You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/232451778

Perceived stress and salivary cortisol in daily life

Article  in  Annals of Behavioral Medicine · January 1994

CITATIONS READS

69 685

2 authors, including:

Nancy Nicolson
Maastricht University
82 PUBLICATIONS   5,742 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Nancy Nicolson on 27 April 2016.

The user has requested enhancement of the downloaded file.


PERCEIVED STRESS AND SALIVARY CORTISOL IN DAII.,Y LIFE 1,z

Maneen M. van Eck, M.S.c and Nancy A. Nicolson, Ph.D.


University of Limburg

ABSTRACT abetes mellitus(8). The diversity ofstress concepts and research


methodologies used in these studies makes it difficult to integraté
Clarifying the nature of endocrine responses to chronic or all of the findings.
intermittent stress in daily life requires repeated measurements Since the beginning of this century, the endocrine system
oJstress, hormone levels, and emotional states. In this study, 42 has received much attention in stress research (9,10) with an
High Stress(HS)and 46 Low Stress(LS)subjects were selected
emphasis on the hypothalamicpituitary—adrenal (HPA) axis.
on the basis ofPerceived Stress Scale scoresfrom a largersample This system forms a theoretically promising pathway for me-
of male white-collar workers. Subjects completed self-reports diating the relationships between psychosocio] stress and sub-
[Experience Sampling Method(ESM)J and collected saliva sam- sequent disorder (1 1). Cortisol secretion appears to play an im-
plesjor cortisol determination ten times a day overfive consec- portant role in the regulation of physiological and behavioral
utivedays, incl:~ding three work and two non-work days. We test responses under stressful situations (12,13j. Cortisol increases
the hypothesis that high perceived stress is associated with ele- after strenuous physical exercise(14)and mental task load, spe-
vated cortisol. ci5cally in response to distress as opposed to effort or general
The HS group scored higher than the LS group on measures arousal (15,16). Cortisol is also the most important glucocor-
of trail anziet}~, depression, anger, and psychosomatic com- ticoid, with the potential of exerting its effects on practically all
plaints, as wel! as on ESM measures oJpositive and negative cells and tissues. Some of the most important effects of cortisol
affect and stressful daily events. Although negative a}ject was are stimulation ofgluconeogenesis,inhibition ofglucose uptake,
higher and stressful events morefrequent on workdays than week- suppression ofinflammation,and suppression of numerous ini-
ends (especially for HS subjects), no dijjerence in workday vs. mune functions. Glucocorticoids almost invariably suppress
weekend cortiso! levels was found in the subsample ofsubjects
á rather than enhance biological defense mechanisms. They are
with sujjrcienl data. On workdays, HS subjects had higher mean thought to protect against the normal defense reactions that are
cortisol levels than LS subjects at each ofthe ten sampling times activated by stress by preventing these defense reactions from
between 8 a.m. and 10 p.m. (repeated measures ANOVA, p <
overshooting and by preserving the specificity of immune
.02). Mean workday conisol wcu correlated with higher trait
reactions (17).
anxiety, depression, and the low arousal ~fimension ofESM neg- Indeeà, moss of the physiological reactions to stress are
ative a,~ect. _
thougt-~t to b:, adaptive, i.e. they counter the effects of physical
7 stress (trauma, bleeding), or are seen as a preparation íor fight
(Ann Behav Med 199A, 163):221-227)
or flight responses to acute danger (18). The question now is,
why and when could this adaptive mechanism be a pathway to
disease? There are indications that the same suppressive effects
INTRODUCT7[ON of glucocorticoids on several functions of the immune system
may leave a subject more susceptible to infections or even tu-
Over the last 20 years, a wide range of studies bas estab- mors(19). Other studies indicate that increased levels ofcortisol
lished an association between psychcsocial stress and a variety may be involved in the etiology ofcoronary heart diseases (20).
of psychological and physical disorders. The strengths of these The general idea is that stress causes an overproduction ofanti-
associations aze usually small and inconsistent, but stressful life inflammatory and immune suppressive actions, leading to ad-
events, for instance, have been linked to depression, neurotic verse consequences (e.g. breakdown of the immune system)
impairment, and other psychological symptomatology (1-3). (21,22). There is also evidence from animal studies for a gradual
Suess has also been linked empirically with increased suscep- loss ofadaptation within the pituitary-corticoadrenal response.
tibility to infectious disease (4), with other immune-related dis- Ifstress is prolonged, hypophyseal receptors appear to lose their
orders like recurrences of herpes infections (5) and asthmatic sensitivity to cortisol,and as a consequence, prolonged increases
exacerbations (6), with coronary heart disease (7), and with di- in cortisol tend to further slow the recovery of wrtisol after
stress because of the ineffectiveness of the negative feedback
loop (23,24). Recent theoretical overviews by Henry and Ste-
'We thank Dr. J. Sulon, University of Liege, for performing RIA ana]- phens (25) and Siegrist (26) are more or less in line with this
yses of salivary cortisol, and C. Dijlanan for assistance in all aspecu of concept of loss of adaptation.
the study. It is reasonable to assume that stress reactions will only
'..Funding was provided by the Netherlands Organization for Scientific lead todisease when they are prolonged or occur. very often.
Research(NWO)and the University of Limburg. t The effects of chronic or .internuttent stress on the HPA-axis,
Reprint Address: M. M.van Eck, Social Psychiatry Section, Department however, have received far less attention in the literature than
of Psychiatry and Neuropsychology, University of Limburg, P.O. Box the effects of acute suessors, and studies on naturally occurring
616, 6200 MD Maastricht, the Netherlands. stressors and their effects are even more rare. Data on cortisol
'~;"; m .1994 by The Society of Behavioral Medicine. levels during chronic psychosocial stress are inconsistent, wíth
222 ANNALS OF BEHAVIORAL MEDICINE van Eck and Nicolson

enhanced as well as decreased concentrations repoRed,and large but not in evening or overnight urine samples (40). Depressive
variability among individuals. Elevated levels of cortisol were symptoms could also be related to increased coRisol secretion
found in subjects living near the damaged nuclear power plant at specific times ofthe day. Concerning day-of-the-week effects,
at Three Mile Island (27). Higher levels of urinary coRisol cor- we hypothesized that cortisol would be elevated on workdays
related significantly with urinary catecholamines, self-reports of compared to the weekend,possibly in interaction with perceived
physical and mental symptoms, and decrements in task perfor- stress. Although- subject groups were selected according to a
mance. Astudy following parents ofleukemic children into the global measure of stress and not on the basis of work-related
period of bereavement for as long as three years after the death stress, we reasoned that work might be a major source of stress.
ofthe child showed that l 7-hydroxycoRicosteroid excretion rates Studies by Frankenhaeuser and colleagues (41) have demon-
were related to the effectiveness of subjects' psychological de- stratedthat cardiovascular and neurcendocrine activity are higher
fenses (28,29). Elevated levels of plasma and salivary cortisol at work than at home, and that slow recovery of responses, or
were found in Iran hostages after release from captivity, reflect- "unwinding," after work might represent along-term health
ing distress, anxiety, and elation. Salivary cortisol was the only risk.
physiological measurement that demonstrated a significant cor- Finally, we examined the extent to which psychological
relation with psychiatrists' ratings ofthe hostages' psychological variables (trait amciety, depression, anger, and psychosomatic
disturbance (30). complaints; ESM measures of mood and stressful events) might
The above studies have examined stress under rather ex- contribute to individual differences in workday cortisol level.
treme or unusual situations. With respect to more normal set-
SUBJECTS
tings, studies of air traffic controllers suggest that occupational
stress is associated with increased cortisol production (31,32). Local industries and government agencies were approached
Caplan et al. (33) studied white-collar workers in a relatively via their personnel departments to participate in the study. A
everyday organizational setting. While there was no main effect decision as to which categories of employees fell under the def-
of work load or personality traits on mean cortisol, high per- iiution:of "white-collar" was made by each personnel depart-
ceived work load was associated with lower morning coRisol ment on the basis of standard job function descriptions. Ques-
values. tionnaireswere distributed among these employees,accompanied
The results presented here are part of a larger study con- by a letter explaining the goals of the study. Participation was
cerningemotional and hormonal responses[o naturally occuring voluntary, and care was taken to insure anonymity. Three hun-
stress in daily life. Stress is cuirentiy conceptualized as a dy- 3red sixteen male employees from six different industries or
namic process, whi~•h changes over time and in relation to the agencies completed the screening questionnaire. The mean score
environment. It is therefore necessary to include repeated mea- on !hc Perceived Stress Scale(PSS)for this sample was 12.7 (sd
surements of stress, hormone levels, and emotional states to = 6.i1), similar to U.S. norms(mean 13.02, sd = 6.45)(38).
investigate the stress—hormone relationship. This relationship Ninety-two subjects with scores in the upper or lower ter-
should ideally be studied in naturalistic settings and over time, tiles of the screening sample distribution (PSS-10 score <_ 10 or
so that we can begin to understand how adaptation occurs in >_ 1 b) were recruited to participate in the study, excluding any
response to intermittent daily hassles in individuals who are individuals with a history of endocrine disorder, medications
observed in their normal social networks,settings,and activities known to affect cortisol levels, psychopathology(major affective
(34). For these reasons, we chose to use two relatively new disorder, psychoses), or currently in treatment for mental health
techniques, Experience Sampling Method (ESM)(35,36) and problems. High- and low-stress subjects were matched for age
monitoring of salivary cortisol (37), to assess the relationship group, marital status, and household composition. Because it
between an indiviáual's affective state and neuroendocrine was not possible to match all subjects, numbers of subjects in
changes over time. Subjects were selected on the basis of Per- the final two groups differ slightly.
ceivedStress Scale scores(3H)from a large sample of male white- Four subjects failed to meet Experience Sampling compli-
collar workers and were then monitored during their daily ac- ance criteria (see Daily Experience section, below) and were
tivities. Subjects completed self-repoRs (ESM) and collected therefore excluded from further analysis. Of the 88 remaining
saliva samples for cortisol determination at semi-random in- subjects, 42 subjects comprised the High-Stress(HS)group and
tervals ten times a day over five consecutive days, including 46 subjects the Low-Stress(LS)group. Mean age was 42.1 years
three work and two non-work days. (range 27 to 57 years), with no significant difference between
We test the hypothesis that high perceived stress is asso- the two groups. Eighty-nine percent were marred,and 81%had
ciated with elevated cortisol, either in general, or during certain children living at home.
phases of the circadian cycle or certain days of the week (e.g.
work versus non-work days). With respect to diurnal patterns, ME.~SURES
no specific hypothesis was formulated. The few studies in which Questionnaires
conisol has been measured at more than one time of day in In addition to the PSS, questionnaires concerning psycho-
healthy subjects report significant relationships between cortisol social stress, coping style, and psychological and physical symp-
and psychological trait measures at some time points and not toms were completed. Self-report instruments were chosen on
at others. For example, in a large German survey study in which the basis of their theoretical relevance to the stress process and
salivary cortisol was measured three times on a single day, only psychometric reliability and validity. The following measures
8 a.m. levels showed clear associations with psychological and are used in the current analysis:
demographic variables(39).In a similar fashion, perceived work
load had an effect on cortisol levels in white-collar workers that Perceived Stress: The 10-item version ofthe PSS was trans-
was dependent on the time of day when blood was sampled lated into Dutch by the first author and then back-translated as
(33), and hostility was associated with high cortisol in daytime, a check into English by the second author. The items were rated
r

VOLUME 16, NUMBER 3, 1994 223


Stress and Cortisol

missing and invalid responses on weekends could be attributed


~n a five-point frequency scale, ranging from 0 = "never," to 4
="very often" in the last month. A total perceived stress score to the factthatsubjects veere still asleep.
The ESM form contained open-ended questions concerning
avas obtained by reversing the scoring on the positive items and
thought content, the physical and social context, and what the
Then summing across the ten items. All subjects completed the
individual was doing when signalled. The forms also included
PSS twice, during the initial screening and again immediately
preceding Experience Sampling. The two PSS scores were highly seven-point Likert scales (from 1 = "not at all;" to 7 = "very
much"), for rating aspects of thoughts, mood, physical well-
.correlated (rho = .73, p c .001); the mean score will be used
being, individually defined (psycho)somatic complaints, and
in the analysis.
present activity. Subjects were asked to describe any stressful
Psychosomatic Symptoms: The SUNYA Psychosomatic events or situations which may have taken place in the interval
Symptom Checklist(PSC)includes 17 common psychosomatic since the last ESM report and to rate these events on a number
complaints (e.g. headaches, backaches, and nausea). The scale of dimensions. Information about smoking (49), food (50), cof-
was originally developed by Cox, Freundlich, and Meyer (42) fee (51), and alcohol intake, medications, and physical exertion
and revised by Attanasio et al.(43). Subjects rate each complaint (52) since the last beep was also obtained, to help control for
on 5-point scales for frequency (0 = "never or rarely occurs;" possible confounding influences on cortisol secretion.
to 4 = "occurs daily") and intensity(0 = "not bothersome;" to
4="extremely bothersome"). A Total Score,reflecting the over- Salivary Cortisol
all level of psychosomatic distress, is obtained by summing the Salivary cortisol is a reliable and valid indicator of the free
cross-products of each item's frequency by intensity. cortisol in plasma, which is considered to be the biologically
Depression: Depressive symptomatology was assessed with active hormone. Cortisol concentrations aze independent ofthe
the Dutch translarion (44) of the Zung Self-Rating Depression flow rate of saliva (53). Salivary cortisol increases within min-
Scale (45). utes in response to acute stressons (54) and has ahalf-life of
approximately one hour(55). We found no difference in cortisol
Anxiety: Trait anxiety was measured with the Dutch ver-
levels in saliva samples either frozen immediately or kept at
sion (46) of the State-Trait Anxiety Inventory (STAD.
room temperature for two days(56); others report no change in
Anger: Trait anger was measured with the Dutch version unfrozen samples up to 30 days (57).
(47)ofthe Spielberger Trait Anger Scale. Although two subscales At the same time ESM forms were being completed, sub-
exist("anger-temperament" and "anger-reaction"), only the to- jects collected saliva by holding a cotton dental roll in the mouth
tal score is used here. for approximately one minute. The saturated roll was placed in
Daily Experience a capped plastic vial (Salivette; Sarstedt), which was stored in
a specially designed wallet. At the end of each day, subjects
The Experience Sampling Method (35,36,48) was used to
placed the vials in their home freezers. At the end of the sam-
collect data from subjects at selected moments during their nor-
plingperiod, uncentrifuged samples were stored at -20 C. Com-
mal daily activities. Subjects received auditory signals, after
pliance with the saliva sampling procedure »vas good in both
which they filled in a questionnaire and collected asaliva sám-
high and low. stress.groups, with the same mean response rate
ple. Although additional reports were completed at the end. of
(83%) as for the ESM reports.
each day, for the purpose of this analysis we will refer only to
Salivary cortisol levels were determined in duplicate by
the "beep level" data.
direct radioimmunoassay (58), using123I-cortisol (Panmos diag-
After a briefing session, in which all procedures were ex-
nostica, Finland) and antiserum madeagainst the 3-CMO-BSA
plained in detail and informed consent was obtained, subjects
conjugate by Dr. J. Sulon, University of Liege, Belgium. The
- were sampled for a period of five consecutive days, begnning
In lower detection limit of the assay was 12 ng/dl, with a mean
early on a Thursday morning and ending late on Mónday.
non-w orkdays. A Seiko wrist- intra-assay coefficient of variation of 4.8% (range: 2.2%-7.5%
this sample, the weekends were
at for fow assays). Each subjects samples were analyzed in .the
watch was programmed to emit beeps ten rimes each day, '
tely 90 minutes between same assay to reduce sources of variability.
.t semi-random intervals of approxima
~;, ._ the hours of8 a.m. and 10 p.m. Beeps were clustered around statistical Analysis
11:15
the midpoint ofeach time block (e.g. 8:15 a.m.,9:45 a.m.,
on); the exact time sequence of beeps was varied The 16 ESM mood items were reduced to three mood mea-
a.m. and so
session, sures, based on the results of a principal component analysis
éach day to decrease predictability. In afinal debriefing
for legibility, and subjects were asked with varimax rotation, which accounted for 78% of the total
_ESM forms were checked
variance when subject mean scores were used. Ratings on Zhe
- - to clarify reasons for missing data.
items cheerful, satisfied, relaxed,-energetic, self-assured., con-
Compliance with the procedures was generally -good. The
reports centrated, and enthusiastic were summed to form a Posirive
criteria we set for inclusion in the analysis (>_20 ESM
after being signalled and no miss- AfFect(PA)scale (Cronbaoh'salpha = :95). Two separate com-
completed within 20 minutes
entire days) were met by all but four subjecu (two ponents of negative affect (NA) were identified: "NA-]ow
ing data for
88 subjecu completed an av- arousal," including the items depressed,anxious, worried,lonely,
from each group). The remaining
of all possible responses within the time limit, for tired, and miserable(alpha =.87),and"NA-high arousal," with
erage of 83°ró
HS and IS groups did the items restless, irritated, hurried, and nervous(alpha = .93).
àn average of ~41 responses per subject.
per sub- The sums of the scale items were divided -by •the number of
not differ in compliance rates(40.1 vs. 42.3 responses
Complian ce was lowest items in the scale, so that all mood measures have ranges from
jest, Mann-Whitney U-test, p > .OS). 1 to 7.
approxima tely 8:15 a.m:), with an average
for the first beep (at
for first beeps Because the distribution of cortisol values was positively
of73%valid reports. On Saturdays,response rate
59°rb. Seventy-f our percent of all skewed, isansformation to natural logarithms of cortisol con-
~:. was 61%, and on Sundays,
224 ANNALS OF BEHAVIORAL MEDICINE van Eck and Nicolson

T~+~BLE 1 -~- hlgh alrni


Differences in Psychological Measures behveen Low and -.- ww eras
High Sáess Groups

Low PSS High PSS


Mean Mean p-value
(st. dev.) (st. dev.) (2-tailed)

Trait measures
Perceived Stress
(p$S) 7.2(2.2) 18.1 (3.4) <.000I
Trait amciety 28.3(4.4) 39.8 (7.6) <.0001
Trait anger 18.8 (4.6) 23.0(5.2) <.0001
Zung Depression 36.5 (5.5) 48.4(7.7) <.00O l
Psychosomatic
Symptoms(PSC) 6.1 (5.6) 27.5 (23.2) <.0001 ~~r~

ESM measures'
pp 5.7 (0.6) 5.3(03) <.00 I
NA-low arousal 1.2 (0.1) l.4 (0.4) <.0001 Q L
<.0001 a.~s 9.45 11.15 12.45 14.15 15.45 17.15 78.45 20.75 21.45
NA-high azousal l.3 (0.3) 1.& (0.6)
TIME OF DAY
Frequency of
stressful events 12.3%(11.91 22.0%(20.0) <.01 FIGURE 1: Mean cortisol (ng/dl; ±s.e.) during workdays
for the High Stress group(N = 34) and the Low Stress group
'Subject means, aggregated From beep-level data.
(N = 39).

centrations was performed prior to statistical analysis. Non-


parametric tests were ~~sed for univariate group(Mann-Whiney were rejected because of missing data for any ofthe 20(10 time
U-testj and within-sub;ect{Wilcoxon watched-pau-s signed-rank blocks x Z day types) measures. The remaining 49 subjects
:est)comp~zsons. Unless noted otherwise, significance tests are included 28 LS and 21 HS subjects.
two-tailed. Analysis of variance (SPSS procedure MANOVA) Cortisol concentrations showed the well-defined circadian
was used to test differences (and interactions) between HS and rhythm with peak salivary concentrations in the early morning
LS groups, workdays and weekends, and the ten times of day samples, declining values in the afternoon, and lowest concen-
sampled. Analyses were performed with SPSS-Macintosh trations at the end of the day. The main effect tór time of day
version. was highly significant (F(9,423) = 182.04, p < .001), with no
significant interactions with either group or day type. Contrary
RESULTS to our hypothesis, no significant main effect for group was found.
Given the general nature of the PSS, which was used to Further, there was no workday-weekend difference in cortisol
define high and low stress groups in this study, it is useful to levels, either as a main effect or in interaction with group.
contrast the two groups on more specific measures of distress. As noted earlier (see Daily Experience, above) the per-
In addition to trait measures, we compared groups on a number centage of missing responses was higher on weekends than on
of aggregated state measures derived from the ESM data. Mean workdays, especially for early morning samples. Inclusion of
scores on each of the three ESM mood scales were computed weekend data ín the model is thus largely responsible for the
across a(1 valid beeps for each subject. Frequency of stressful considerable data attrition. We therefore chose to examine
daily events was calculated as the percentage of total beeps for workdays and weekends again separately. While this does not
a subject on which a stressful event was reported, regardless of address the issue of workday vs. weekend differences, it allows
rated severity. us to test for group differences in cortisol in a larger sample.
As shown in Table 1, subjects with a high level ofperceived For workdays, I S subjects were forced out of the analysis
stress were significantly more trait aiixious and depressed than due to missing data, leaving 39 LS and 34 HS subjects. Time
subjects with a relatively lower level of perceived stress. They ofday again showed the lazgest effect on cortisol levels(F(9,639)
also scored higher on trait anger and repoRed more psychoso- = 262.98,. p < .001), but here the difference between HS and
matic symptoms. HS subjects had significantly lower ESM pos- LS groups was also significant (F(1,71) = 6.02, p < .02). The
itive affect, higher negative affect-low arousal, and higher neg- interaction effect was not significant (F(9,639) _ .67, p > .OS).
ative affect-high arousal scores. They were almost twice as likely Cortisol concentrations for HS and LS groups on workdays are
to report stressful daily events as LS subjects. In absolute terms, shown in Figure 1.
the mean number ofstressful events reported over the five days In the analysis ofdifferences in cortisol levels on weekends,
of ESM was 9.0 for the HS group and 5.3 for the LS group. only 56 subjects (32. LS and 24 HS) had sufficient data for
We next examined whether high perceived stress was as- inclusion. For this subsample, the usual time of day effect was
sociated with elevated salivary cortisol. Mean cortisol levels found (F(9,486)= 85.80, p < .001), but there was no consistent
were first computed for each subject at each of the ten time difference between the two stress groups (F(1,54) _ .79, p >
blocks for workdays and again for weekend days. To test for .OS). The group by time of day interaction approached signifi-
main effects on cortisol levels as well as interactions, analysis cance (F(9,486) = 1.83, p = .06), with a tendency for the high
ofvariance was performed with the between-subject factor group suess group to have higher cortisol levels in the morning through
(HS, I.S) and within-subject factors time ofday(ten beeps) and eaziy afternoon (beeps 2, 3, 4, and 5, between 9:30 a.m. and
day type(workday, weekend). Thirty-nine ofthe 88 subject cases 2:30 p.m.)and in the late evening(beeps 9 and !0, after 8 p.m.)
Stress and Cortisol VOLUME 16, NUMBER 3, 1994 225

only. Again, it is important to note that this (non-significant) TABLE 2


pattern is based on a relatively small subset ofthe entire sample. Correlations between Psychological Variables and Mean Salivary
To summarize, cortisol levels showed the expected strong Cortisol during Workdays
diurnal pattern in all analyses we performed. More interestingly,
high stress subjects had elevated cor[isol throughout the day on r N
workdays, confirming our hypothesis, at least for these days. It Trait anxiety .18" 88
would seem logical to assume that such a pattern would be Trait anger .08 88
mediated by group differences in stress exposure or, more spe- Zung Depression .18' $7
cifically, in emotional response to daily stress. Questions con- Psychosomatic Symptoms
cerning reactivity of mood and cortisol to stressful events are (PSC) .14 88
PA —.10 87
beyond the scope ofthis paper. However, measures of workday
NA —low arousal .20• 87
stress and distress could provide additional insights into the NA —high arousal .08 87
perceived stress--cortisol relationship.
Within subjects, workdays in comparison to weekends were p < .OS (Spearman; one-tailed tesu).
associated with a higher rate of stressful events (Wilcoxon
matched-pairs test, p c .0001) and greater NA-high arousal (p
< .0001), with no differences in either NA-low arousal or PA.
The coded descriptions of reported stressors suggested that the DISCUSSION
work environment was a major arena for such daily hassles, In this study, two groups of male white-collar workers dif-
with stressons more often occurring in work than in household, fering in levels of perceived soesa were compared on subjective,
leisure, or social network contexts. Forty-three percent ofevents self-report measures ofdistress and on levels ofsalivary cortisol,
in LS subjects and 47% of events in HS subjects were work- as measured repeatedly over five days. The High Stress group
related. While this general pattern ofworkday distress was found scored higher than the Low Stress group on measures of trait
in both groups, within-subject differences in workday vs. week- anxiety, depression, anger, and psychosomatic complaints, as
end NA-high arousal were greater in HS than in LS subjects well as on ESM measures of positive and negative affect and
(Mann-Whitney, p < .001); that is, HS subjects were dispro- stressful daily events. Cortisol showed a clear diurnal secretory
portionately more negatively aroused on workdays. Workday/ pattern in both groups, with values in the normal range for
weekend differentials in NA-low (.OS c p < .10) and stressful healthy men (37). On workdays, cortisol was elevated through-
events (.OS < p c .10) did not distinguish the two groups, out the day in the High Stress relative to the Low Stress group.
although results were in the expected direction of relatively Several measures of distress, including both trait (anxiety, de-
greater work-related stress/distress in HS subjects. pression) and concurrent state(NA-low arousal) measures, cor-
Since HS subjects scored higher than LS subjects on trait related significantly with mean workday cortisol.
anxiety, depression, anger, and psychosomatic complaints, as The observed pattern of higher workday cortisol levels in
well as. on ESM measures of positive anti negative affect, we the High Stress group is consistent with our initial hypothesis;
expected positive relationships between these variables and cor- a straightforward interpretation would be that mild chronic or
tisol levels, at least on workdays. Mean levels of cortisol over intermittent stress in situations like those subjects reported dur-
the three workdays were determined by first calculating the mean ing the ESM sampling period was sufficient to increase secretion
at each of the ten time blocks for every subject and then av- ofthe hormone. Since the half-life ofcortisol in saliva is shorter
eraging the ten values for each subject. This aggregated measurer,..' than the intervals between beeps, consistently elevated cortisol
of mean cortisol was slightly higher in the HS (116 ng/dl, sd from morning through late evening indicates continued secre-
37.5)compared to the LS group(101 ng/dl, sd 28.2)(one-tailed tion. More detailed temporal analyses are needed, however, to
Mann-Whitney test, p c .OS). understand cortisol dynamics in response to daily svess.
Spearman rank order correlations between mean workday Of the few studies of stress and cortisol in everyday envi-
conisol and both psychological measures and ESM mood scales ronments to date, results have been inconsistent. Caplan et al.
are shown in Table 2. Small but significant (one-tailed tests) (33)tested the effect ofperceived work load(a summary measure
positive correlations were found between cortisol and trait anx- of quantity, deadlines, and calls and office visits) on plasma
iety, depressive symptomatology, and NA-low arousal. cortisol and found that high work load was associated with low
Although these results indirectly support the conclusion morning cortisol levels. The authors suggested that this finding
that daily soesa and distress lead to increases in cortisol, the might reflect a shift in the circadian rhythm ofcortisol as a result
role of possible confounding factors should be explored. If, for ofjob soesa. A major disadvantage of this study is the cross-
example, HS subjects smoked more than IS subjects, this might sectional design, in which blood samples were drawn at various
explain the observed differences in cortisol secretion. HS and times of day, with only one cortisol determination per subject.
LS groups were therefore compared on the percentage of work- In other words, comparisons between morning, midday, and
àay beeps with reported smoking, strenuous physical exertion afrernoon cortisol concentrations were comparisons between
(rated 7 on the 7-point scale), coffee, food,and alcohol intake— subjects and not within subjects. Although perceived stress may
all activities that could theoretically result in elevated cortisol differ in important ways from perceived work load, our results
secretion (see Daily Experience, above). No significant differ- clearly contradict those of Caplan and colleagues; we found
ences were found, with mean rates for HS and LS groups as higher early morning cortisol in high stress subjects and no
follows(HS/IS): smoking 17~k/16%,exertion 0.7%/1.3%,cofïee evidence at all for a circadian shift. Brandtst~dter et al. (39),
46%/41%,food 42%/39%,and alcoho18.0°x/7.8%. Although we who measured salivary cortisol concentration in 767 adults three
cannot conclude that these behaviors had no influence on cor- times over the course ofone day,found higher morning cortisol
tisol secretion, effects seem equally likely to have occurred in concentrations in men with high lífe satisfaction, high psycho-
either group. logical well-being, and high level ofemployment. Cortisol levels
226 ANNAIS OF BEHAVIORAI. MEDICINE vair Eck and Nicolson

(5) Hoon EF, Hoon PW,Rand KH,et al: A psychobehavioral model


in the afternoon or early evening were not predicted by psy-
of genital herpes recurrence. Journal ojPsychosomatic Research.
chological variables. Again,these results are difficult to reconcile
1991, 35:25-36.
with our findings. (6) Isenberg SA, Lehrer PM,Hochron S:The effects ofsuggestion and
Trait negarive affectivity(NA)is conceptualized as the ten- emotional arousal on pulmonary function in asthma: A review and
dency to experience a wide range of negative emotions (59). In a hypothesis regarding valtal mediation. Psychosomatic Medicine.
studies of stress-illness relationships, trait NA has been held 1992, 54:192-216.
responsible for inflated correlations between stressons and self- (7) Bassett JR:Psychologicalstress and the coronary artery in ischemic
reported health symptoms (60,61). In general, NA has not been heart disease. In Kalmer S (ed} The Coronary Anery. London:
found to predict objective health problems or physiological stress Croom-Helm; 1982, 474-500.
responses and has therefore been considered a nuisance in the (8) Gcetsch VL,Vlliebe DJ, Veltum LG,et al: Case history and shorter
communicatioa..BehaviorResearch and Therapy. 1990, 28(6):531-
analysis of stress data (62). The High Stress group in our study
53T:
was defined on the basis of scores on a very clearly subjective
(9) Selye H: A syndrome produced by diverse noxious agents. Nature.
self-report measure, the Perceived Stress Scale, and the fact that 1936, 138:32-36.
HS subjects scored higher on state and trait measures ofdistress (10) Mason JW:A review ofpsychoendocrine reseazch on the pituitary-
as well as self-reported stress exposure suggests the pervasive adrenal cortical system. Psychosomatic ~Lledicine. 1968, 30:576-
influence oftrait NA.In this light, our findings ofan association 607.
between subjective stress, trait and state NA measures, and (I1) Curtis GC: Psychosomatics and chronobiology: Possible impli-
salivary cortisol levels is particularly noteworthy. cations of neuroendocrine rhythms.Psychosomatic Medicine. i 972,
As is usually the case in ps~c:hophysiological studies, the 34(3):235-256.
psychological variables we measured accounted for only a small (12) Mason JW: A historical view of the stress field. Journal ojHuman
Stress. 1975, 1(1):7-12.
percentage ofthe variability in cortisol. Moreover, the observed
(13) Rose RM: Overview of endocrinology of stress. In Brown GM,
elevations in workday cortisol secretion in HS subjects were Koslow SH, Reichlin S (eds), Neuroendocrinology and Psychiatric
slight, relative to pathophysiological levels seen in Cushing's Disorder. New York: Raven Press, 1984, 95-122.
disease or some cases of major depression. Since cortisol levels (14) Mason JW, Hartley H, Kotchen TA, et ai: Plasma coRisol and
are not routinely measured in prospective eoidemiolugical stud- norepinepluine responses in anticipation of muscular exercise.
ies of healthy individuals, we have no guidelines at present for Psychosomatic ~Nedicine. [973, 35(5):406-~tt4.
assessing the clinical relevance or health implications o:these (15) Lundberg U, Frankenhaeuser M: Pituitary-adrenal and sympa-
findings. thetio-adrenal wrrelates óf distoss and effon. Journal ojPsycho-
Analysés in this study were done on aggregated data, which somatic Research. 1980, 14:125-130.
has the advantage of increasing the reliability of measurement (16) Fibiger W, Singer G: Biochemical assessment and differentiation
of mental and physical effort. Work and ,Stress. 1989, 3(3)237-
when only the chronic level of variables is of interest. On the
247.
other hand, aggregated measures may be based (63), and they (17) Munck A, Guyre PM, Holbrook NJ: Physiological functions of
obfuscate the important interplay between experiential and glucocorticoids in stress and their relation to pharmacological ac-
physiological states and processes which momentary measures tions. Endocrine Reviews. 1984, S(I):25-44.
are intended to clarify. To gain more insight into the relationship (18) Cannon WB: Organization for physiological homeostasis. Physi-
between perceived stress and cortisol, we need to move to beep- ological Review. 1929, 9:399.
level analyses. We can then investigate fluctuations over time (19) Claman HN: CoRicosteroids and lymphoid cells. New England
in stressful events, mood, and psychosomatic complainu and Journal ojMedicine. 1977, 287:388-397.
determine how these Ructuations relate to endocrine activity (20) Troxler RG, Sprague EA, Albanese RA, et al: 1'he association of
(concurrent and lagged relationships). One promising method elevated plasma cortisol and early atherosclerosis as demonsUated
by coronary angiography. Atherosclerosis. 1977, 26:151-162.
for the analyses oftemporal data is hierarchical linear modeling
(21) Selye H: The Stress ofLije. (Rev. Ed.). New-York: McGraw-Hill,
or multi-level analysis (64). This approach can account for the
1976.
dependency of data within a subject, can deal with the problem (22) Gaillazd R-C, Al-Damluji S: Stress and the pituitary-adrenal axís.
of missing data, and allows for individual differences in inter- Ballière's Clinical Endocrino%gy and Metabolism. I987,1(2):319-
cepts, slopes, and error structures. We are currently undertaking 354.
multi-level analyses, in the belief that these new methods will (23) Bassett JR, Cairncross KD: Changes in the coronary vasculaz sys-
help realize the enormous research pot~tial afforded by mo- tem following prolonged exposure to stress. Pharmacology, Bio-
mentary measures. chemrstry and Behavior. 1977, 6:311-318.
(24) Kloet ERD, Reul JMHM: Feedback acrion and tonic inNuence of
corticosteroids on brain function: A concept arising &om the het-
erogeneity of brain receptor systems. Psychoneuroendocnnology.
REFERENCES
1987, 12:85-105.
(1) Dohrenwend BS, Dohrenwend BP: Overview and prospects For (25) Henry JP, Stephens PM: Stress, Health and the Sxra[ Environ-
research on stressfiil life events. In Dohrenwend BS, Dohrenwend ment:A Sociobio[ogic Approach to Medicine. New York: Springer-
BP(eds),Stressful LifeEvents: Their Nature and Effects. New York: Verlag, 1977.
Wiley &Sons, 1974, 313-331. (26) Siegrist J, Klein D, Matschinger H: Occupational stress, coronary
(2) Dohrenwend BS, Dohrenwend BP: Some issues in research on risk factors, and cardiovascular responsiveness. In Weiner H,Florin
stressfut life events. Journal ofNervous and Mental Disease. 1978, I, Murison R, Hellhammer D (eds), Frontiers ofStress Research.
166(1):7-I 5. Toronto, Canada: Huber, 1989, 323-335.
(3) Brown GW, Harris TO (eds): Life Events and Illness. London: (2~ SchaefferMA, Baum A:Adrenal cortical response to stress at Tóree
Guilford Pass; 1989. Mile Island Journal ofNervous and Mental Disease. 1984, 46(3):
(4) Cohen S, Tyrrell D,~T, Smith AP: Psychological stress and sus- 227-237.
ceptibility to the common cold. New England Journal ofMedicine. (28) Hofer MA,Wo1ffCI',Friedman SB,et al: A psychcendocríne study
1991, 325(9):606--612. of bereavement. Part I. 17-Hydroxycorticosteroid excretion rates
icoLson Stress and Cortisol VOLUME 16, NUMBER 3, 1994 227

~i model of parenu following death of their children from leukemia. Psy- j47) Ploeg HMvd,Defares PB,Spielbeiger CD:Handleiding bijde Zelf-
esearch. ~: chosomatic Medicine. 1972, 34(6):48191. Analyse Vragenlijs!- ZAV- een Nederlandstalige Bewerking van de
(29) Hofer MA,Wo1ffCf,Friedman SB,et al: A psychcendocrine study SpielóergerStale-Trait Anger Scale. Lisse, The Netherlands: Swets
lion and of bereavement. Part II. Observations on the process of mourning & Zeidinger, 1982.
iew and in relation to adrenocortical function. Psychosomatic Medicine. (48) de Vries MW:Introduction: Investigating mental disorders in their
!edicine. 1972, 34(6):492-504. natural settings. Journal of Nervous and Mental Disease. 1987,
(30) Rape RH, Karson S, Howard NJ, et al: Psychological and physi- 175(9):509-513.
schemic ologcal assessments on American hostages freed from captivity (49) Wust S, Kirschbaum C, Hellhammer D: Smoking increases sali-
.,ondon: in Iran. PsvchosomaticMedicine. 1990, 52(1):1-16. vary cortisol. Journal ofClinical Chemistry and Clinical Biochem-
(31) Rose RM,Jenkins CD, Hurst M, et al: Endocrine activity in air istry. 1990, 28.'665-666.
shorter traffic conunllers at work. III. Relationship to physical and psy- (50) Quigley ME, Yen SSC: A midday surge in cortisol levels. Journal
(6):531- chiatric morbidity. Psychoneuroendocrinology. 1982, 7(2):125-134. ojClinical Endocrinology and Metabolism. 1979, 49:945-947.
(32) Rose RM,Fogg LF: Definition of a responder. Analyses of behav- (51) Pincomb GA, Lovallo WR, Passey WR, et al: Caffeine enhances
Nature. ioral, cardiovascular, and endocrine responses to varied workload the physiological response to occupational svess in medical stu-
in air tra~c controllers. Psychosomatic Medicine. 1993, 55:325- dents. Health Psychology. 1987, 6:101-112.
tuitary- 338. (52) Cook NJ, Ng A, Read GF, et al: Salivary cortisol for monitoring
30:57Cr (33) Caplan RD, Cobb S, French Jr 7RP: White collar work load and adrenal activity during marathon runs. Hormone Research. 1987,
cortisol: Disruption of a circadian rhythm by job stress? Journal 25:18-23.
i~p~_ o(Psvchosomatic Research. 1979, 2.3:181-192. (53) Vining RF,McGinley RA,Symons RG:Hormones in saliva: Mode
e. 1972, (34) Dimsdale JE: Generalizing from laboratory to field studies of hu- of entry and wnsequent implications for clinical interpretation.
man stress physiology. Psychosomatic Medicine. 1984,46(5):463- Clinical Chemistry. 1983, 29(10):1752-1756.
Human 469. (54) Hellhammer DH,Kirschbaum C, Belkien L: Measurement of sal-
(35) Csikszentmihalyi M, Larson R: Validity and reliability of the ex- ivary cortisolunder psychological stimulation. In Hingtgen J, I3e11-
~r GM, periencesampling method.JournalofNervous and MentalDisease. hammer D, Huppmann G (eds), Advanced Methods in Psychobi-
chiatric 1987, 175(9):526-536. ology. Toronto, Canada: C. J. Ho~efe Inc., 1987, 281-289.
(36) de Vries MW: The Experience ofPsychopathology: Investigating (55) Fredrikson M, Sundin ~, Frankenhaeuser M: Cortisol excretion
sol and MentalDisordersin Their Natural Settings. Cambridge,UK:Cam- during the defense reaction in humans. Psychosomatic Medicine.
xercise. bridge University Press, 1992. 1985, 47(4):313-319.
(37) Kirschbaum C, Hellhammer DH: Salivary cortisol in psychobio- (56) Nicolson NA, van Poll R,de Vries M: Ambulatory mónitoring of
sympa- lo~cal research: An overview. Neuropsychobiology. 1989,11:15Q salivary cortisol and stress in daily life. In Kirschbaum C, Read
Psycho- 169. GF, Hellhammer D (eds), Assessment ojHormones and Drugs in
(38~ Cohen S, Kamarck T, Mermelstein R: A global measure of per- Saliva in Biobehaviora(Research. Seattle, Washington: Ho~efe R
ntiation ceived stress. Journal ojHealth and Social Behavior. 1983, 24: Huber, 1992, 163-173.
3):237- 385-396. (57) Kirschbaum C,Hellhamer DH: Methodological aspec[s ofsalivary
(39) Brandut~dter J, Baltes-GSu B, Kirschbaum C, et al: Develop- cortisol measurement and iu application in Suess research.Journal
dons of mental and personality correlates of adreAocortical activity as in- of Clinical Chemistry and Clèrica! Biochemistry. 1990, 2E.649-
;ical ac- dexed ty salivary cortisol: Observations in the age range of 35 to 666.
65 years. Journal ojPsychosórriatic Research. 1991, 35(2;3):173- (58) Ansseau M, Sulon .T, Doumont A, of aL The use of salivá c~i-tisol
Physi- 185. in the dexamethasone suppression test. Psychiatry Research. 1984,
(40) Pope MK,Smith TW: Cortisol excretion in high and low cynically 13:201-211.
ïngland hostile men. Psychosomatic Medicine. 1991, 53:386-392. (59) Watson D, Clark LA: Negative affectivity: The disposition to ex-
(41) Franlcenhaeuser M, Lundberg U, Fredrikson M, et al: Stress on perience aversive emotional states. Psychologica/ Bulletin. í 984.
anon of and off the job as related to sex and occupational status in white- 96(3):46590.
nstrated collar workers. Journal ofOrganizational Behavior. 1989, 10:321- (60) Costa PT, McCrae RR: Neuroticism. somatic complaints, and
62. 346. disease: Is the bark worse than the bite? .T~urnal ojPersonalit}~.
aw-Hill, (42) Cox'DJ, ireundlich A, Meyer RG: Differential effectiveness of 1987, 55(2)299-316.
eléctromyograph feedback, verbal relaxation instructions, and (61) Watson D,PennebakerJW: HeaÍth a~mplaints,stress,and distress:
gal axis. medication placebo with tension headaches. Journal ojConsulling Exploring the central role of negative affectivity. Psychologica!Re-
2):319- and Clinical Psychology. 1975, 43:892-898. view. 1989, 96(2):234-254.
(43) Attanasio V, Andrasik F, Blanchazd EB,et al: Psychometric prop- (62) Schaubroeck J, Ganster DC, Fox ML: Dispositional affect and
ilar sys- erties of the SUNYA revision of the Psychosomatic Symptom work-related stress. Journal of Applied Psychology. 1992, 77(3):
~, Bio- Checklist. Journal ofBehavioral Medicine. 1984, 7(2):247-258. 322-335.
{44) Dijkstra P: De zelfbeoordelingsschaal voor depressie van Zuog. In (63) Jascard J,V✓an CK Statistical analysis of'temporal data with many
pence of van Praag HM, Rooymans HGM (eds), Stemming en Onutem- observations: Issues for behavioral medicine data. Annals ojBe-
the het- ming. Amsterdam: De Erven Bohn, 1974. havioralMedicine. 1993, 15(1):43-49. -
inology. (45) Zung WWK: Aself-rating depression scale. Archives of Genera! (64) Bryk AS, Raudenbush SW: Hierarchical Lrnear Models: Applica-
Psychiatry. 1965, 12:63-70. tions and Dara Analysis Methods. London, England: Sage Pubii-
nviron- (46) Ploeg HMvd,Defares PB,Spielberger CD:Handleiding bijde Zelf- cations, 1992.
oringer- Beoordelings Vragenlijst- ZBV- een Nederlandstalige Bewerking
van de Spielberge~ State-Trol! Inventory(a manualjoy the STAID.
~ronary Lisle, The Netherlands: Swett & Zeitlinger, 1980.
., Florin
search.

it Three
. 46(3):

ne study
~n rates

View publication stats

You might also like