J. Human Ergol.

, 30: 203-209, 2001

JOB STRESS, SOCIAL SUPPORT AT WORK, AND INSOMNIA IN

JAPANESE SHIFT WORKERS

Akinori NAKATA, Takashi HARATANI, Masaya TAKAHASHI, Norito KAWAKAMI,

Heihachiro ARITO, Yosei FUJIOKA, Hiroyuki SHIMIZU, Fumio KOBAYASHI and Shunichi
ARAKI

National Institute of Industrial Health, Kawasaki, Japan,Department of Public Health,

Okayama University School of Medicine, Okayama, Japan,Department of Public Health and
Occupational Medicine, Graduate School of Medicine, University of Tokyo, Japan,Department
of Public Health, Gifu University School of Medicine, Gifu, Japan, Department of Health and
Psychosocial Medicine, Aichi Medical University,Aichi, Japan

A cross-sectional study was conducted to clarify the contribution of psychological

job stress to insomnia in shift workers. A self administered questionnaire
concerning job stress, sleep, depressive symptoms and lifestyle factors was
submitted to a sample of 530 rotating shift workers of age 18-59 years (mean age
27) in an electric equipment manufacturing company. Perceived job stress, i.e.,
job demands, job control and social support at work, was assessed using the
Japanese version of the Job Content Questionnaire. Insomnia was regarded as
prevalent if the workers had at least one of the following symptoms in the last
year; less than 30 minutes to fall asleep, difficulty in maintaining sleep, or early
morning awakening almost everyday. Overall prevalence was 37.8%. Logistic
regression analyses while adjusting relevant factors showed that lower social
support at work was significantly associated with a greater risk of insomnia than
the higher social support (adjusted OR 2.5). Higher job strain with lower social
support at work increased the risk, compared to lower strain with higher support
at work (crude OR 1.8; adjusted OR 1.5). Our findings suggest the low social
support at work independently associated with insomnia in shift workers

Introduction
Insomnia, defined as difficulty in initiating and maintaining sleep, is one of the most
commonly reported health problems in shift workers (KNAUTH et al., 1980). Although its
prevalence varies considerably (10-90%®),it is clear that insomnia has important implications for
shift workers since evidence suggests associations with poor health, reduced productivity,
increased accidents at work and absenteeism (STOLLER,1994).
Epidemiological studies designed to determine the risk factors for insomnia suggested that
aging, female genders, shift work, psychological stress, perceived poor health status, low life
quality, and/or environmental disturbances are at higher risks (HARMA et al., 1998; NAKATA et
al., 2000). Among them, shift work and job stress are suspected to be two major occupational risk
factors (KAWACHI et al., 1995; COSTA, 1997; MURATA et al., 1999). In contrast, only a few
studies on the effects of job stress on sleep have been conducted, mostly in industrialized countries.
In these studies, high job demands (CAHILL and LANDSBERGIS, 1996; RIBET and
DERRIENNIC, 1999), low job control (KALIMO et al., 2000), high job strain as defined as the
combination of greater job demands and lower job control according to the job-demands control
model (THEORELL et al., 1988; CAHILL and LANDSBERGIS, 1996), low social support at
workplace (LANDSBERGIS, 1988) and less satisfaction with work (KUPPERMANN et al., 1995)
are reported to relate to an increase in sleep problems. A poor atmosphere at work (JACQUINET-
204 A. NAKATA et al.

SALORD et al., 1993) as well as overall mental workload (URPONEN et al., 1988) are also
proposed as the factors that disturb sleep.
KAGEYAMA et al. (1998) found that poor sleepers had significantly higher scores in job
difficulty, lower scores in achievement in job and support by colleagues. TACHIBANA et al.
(1996) showed over-involvement in job to be a high risk factor for insomnia in male industrial
workers. In our previous report, we found an inverse relationship between social support and sleep
complaints in female daytime employees (NAKATA et al., 2001).
These findings suggest that certain types and amounts of job stress affect sleep or increase
sleep problems. Since shift workers are reported to receive lower social support from family and
friends (HOSSAIN and SHAPIRO, 1999) and higher amount of job stress than do permanent day
workers (TENKANEN et al., 1997), the effects of job stress factors on their sleep should be
examined.
In the present study, a cross-sectional study was conducted in an electric equipment
manufacturing company. Perceived job stress was assessed by the Japanese version of the Job
Content Questionnaire (JCQ) developed by KARASEK (1979) (KAWAKAMI et al., 1996).
Subjective sleep habits (length of time to fall asleep, frequency of awakening during sleep, and
early morning awakening) were assessed by a previously used measure (NAKATA et al., 2000a,
2001). Additionally, in an attempt to assess the main effect and buffering effect of social support
at work on insomnia individually, we tested two research hypotheses on the relationship between
social support at work and insomnia as in Fig. 1.

Fig. 1. Two hypothesized models for the effects of social support on insomnia.

Subjects and methods

A self administered questionnaire concerning demographic information, current job type,
work schedule, job stress, sleep habits, lifestyle factors, and depressive symptoms was submitted to
a total of 2,625 full-time employees in an electric equipment manufacturing company in April
1997. A total of 2,420 (92.2%) workers responded to the survey. We excluded day workers
(n=1367); female workers were also excluded because of the skewed distributions in age and job
type (n=466). Some workers who had a missing response were eliminated from the analyses.
Finally, a total of 530 workers were subjected to the analyses. Job type was categorized into four
groups, i.e., mechanics/repair workers, machine operators, manual laborers, and others (Table 1).
The present study was carried out with the informed consent of all the workers.
 205

Table 1. Selected characteristics, JCQ, and CES-D scale scores among shift workers of an electric equipment manufacturing
company (n=530).

&Jobstrain score was definedas the ratio of thejob demands score to job control score.

Assessment of job stress and social support at work: The Japanese version of the Job Content
Questionnaire (JCQ) was used to assess job stress according to the job demand-control model and
the demand-control-support model (KARASEK, 1979), as previously described (NAKATA et al.,
2000b, 2002). According to our previous study, this version has acceptable levels of reliability and
validity with Cronbach's alpha reliability coefficients from 0.61 to 0.89 for males and from 0.65 to
0.87 for females. (KAWAKAMI et al., 1996).

Assessment of insomnia
A self-administered sleep questionnaire was developed for this study and three questions
about subjective sleep habits during the last one-year period were included as follows:
1. "How long does it usually take you to fall asleep in bed?"(0-10 minutes / 11-30 minutes / 31-59
minutes / 1-2 hours /2+ hours)
2. "How often do you have difficulty staying asleep? "(never or almost never) / few times a year /
more than once a month / more than once a week / almost every day)
3. "How often do you wake up too early and can't fall asleep again?" (never or almost never) / few
times a year / more than once a month / more than once a week / almost every day)
The presence of difficulty initiating sleep (DIS) was defined when one took more than 30
min to fall asleep; difficulty maintaining sleep (DMS) and early morning awakening (EMA) were
defined when one answered "almost always" to the questions 2 and 3, respectively. The presence
of insomnia was defined when at least one of DIS, DMS, or EMA was reported.
Potential confounding factors: Other covariates included age, marital status (married/not married),
smoking (never or past/current smoker), drinking habit (> 4 days/week) (yes/no) caffeine intake (>
3 cups/day) (yes/no), habitual exercise (> 2 days/week) (yes/no), and depressive symptoms. The
 206 A. NAKATA et al.

depressive symptoms were measured by the Center for Epidemiologic Studies Depression Scale
(CES-D).

Statistical analyses: To examine the associations of job control, job demands, and social support at
work with insomnia, we performed a series of univariate and multivariate logistic regression
analyses individually. Age, marital status, smoking, drinking habit, caffeine intake, habitual
exercise and depressive symptoms were adjusted in the multivariate analyses. To assess buffering
effects of social support at work for the relation between job strain and insomnia, analysis of
covariance controlling for the potential confounding factors was used. The significance level of all
statistical analyses was set at a probability of less than 0.05 (two-tailed test). All data in the
present study were analysed by the Statistical Package for the Social Sciences version 10.0 (SPSS
Inc., Chicago, USA).

Results
Overall prevalence of insomnia was 37.8% with a 95% confidence interval (CI) of 33.8-
41.9%. Lower social support at work (Odds ratio (OR) 2.3) was significantly and directly
associated with an increased risk of insomnia; this association was confirmed by multivariate
analyses adjusted for the relevant factors (Table 2). In contrast, job control and job demands did
not show significant differences.

Table 2 Crude and adjusted OR and 95% CI for insomnia at each level of job stress among shift workers (n=515-521).

a All the workers were divided into three groups according to their job stress scores: Lower (or higher) scores for doe control ana
social support at work (or job demands) were one standard deviation below (or above) the mean; higher (or lower) scores were
above (or below) the mean, and medium between the higher and the lower. bAdjusted for all the other covariates in multivariate
logistic regression analyses. *p<0.05, ** p<0.01, n.s, non significant.

Workers with a combination of high job strain and low social support at work showed a
significantly higher risk of insomnia compared to a combination of low strain with high support
(OR 1.8). This association was not confirmed by the multivariate analyses, adjusted for age,
marital status, smoking, drinking habit, caffeine intake, habitual exercise and depressive symptoms
(Table 3). No significant interaction was found between job strain and social support.
 207

Table 3. Crude and adjusted OR and 95% CI for insomnia at each level of a combinationof job strain and social support at work
amone shift workers (n-5O7.

aJob strain and social support scores were dichotomised into higher or lower groups by respective mean scores, and created the
four job stress level groups (thereby constructing a higher and lower group for each scale). b See the footnote as in Table 2. *
p<0.05, n.s, non significant.

Discussion
The principal findings in this study are that the overall prevalence of insomnia in this
population was 37.8% and that insomnia appeared to be directly associated with social support at
work, rather than with job control or job demands. The results supported the hypothesis in the
present study that social support at work had a main effect on insomnia but did not support a
buffering effect, since no significant interaction was found between job strain and social support
(Fig. 1). Social support at work in shift workers is thus independently associated with insomnia,
even after controlling for important confounding factors.
The prevalence of insomnia was the highest among the workers with lower social support at
work (Table 2). The finding agrees with our previous study for the daytime female employees,
which found inverse relationships between social support from supervisors and frequency of early
morning awakening, and support from colleagues and frequency of awakening during sleep
(NAKATA et al., 2001).
The results of the current study are also in agreement with two previous studies.
LANDSBERGIS (1988) reported an inverse relationship between social support and sleep
problems in female health care workers, which also used JCQ as job stress measure. Another study
by KAGEYAMA et al. (1998) revealed a significant decrease in scores of support by colleagues in
poor sleepers compared to good sleepers in white-collar male workers. These reports support the
findings in the present study that social support at work has a main effect on insomnia. By
contrast, although the prevalence of insomnia was higher in the high strain/low support group (OR
1.8) compared to high strain/high support group (OR 1.0), a buffering effect was not supported
since there was no significant interaction between job strain and social support. Social support at
work itself may play an important role in the occurrence and/or persistence of insomnia in
Japanese shift workers. Further well-designed studies are required to confirm this relationship.
No clear associations were found between job control or job demands and prevalence of
insomnia in this study. In contrast, three studies reported significant associations with job control
or job demands with sleep. LANDSBERGIS (1988) found a significant positive correlation
between job demands and sleep disturbances. THEORELL et al. (1988) reported that sleep
disturbances increased from the lowest to the highest degree of job strain in a sample of 71 men
and women. In a recent study, an increase in job demands or a decrease in job control have been
observed to increase the prevalence of insomnia in male shift workers (KALIMO et al., 2000); the
prevalence of insomnia was higher in the highest quartile of job strain. The discrepancy between
the current and these previous studies might be attributable to the fact that our samples were from
a single occupation in an electric equipment manufacturing company compared to a variety of
occupational categories in other three studies; the job demands-control model is not clear in studies
208 A. NAKATA et al.

with a single occupation group since limited distribution of job demands and job control could be
observed (SPECTOR et al., 1987), as suggested by the standard deviations in the Table 1. Thus a
wide variety of occupational categories of shift workers should be examined in future studies.
There is evidence that shift work contributes to excess cardiovascular morbidity and
mortality (e.g. AKERSTEDT et al., 1984; KNUTSSON et al., 1986). On the other hand, it has
been reported that insomnia is observed prior to the occurrence of myocardial infarction
(CARNEY et al., 1990). Thus it could be hypothesized that insomnia induced by low social
support at work in addition to the effect of shift work may give rise to excess in the development of
cardiovascular diseases. Additional studies are needed to confirm this hypothesis.
Finally, several limitations of our study should be noted. The study design was cross-
sectional, and therefore only associations but not causal relationships could be identified. Also, the
sample size was small, and only male workers from a single company were included.
Comprehensive research is required to provide more practical knowledge about the role of social
support at work on insomnia in our working life.

Acknowledgement
This work was supported by the Special Coordination Funds of the Ministry of Education,
Culture, Sports, Science and Technology of the Japanese Government.

References
AKERSTEDT, T., KNUTTSON, A.,ALFREDSSON, L. andTHEORELL, T. (1984)Shiftworkandcardiovascular disease.
Scand.J. WorkEnviron. Health,10:409-414.
CAHILL, J. andLANDSBERGIS, P.A.(1996)Jobstrainamongpostofficemailhandlers. Int.J. Health.Services,26;731-751.
CARNEY,R.M.,FREEDLAND, K.E.and JAFFE,A.S.(1990)Insomniaand depression prior to myocardial infarction.
Psychosom. Med.,52:603-609.
COSTA, G. (1997)Theproblem: shiftwork.Chronobiol, Int.,14: 89-98.
HARMA, M.,TENKANEN, L.,SJOBLOM, T.,ALIKOSKI, T. andHEINSALMI, P. (1998)Combined effectsof shiftwork
andlife-style ontheprevalence of insomnia,sleepdeprivation anddaytimesleepiness. Scand.J. WorkEnviron.Health,
24:300-307.
HOSSAJN, J.L.andSHAPIRO, CM. (1999)Considerations andpossibleconsequences of shiftwork.J. Psycosom. Res.,47:
293-296.
JACQUINET-SALORD, MC.,LANG,T., FOURIAUD, C., NICOULET, I. and BINGHAM, A. (1993)Sleepingtablet
consumption, selfreported
qualityofsleep,andworking conditions.
J. Epedemiol. Community Health,47:64-68.
KAGEYAMA, T., NISHIKIDO, N., KOBAYASHI, T. et al. (1998)Self-reported sleepquality,job stress,and daytime
autonomic activities
assessedintermsof short-term heartratevariability
amongmalewhite-collar workers. Ind.Health,
36:263-272.
KALIMO, R., TENKANEN, L., HARMA,M.,POPPIUS, E. and HEINSALMI, P. (2000)Job stressandsleepdisorders:
findings fromtheHelsinki HeartStudy.StressMed.,16:65-75.
KARASEK, R.A.(1979)Jobdemand, jobdecision latitude,andmentalstrain:Implications forjobredesign.Adm.Sci.Q.,24:
285-308.
KAWACHI, I., COLDITZ, G.A.,STAMPFER, M.J. etal.(1995)Prospective studyofshiftworkersandriskofcoronary heart
diseaseinwomen. Circulation,
92:3178-3182.
KAWAKAMI, N. andFUJIGAKI, Y. (1996)Reliability andvalidityof the Japanese versionof Job ContentQuestionnaire:
Replication andextension incomputer company employees. Ind.Health,34:295-306.
KNAUTH, P., LANDAU, K.,DROGE, C., SCHWITTECK, M.,WIDYNSKI, M. andRUTENFRANZ, J. (1980)Durationof
sleepdepending onthetypeof shiftwork.Int.Arch.Occup.Environ.Health,46:167-177.
KNUTSSON, A.,AKERSTEDT, T.,JONSSON, B.G.et al.(1986)Increased riskof ischaemic heartdiseasein shiftworkers.
Lancet,2:89-92.
KUPPERMANN, M.,LUBECK, D.P.,MAZONSON, P.D.et al.(1995)Sleepproblems andtheircorrelatesina working
population. J. Gen.Intern.Med.,10:25-32.
LANDSBERGIS, P.A.(1988)Occupational stressamonghealthcareworkers: A testofthejobdemands-control model.J. Org.
Behav.,9:217-239.
MURATA, K.,YANO,E.andSHINOZAKI, T. (1999)Cardiovascular dysfunctionduetoshiftwork.J. Occup.Environ.Med.,
41:748-753.
 209

NAKATA, A., HARATANI, T., KAWAKAMI,N., MIKI, A., KURABAYASHI,L, and SHIMIZU, H. (2000a) Sleep problems
in white-collar male workers in an electric equipmentmanufacturing company in Japan. Ind. Health, 38: 62-68.
NAKATA, A., ARAKI, S., TANIGAWA, T. et al. (2000b) Decrease of suppressor-inducer(CD4+CD45RA+) T lymphocytes
and increase of serum immunoglobulin G due to perceived job stress in Japanese nuclear power plant workers. J. Occup.
Environ. Med., 42:143-150.
NAKATA, A., HARATANI, T., KAWAKAMI, N. et al. (2001) Relationship between perceived job stress and sleep habits in
daytime female workers -An epidemiological study of employees in an electric equipment manufacturing company in
Japan-Jap. J. Behav. Med., 7: 39-46 (in Japanese with English abstracts).
NAKATA, A., TANIGAWA, T., FUJIOKA, Y., KITAMURA, F., ISO, H. and SHIMAMOTO, T. (2002) Association of low
job control with a decrease in memory (CD4+CD45RO+) T lymphocytes in Japanese middle-aged male workers in an
electric power plant. Ind. Health, 40:142-148.
RIBET, C. and DERRIENNIC, F. (1999) Age, working conditions, and sleep disorders: a longitudinal analysis in the French
cohort E.S.T.E.V. Sleep, 22: 491-504.
SPECTOR, P.E. (1987) Interactive effects of perceived control and job stressors on affective reactions and health outcomes for
clerical workers. Work& Stress, 1: 155-162.
STOLLER, M. (1994) Economic impact of insomnia. Clinical. Therapeutics , 16: 873-897.
TACHIBANA, H., IZUMI, T., HONDA, S. et al. (1996) A study of the impact of occupationaland domesticfactors on insomnia
among industrial workers of a manufacturing company in Japan. Occup. Med., 46: 221-227.
TENKANEN, L., SJOBLOM, T., KALIMO, R., ALIKOSKI, T. and HARMA, M. (1997) Shift work, occupation and coronary
heart disease over a 6-years of follow up in the Helsinki Heart Study. Scand. J. Work.Environ. Health, 23: 257-265 .
THEORELL, T., PERSKI, A., AKERSTEDT, T. et al. (1988) Changes in job strain in relation to changesin physiological state.
Scand. J. Work. Environ. Health, 14: 189-196.