 .010 .337 .189 .981 .263 .526 .169 .024 wake-up time and ACR). Only the confounding variable tobacco, reached a significant level in the hierarchical analyses. Our significant findings are limited to the afternoon decline and the evening values of cortisol. The decrease during the day relates to decision authority, physical functioning, general health, and vitality in the single, unadjusted regression analyses. The decrease also related to coffee intake, which we included originally as a confounding variable. In the final hierarchical regression of the evening values, only decision authority and coffee were significant. ARTICLE IN PRESS 328 A. Harris et al. Table 5 Single and hierarchical multiple linear regression analyses of age, coffee, vitality (SF-36), decision authority and coping with cortisol level in the evening (2200 h) as dependent variable (n ¼ 39). Not adjusted 1. 2. 3. 4. 5. Age Coffee Vitality Decision authority Coping (IMOC) Step one b P .140 .426 .331 .459 .316 .390 .006 .040 .003 .047 Adjusted R R change Step two Step four b P b P b P b P .418 .008 .340 .194 .043 .238 .246 .147 .329 .135 .351 .038 .222 .140 .315 .075 .201 .382 .052 .643 .152 .175 .162 .032 low score 25 Step three .240 .093 .222 .004 high score Cortisol nmol/l 20 15 10 5 0 Awakenig Awake+0030h awake+0045h Timepoint 1500h 2200h Figure 1 Cortisol levels at each time point in participants with high versus low score on decision authority. The graph shows the mean value over time, with error bars representing the standard error of the mean. Participants with high score on decision authority had significantly lower cortisol levels in the evening (2200 h) (po.05). The surprising effect in this study is the effect of the coffee intake. Coffee alone accounted for 8.4% of the cortisol decline during the day and 15.2% of the cortisol level in the evening (2200 h). This may be more than just an irritating confounding factor. Coffee is a strongly activating and stimulating drink. It is well known that the caffeine in coffee activates the human stress axis but the underlying mechanisms are not yet fully delineated (Al ‘Absi and Lovallo, 2004). The caffeine’s stimulatory effects influence sleep and it is well known that the effect of coffee last for hours (Shilo et al., 2002). Drinking coffee during the day may lead to longer time before the general activation goes down, and may interfere with the restitution process. Some previous studies have not found any relationship between caffeine and cortisol in healthy volunteers (Spindel et al., 1984; Uhde et al., 1985; Zanoboni and Zanaboni Muciaccia, 1987), while other studies have found that caffeine may elevate cortisol secretion in humans at rest and during mental stress (Lovallo et al., 1996; Lane et al., 2002). There is also evidence of adaptation to coffee intake. Lovallo et al. (2005) found that healthy consumers taking repeated doses of caffeine during the day have statistically significant increases in cortisol secretion during the afternoon hours. The study also showed that cortisol responses to caffeine are reduced, but not eliminated, in healthy young men and women who consumed caffeine on a daily basis. In a recent animal study, injections of caffeine led to prolonged elevations in corticosterone, the highest doses yielding elevations for up to two hours (Patz et al., 2006). Low-to-moderate doses failed to modulate the corticosterone response to loud white noise, but modulatory effects of high caffeine doses were not tested. The interaction between stress factors and caffeine effect suggested from our studies is supported by findings from medical students. They got more effect from oral doses of caffeine (coffee or caffeine tablets) during an examination period than in periods without examination (Shepard et al., 2000). Clinically, persons with high risk of hypertension have more ARTICLE IN PRESS Coffee, stress and cortisol in nursing staff rapid and prolonged cortisol responses to caffeine than do low-risk persons (Al’Absi et al., 1998). The interaction between caffeine and stress or activation may improve performance but may also have negative health effects on blood pressure (Al’Absi et al., 1998; Shepard et al., 2000), or restitution, as in our study. However, other studies found that moderate consumption of both caffeinated and decaffeinated coffee may lower risk of type 2 diabetes in younger and middle-aged women (van Dam et al., 2006). This study showed that constituents other than caffeine may have a positive effect on health. In the present study we only have data on coffee but not other caffeine sources like tea or caffeinated soft drinks, so we cannot calculate the total caffeine intake. Situations where individuals are faced with strong stressors and only few possibilities to influence situations may lead to ‘‘sustained activation’’ and an inability to recover (Ursin and Eriksen, 2004). The need for recovery after a working day is one possibility to operationalise ‘‘sustained activation’’ (Kristenson et al., 2004). In our study there was a relationship between high score on ‘‘decision authority’’ (influence on the job, how to perform work and the freedom to make decisions) and low cortisol levels in the evening. This may be an indication of insufficient recovery and insufficient restitution (Sluiter et al., 1999). Insomnia gives an increased risk for permanent work disability (Sivertsen et al., 2006). Our findings support Theorell’s (Theorell et al., 2001) findings of a relationship between decision authority (control) and low cortisol in the evening. The results are also supported by the findings of Sluiter et al. (1999) showing low control and lack of recovery after work and Bollini et al. (2004) showing that more internal locus of control was related to a reduced cortisol response. The conclusions from the present study are limited by the small sample size, which reduces statistical power. However, the data was performed in a relative homogeneous healthy sample, collected at the same time of the year, almost in the same time in the morning on two consecutive working days. These facts reduced the influence of potential confounders and increase the reliability of the findings. 5. Conclusion In conclusion, the present study showed that high cortisol levels in the evening were related to coffee and to low scores on decision authority. Coffee consumption should therefore be controlled in future studies. The present study provides preliminary support for the idea that poor recovery may be an important factor for the relationship between sustained activation and pathology. Role of funding source The study was funded by grants from the University of Bergen, the Norwegian Research Council, and the Ministry of Health and Social Affairs through the Research Unit of the Norwegian Back Pain Network. They have had no further role in study design, data collection, analysis and interpretation of data, in the writing of the report or in the decision to submit the paper for publication. 329 Conflict of interest None declared. Acknowledgements The authors would like to thank Eli Nordeide and Randi F. Espelid for analysis of cortisol, Aud Skogen for administration of the data collection, Nina Konglevoll and Linda Sandal for technical assistance and Stein Atle Lie for assistance with the statistical analyses. References Al ‘Absi, M., Lovallo, W.R., 2004. Caffeine effects on the human stress axis. In: Nehlig, A. (Ed.), Coffee, Tea, Chocolate and the Brain. CRC Press, Boca Raton, FL, pp. 113–131. Al’Absi, M., McKey, B., Sung, B.H., Whitsett, T.L., Wilson, M.F., 1998. Hypothalamic–pituitary–adrenocortical responses to psychological stress and caffeine in men at high and low risk for hypertension. Psychosom. Med. 60, 521–527. Bollini, A.M., Walker, E.F., Hamann, S., Kestler, L., 2004. The influence of perceived control and locus of control on the cortisol and subjective responses to stress. Biol. Psychol. 67, 245–260. Clow, A., Thorn, L., Evans, P., Hucklebridge, F., 2004. The awakening cortisol response: methodological issues and significance. Stress 7, 29–37. Cooper, C.L., 1981. The Stress Check. Prentice Hall, New York. Eller, N.H., Netterstrøm, B., Hansen, Å.M., 2006. Psychosocial factors at home and at work and levels of salivary cortisol. Biol. Psychol., July 7 (Epub ahead of print). Endresen, I.M., Ellertsen, B., Endresen, C., Hjelmen, A.M., Matre, R., Ursin, H., 1991. Stress at work and psychological and immunological parameters in a group of Norwegian female bank employees. Work Stress 5, 217–227. Engstrom, M., Ljunggren, B., Lindqvist, R., Carlsson, M., 2006. Staff satisfaction with work, perceived quality of care and stress in elderly care: psychometric assessments and associations. J. Nurs. Manage. 14 (4), 318–328. Eriksen, H.R., Ursin, H., 2004. Subjective health complaints, sensitization, and sustained cognitive activation (stress). J. Psychosom. Res. 56, 445–448. Eriksen, H.R., Olff, M., Ursin, H., 1997. The CODE: a revised battery for coping and defence and its relations to subjective health. Scand. J. Psychol. 38, 175–182. Eriksen, H.R., Ihlebæk, C., Ursin, H., 1999. A scoring system for subjective health complaints (SHC). Scand. J. Public. Health 1, 63–72. Hjortskov, N., Garde, A.H., Ørbæk, P., Hansen, Å.M., 2004. Evaluation of salivary cortisol as a biomarker of self-reported mental stress in field studies. Stress Health 20, 91–98. Karasek, R., Theorell, T., 1990. Healthy Work: Stress, Productivity, and the Reconstruction of Working Life. Basic Books Inc. Publishers, New York. Kristenson, M., Eriksen, H.R., Sluiter, J.K., Starke, D., Ursin, H., 2004. Psychobiological mechanisms of socioeconomic differences in health. Soc. Sci. Med. 58, 1511–1522. Kudielka, B.M., Kirschbaum, C., 2003. Awakening cortisol responses are influenced by health status and awakening time but not by menstrual cycle phase. Psychoneuroendocrinology 28, 35–47. Kunz-Ebrecht, S., Kirschbaum, C., Marmot, M., Steptoe, A., 2004. Differences in cortisol awakening response on work days and weekends in women and men from the Whitehall II cohort. Psychoneuroendocrinology 29, 516–528. ARTICLE IN PRESS 330 Lane, J.D., Piepier, C.F., Phillips-Bute, B.G., Bryant, J.E., Kuhn, C.M., 2002. Caffeine affects cardiovascular and neuroendocrine activation a work and home. Psychosom. Med. 64, 595–603. Levine, S., Ursin, H., 1991. What is stress? In: Brown, M.R., Rivier, C., Koob, G. (Eds.), Stress, Neurobiology and Neuroendocrinology. Marcel Dekker, New York, pp. 3–21. Lindfors, P., Lundberg, U., 2002. Is low cortisol release an indicator of positive health? Stress Health 18, 153–160. Loge, J.H., Kaasa, S., Hjermstad, M.J., Kvien, T.K., 1998. Translation and performance of the Norwegian SF-36 Health Survey in patients with rheumatoid arthritis. I. Data quality, scaling assumptions, reliability, and construct validity. J. Clin. Epidemiol. 5, 1069–1076. Lovallo, W.R., Al’Absi, M., Blick, K., Whitsett, T., Wilson, M.F., 1996. Stress-like adrenocorticotropin responses to caffeine in young healthy men. Pharmacol. Biochem. Behav. 55, 365–369. Lovallo, W.R., Whitsett, T.L., Al’Absi, M., Sung, B.H., Vincent, A.S., Wilson, M.F., 2005. Caffeine stimulation of cortisol secretion across the waking hours in relation to caffeine intake levels. Psychosom. Med. 67, 734–739. Lundberg, U., Frankenhauser, M., 1999. Stress and workload of men and women in high ranking positions. J. Occup. Health Psychol. 4, 142–151. Lundberg, U., Hellstrøm, B., 2002. Workload and morning salivary cortisol in women. Work Stress 16, 356–363. McEwen, B.S., Seeman, T., 1999. Protective and damaging effects of mediators of stress: Elaborating and testing the concepts of allostasis and allostatic load. Ann. NY Acad. Sci. 896, 30–47. Patz, M.D., Day, H.E., Burow, A., Campeau, S., 2006. Modulation of the hypothalamo-pituitary-adrenocortical axis by caffeine. Psychoneuroendocrinology 31, 493–500. Pruessner, J.C., Wolf, O.T., Hellhammer, D.H., Buske-Kirschbaum, A., von Auer, K., Jobst, S., Kaspers, F., Kirschbaum, C., 1997. Free cortisol levels after awakening: a reliable biological marker for the assessment of adrenocortical activity. Life. Sci. 61, 2539–2549. Pruessner, J.C., Hellhammer, D.H., Kirschbaum, C., 1999. Burnout, perceived stress and cortisol responses to awakening. Psychosom. Med. 61, 197–204. Pruessner, M., Hellhammer, D.H., Pruessner, J.C., Lupien, S.J., 2003a. Self-reported depressive symptoms and stress levels in healthy young men: associations with the cortisol response to awakening. Psychosom. Med. 65, 92–99. Pruessner, J.C., Kirschbaum, C., Meinlschmid, G., Hellhammer, D.H., 2003b. Two formulas for computation of the area under the curve represent measures of total hormone concentration versus time-dependent change. Psychoneuroendocrinology 28, 916–931. Schreurs, P.J.G., Tellegen, B., Van De Willige, G., Brosschot, J.F., 1988. De Utreschtse Coping Lijst: Handleiding. Lisse, Swets en Zeitlinger. Schreurs, P.J.G., Van De Willige, G., Brosschot, J.F., Grau, G., 1993. De Utrechtse Copinglijst: UCL. Handleiding, second ed. Lisse, Swets en Zeitlinger. Seeman, T.E., Singer, B., Wilkinson, C.W., McEwen, B., 2001. Gender differences in age-related changes in HPA axis reactivity. Psychoneuroendocrinology 26, 225–240. Shepard, J.D., AlÀbsi, M., Whitsett, T.L., Passay, R.B., Lovallo, W.R., 2000. Additive pressor effects of caffeine and stress in male medical students at risk for hypertension. Am. J. Hypertens. 13, 475–481. Shilo, L., Sabbah, H., Hadari, R., Kovatz, S., Weinberg, U., Dolev, S., Dagan, Y., Shenkman, L., 2002. The effects of coffee consumption on sleep and melatonin secretion. Sleep Med. 3, 271–273. Siegrist, J., 1996. Adverse health effects of high-effort/low-reward conditions at work. J. Occup. Health Psychol. 1, 27–43. Sivertsen, B., Overland, S., Neckelmann, D., Glozier, N., Krokstad, S., Pallesen, S., Nordhus, I.H., Bjorvatn, B., Mykletun, A., 2006. View publication stats A. Harris et al. The long-term effect of insomnia on work disability. The HUNT-2 Historical Cohort Study. Am. J. Epidemiol. 163 (11), 1018–1024. Sluiter, J.K., Van der Beek, A.J., Frings-Dresen, M.H.W., 1998. Workstress and recovery measured by urinary catecholamines and cortisol excretion in long distance coach drivers. Occup. Environ. Med. 55, 407–413. Sluiter, J.K., Van der Beek, A.J., Fring-Dresen, M.H.W., 1999. The influence of work characteristics on the need of recovery and experienced health: a study on coach drivers. Ergonomics 42, 573–583. Sluiter, J.K., Croon, E., Meijman, T., Frings-Dresen, M., 2003. Need for recovery from work related fatigue and its role in the development and prediction of subjective health complaints. Occup. Environ. Med. 60, 62–70. Spindel, E.R., Wurtman, R.J., McCall, A., Carr, D.B., Conlay, L., Griffith, L., Arnold, M.A., 1984. Neuroendocrine effects of caffeine in normal subjects. Clin. Pharmacol. Ther. 36, 402–407. Steptoe, A., Ussher, M., 2006. Smoking, cortisol and nicotine. Int. J. Psychophysiol. 59 (3), 228–235. Steptoe, A., Cropley, M., Griffith, J., Kirschbaum, C., 2000. Job strain and anger expression predict early morning elevations in salivary cortisol. Psychosom. Med. 62, 286–292. Steptoe, A., Siegrist, J., Kirschbaum, C., Marmot, M., 2004. Effort–reward imbalance, overcommitment, and measures of cortisol and blood pressure over the working day. Psychosom. Med. 66, 323–329. Tabachnick, B.G., Fidell, L.S., 1996. Using Multivariate Statistics, third ed. Harper Collins, New York. Theorell, T., Michèlsen, H., Nordemar, R., Music 1 study group, 1993. Validitetsprøvning av psykososiala indexbildningar. In: Hagberg, M., Hogstedt, C. (Eds.), Stockholmsundersøkningen 1. Music Books, Stockholm, pp. 163–177. Theorell, T., Emdad, R., Arnetz, B., Weingarten, A.M., 2001. Employee effects of and educational program for managers at an insurance company. Psychosom. Med. 63, 724–733. Uhde, T.W., Bierer, L.M., Post, R.M., 1985. Caffeine-induced escape from dexamethasone suppression. Arch. Gen. Psychiatry 42 (7), 737–738. Ursin, H., 1988. Expectancy and activation: an attempt to systematize stress theory. In: Hellhammer, D., Florin, I., Weiner, H. (Eds.), Neurobiological Approaches to Human Disease. Hans Huber, Toronto, pp. 313–334. Ursin, H., 2000. Psychosomatic medicine: state of the art. Ann. Med. 32, 323–328. Ursin, H., Eriksen, H.R., 2004. The cognitive activation theory of stress. Psychoneuroendocrinology 29, 567–592. van Dam, R.M., Willett, W.C., Manson, J.E., Hu, F.B., 2006. Coffee, caffeine, and risk of type 2 diabetes. Diabetes Care 29, 398–403. Von Thiele, U., Lindfors, P., Lundberg, U., 2006. Self-rated recovery from work stress and allostatic load in women. J. Psychosom. Res. 61, 237–242. Ware, J.E., Gandek, B., IQOLA Project Group, 1994. The SF-36 health survey: development and use in mental health research and the IQOLA Project. Int. J. Ment. Health 23 (2), 49–73. Ware, J.E., Snow, K.K., Kosinski, M., Gandek, B., 2000. SF-36 Health Survey: Manual and Interpretation Guide. QualityMetric Incorporation, Lincoln, RI. Weitzman, E.D., Fukushima, D., Nogeire, C., Roffwarg, H., Gallagher, T.F., Hellman, L., 1971. Twenty-four hour pattern of the episodic secretion of cortisol in normal subjects. J. Clin. Endocrinol. Metab. 33 (1), 14–22. Wüst, S., Wolf, J., Hellhammer, D.H., Federenko, I., Schommer, N., Kirschbaum, C., 2000. The cortisol awakening response—normal values and confounds. Noise Health 2 (7), 79–88. Zanoboni, A., Zanaboni Muciaccia, W., 1987. Effects of naloxone and coffee on anterior pituitary hormones. Drugs Exp. Clin. Res. 13 (7), 443–446.