Background
No prospective studies have examined the association between occupational stress according to the job demand−control model and the risk of stroke in Asian populations.
Methods
We conducted a multicenter community-based prospective study of 6553 Japanese male and female workers. Occupational stress was evaluated using a Japanese version of the job demand−control model questionnaire. We used the Cox proportional hazards model to evaluate the association between occupational stress and stroke.
Results
During a mean follow-up of 11 years, we identified 147 incident strokes. Multivariable analysis revealed a more than 2-fold increase in the risk of total stroke among men with job strain (combination of high job demand and low job control) (hazard ratio, 2.73; 95% confidence interval, 1.17-6.38) compared with counterpart men with low strain (combination of low job demand and high job control) after adjustment for age, educational attainment, occupation, smoking status, alcohol consumption, physical activity, and study area. Additional adjustments for biologic risk factors attenuated the hazard ratio, but there continued to be statistical significance (hazard ratio, 2.53; 95% confidence interval, 1.08-5.94). In women, no statistically significant differences were found for any stroke incidence among the job characteristic categories.
Conclusion
Occupational stress related to job strain was associated with incident strokes among Japanese men.
Stress is considered a risk factor for stroke.1-3 Chronic stress associated with occupations can be avoided; modifiable job characteristics have been conceptualized as stress models, and these may provide clues for concrete interventions.4 The job demand−control model is the most often used occupational stress model.4,5 It posits that workers who face high psychological demands in their occupation and have little control over their work (ie, those who have job strain) are at a greater risk of becoming ill than are workers with low psychological demands and a high degree of control in their occupation (ie, those with low-strain occupations). Many prospective studies have supported this hypothesis using the outcome of coronary heart disease.6,7 Furthermore, stress-reduction approaches based on the model have been shown to be effective.8
There is limited prospective evidence showing an association between job strain and the risk of stroke9-12; all findings are from studies in Nordic countries. Two Swedish studies that evaluated job strain on the basis of workers' self-reported job characteristics failed to show statistically significant associations.9,10 Two additional large-scale registry-based studies showed that low levels of job control were significantly associated with death from stroke.11,12A crude assessment aggregated to the data by a secondary data source (job exposure matrix) indicated that a more accurate evaluation of occupational stress was warranted.6 Furthermore, conventional cardiovascular risk factors were not taken into account. In the present study, we sought to estimate the risk of stroke onset associated with job strain in a Japanese working population. We used individual-level data with a valid instrument. We also examined the possible confounding or mediating effects of socioeconomic, behavioral, and biologic risk factors by controlling for them in the analyses.
To investigate the risk factors for cardiovascular diseases in Japan in the Jichi Medical School Cohort Study, data for 12 490 Japanese adults from 12 communities located across Japan were collected between April 1992 and July 1995 using a standardized questionnaire and physical examination findings.13 Routine mass screening examinations for cardiovascular diseases in the older adults are held in Japan in accordance with legal regulations and provided the data for the present study. The regulations require municipal governments to offer screenings to all residents who are willing to participate. In each community, potential participants are invited to partake in screenings through letters or public information. Potential participants are told that persons who are treated at a hospital or clinic for a cardiovascular disease do not have to participate in the examination and screening. The overall response rate was 65.4%.
The aim of the present investigation was to explore the effect of job strain on incident stroke, so we limited the study population to 3659 male and 3995 female workers with a baseline age of 65 years or younger. We excluded workers with a history of stroke or myocardial infarction and those without complete information regarding occupational stress. The final study group consisted of 3190 men and 3363 women. The occupations of the participants were manager (759 men and 206 women); professional, technician, or clerk (251 men and 483 women); sales or service worker (280 men and 807 women); farming, forestry, or fishery worker (1055 men and 1110 women); and security, transportation, communications, or craft worker, laborer, or unclassified worker (845 men and 757 woman). Compared with the general working population of Japan, the study population included larger proportions of older workers and workers engaged in nonindustrial occupations (farming, forestry, and fishery).14 More than 99% of the participants were employed by companies with fewer than 300 employees. Japanese companies are required to provide an annual health examination of employees. For those who are not offered a health examination at their workplace, such as workers in a nonindustrial occupation or who are self-employed, the mass screening examination program is an opportunity to have their health status determined. Employees who are offered a health examination at their workplace may also participate in the mass health examinations. Many small companies (local industry) or local government offices in rural districts such as those constituting our study areas rely on the mass screening examinations to provide their employees with a health examination. We inferred from repeated surveys that changes in occupation or job position were not frequent in the rural settings included in the current cohort.15 Some part-time employees may have been included in the study population, but this was not confirmed.
Surveillance of stroke and classification of stroke subtypes
The study follow-up system ensures contact with the participants annually through direct interview or via telephone or letter to determine the participant's current health status. Participants are asked whether they had a stroke or diagnosis of cardiovascular disease that occurred after enrollment in the study. Those who did were asked which hospital they attended and the date of hospitalization. If an incident case was suspected, all the medical records were reviewed and duplicate computed tomography or magnetic resonance imaging films were obtained. Diagnosis of stroke was determined by the presence of a focal and nonconvulsive neurologic deficit lasting 24 hours or longer with a clear onset. Stroke subtype was determined according to the criteria of the National Institute of Neurological Disorder and Stroke.16 The International Classification of Diseases, Tenth Revision(ICD-10), codes that were applied to study patients were hemorrhagic stroke ICD-10 codes I60 (subarachnoid hemorrhage) and I61 through I62 (intracerebral hemorrhage) and nonhemorrhagic stroke ICD-10 code I63 (ischemic stroke). Patients with transient ischemic attack were excluded from the analysis. The diagnosis was determined independently by a diagnosis committee composed of a radiologist, a neurologist, and 2 cardiologists. With the permission of the Agency of General Affairs and the Ministry of Health, Labour, and Welfare, we determined the causes of death for all participants who died between the date of their first health examination and the end of 2005, using the Cause-of-Death Register found at the public health center located in each community.
Assessment of occupational stress
Job characteristics were derived at baseline using a Japanese version of the job demand–control model questionnaire from the World Health Organization Multinational Monitoring of Trends and Determinants in Cardiovascular Disease (WHO-MONICA) Psychosocial Study Questionnaire. The job characteristics studied were job control and psychological demands. Job control was defined as the sum of 2 subscales given equal weight: (1) skill discretion, measured by 4 items (possibility for learning new things, skills required for the job, requirement for creativity, and repetitious nature of the work) and (2) decision authority, measured by 2 items (right to make one's own decisions and freedom to choose the manner in which the work is performed). Psychological job demands were defined by 5 items (speed in completing work, degree of difficulty of the work, excessive workload, insufficient time allowed to complete the work, and conflicting demands). All questions were scored on a Likert scale of 1 to 4. The Cronbach α coefficients for the job control index and psychological demand index were .65 and .69, respectively. Interrelationships among the job characteristic scores, sex, age, educational level, and occupational status were in concordance with the current literature.17 The job conditions assessed as part of this cohort during the follow-up demonstrated a moderate degree of stability with 5-year−interval intraclass correlation coefficients of 0.63 (n = 377) for job control and 0.55 (n = 378) for job demands.15 Cross-classification of the job control and job demand scales according to their sex-specific median values produced a quadrant scheme with 4 exposure categories, with low job demand and high job control representing a low-strain job (reference category), high job demand and high job control representing an active job, low job demand and low job control representing a passive job, and high job demand and low job control representing a strain job.
Assessment of risk factors
We measured the following demographic characteristics and conventional risk factors at baseline: age (18-39, 40-49, 50-59, or 60-65 years), educational attainment (≤15 years [age at completion of compulsory education], 16-18 years [age at completion of senior high school], or ≥19 years [age at entering college or further education]), smoking status (lifetime nonsmoker, ex-smoker, or current smoker), alcohol consumption (nondrinker, <1 go daily (go is a traditional Japanese alcohol unit; 1 go = 28.9 g of alcohol), or ≥1 go daily), physical activity index18 (<29, 29-36, or ≥37), body mass index (calculated as weight in kilograms divided by height in meters squared) (<22, 22-24.9, or ≥25), hypertension (physician-diagnosed hypertension or systolic/diastolic blood pressure ≥140/90 mm Hg), diabetes mellitus (under treatment or a fasting/casual blood glucose level of at least 126/200 mg/dL [to convert to millimoles per liter, mutiply by 0.0555]), and hypercholesterolemia (physician-diagnosed hypercholesterolemia or total cholesterol ≥220 mg/dL [to convert to millimoles per liter, multiply by 0.0259]).
Analysis was based on the stroke incidence rate during the 11 years of follow-up. For each participant, person-years of follow-up were allocated according to the dates of her or his health examinations until death, date of movement outside the study community, the endpoint of stroke, or December 31, 2005, whichever occurred first. Data regarding the movements of the study population were obtained every year from the participant's municipal government, which records movements of residents in and out of the particular community. A total of 193 subjects (2.9% of the analytic cohort) moved out of their community during the study period and were treated as censored cases. The total observed person-years was 71 385. Cox's proportional hazard regression analysis was used to examine the association between psychosocial job characteristics and an incident stroke. The hazard ratios were estimated first after adjusting for age and study area (community), and then after adjusting for age, educational attainment, occupation, smoking status, alcohol consumption, physical activity index, and study area. Body mass index, hypertension, diabetes, and hypercholesterolemia were also included in the multivariable model as intermediate variables to examine how the inclusion of these variables influenced the association between psychosocial job characteristics and the risk of stroke. Ordinal variables were represented by dummy variables. All probability values were 2-tailed, and values of P < .05 were considered statistically significant. All analyses were conducted with SPSS for Windows, release 15 (SPSS Inc, Chicago, Illinois).
The study design and procedures were reviewed and approved by each municipal government and the Ethics Committee for Epidemiological Research at Jichi Medical School. Written informed consent was obtained from all prospective participants.
Table 1 shows the relationships between psychosocial job characteristics and the examined variables at baseline. Men reporting active jobs were younger than those reporting passive jobs. The socioeconomic status was lower for men with passive or strain jobs than for those with active or low-strain jobs; the former group was more likely to be engaged in blue-collar work and to have less education. Nonindustrial occupations were more prevalent among men with low-strain jobs, while managerial positions were prevalent among men with active jobs. Men exposed to high-strain jobs were more likely to be heavy drinkers than were men with low-strain occupations. Men with active jobs had a higher level of physical activity than did men with passive jobs. Body mass index was lowest in men with passive jobs. Women with low-strain jobs were older and more obese than women with job strain. The relationship between psychosocial job characteristics and socioeconomic status of women was similar to that of men. Active jobs for women were associated with alcohol consumption and a high level of physical activity.
During the follow-up period, we identified 147 incident strokes: 91 in men and 56 in women. There were 90 ischemic strokes (64 in men and 26 in women), 33 intracerebral hemorrhages (21 in men and 12 in women), and 24 subarachnoid hemorrhages (6 in men and 18 in women). Due to the small number of cases in each category, we pooled intracerebral and subarachnoid hemorrhage cases into the category of hemorrhagic stroke.
Age- and area-adjusted hazard ratios showed that men with high-strain jobs had a more than 2-fold higher risk of total stroke than did men with low-strain jobs. Further adjustment for socioeconomic status and behavioral risk factors strengthened the associations. The hazard ratio for job strain decreased after biologic risk factors were added to the model but continued to be statistically significant. Multivariable analyses of stroke subtypes revealed that men with high-strain jobs had more than 2-fold higher risks of ischemic and hemorrhagic strokes than men with low-strain jobs, but the associations were not statistically significant due to large confidence intervals. Although women with high-strain jobs tended to have a higher risk of stroke than women with low-strain jobs, no statistically significant differences were found for any stroke incidence among the job characteristic categories for women (Table 2).
To our knowledge, the present study is the first to demonstrate a significant association between job strain and the risk of an incident stroke. A more than 2-fold increase in the risk of incident total stroke was found among men with high-strain jobs compared with men with low-strain jobs. The association remained statistically significant after adjustment for various conventional risk factors. The more than 2-fold higher risks of job strain for ischemic and hemorrhagic strokes suggested that the association between job strain and stroke was attributable to increased incidences of both subtypes, although the effects of job strain on the risks of these subtypes separately were uncertain due to the limited number of cases. Our evidence was based on the largest available Japanese cohort of its kind, and data were obtained in a standardized fashion. Information about exposure to job strain was obtained from self-reports with a validated instrument rather than by assigning scores based on job description. Hence, each score more accurately represents the individual work environment.6 Diagnosis of stroke was ascertained by an independent committee using accepted diagnostic criteria. Self-reporting bias is unlikely to be important due to the hard end point and prospective study design, while bias attributable to sample attrition is thought to be implausible because the follow-up rate was high.
Most of the literature regarding job strain and cardiovascular diseases relates to white populations, with health outcomes fairly restricted to coronary heart diseases. Only large-scale prospective studies with aggregated job characteristics data have demonstrated associations between low job control and stroke mortality.11,12 The relatively higher incidence of stroke among Japanese men may produce sufficient statistical power to detect associations that have not previously been well addressed. Differences in the predominant lesions and their pathologic condition, or the most attributable risk factors, between Japanese and white populations may provide other explanations for race/ethnicity differences. Lacunar infarcts and small intracerebral artery lesions appear to be important for stroke pathogenesis among Japanese populations, in contrast to the large-artery occlusive infarctions found in white populations.19 Blood pressure appears more strongly associated with stroke in Asian populations, whereas dyslipidemia and a cholesterol-rich diet seem to be more strongly associated with stroke in white populations.19,20
Our findings did not show any statistically significant associations in women. Low statistical power among the women is obvious due to the limited number of outcome cases. The power analysis revealed that the minimal hazard ratio required to achieve statistical significance with 80% power among the women was around 2.5. Although this may be possible, it is hard to achieve based on the current literature.6,7 Furthermore, the characteristics of our female sample indicate that further research is necessary in women. Specifically, the labor force participation rate of Japanese women is lower than that in Western societies and, therefore, with the exception of large enterprises, the attitudes of Japanese women toward work may be less proactive. In addition, the possible inclusion of part-time workers, most of whom are assumed to be women, may have affected the results.
We found that adjustments for biologic risk factors slightly attenuated the association between job strain and an incident stroke. These findings suggest that the association between job strain and cardiovascular diseases is mediated by the presence of 1 or more chronic diseases, such as obesity, hypertension, glucose intolerance, and dyslipidemia.21 However, adjustments for such variables did not fully account for the associations between job strain and stroke. Other than the studied variables, possible mechanisms through which job strain leads to stroke may include poor adaptation to stress,1,22 enhanced sympathetic activation,22 and hemostatic or inflammatory conditions.23
The participants in this study were relatively older adults and thus a considerable number may have sustained a long career in the same job. Several studies regarding the association between job strain and cardiovascular disease risk have encountered the problem that many participants cease gainful employment during the follow-up period, such that exposure would have stopped and the associations become attenuated.6 Around the time when the study population was recruited, the average retirement age of Japanese employees was 63 years,24 but some workers in the cohort are thought to have retired later because of the rural study setting. A stratified analysis by age at the end of the follow-up (≤63 or >63 years) showed elevated risks of job strain among both strata, but the association was stronger among men who were followed up until an older age (≤63 years: age- and area-adjusted hazard ratio, 2.23; 95% confidence interval, 0.59-8.98; >63 years: 3.23; 1.09-9.59). Most subjects probably continued to work until an older age, and, therefore, would also have reached an age when there really is a risk of developing stroke (greater duration of exposure). Alternatively, the clinical manifestation could become overt after retirement due to the pathophysiological changes anticipated by job-strain exposure during their working life. Because we did not capture the exact length of job-strain exposure or the actual retirement age, these hypotheses need to be examined in a future study.
The study population was composed of relatively healthy Japanese adults. The mass screening examination program is not mandatory and employees who undergo health checks at their workplaces do not have to participate. Thus, participants may have a more health-oriented predisposition than do nonparticipants. Furthermore, the invitation to participate in screenings did not insist that those receiving care for cardiovascular diseases sign up. All these conditions probably accounted for the small number of outcome events.13 In addition, the relatively low response rate may imply that those with the worst work conditions chose to opt out of the study. The age-adjusted incidence rates for total stroke per 100 000 population (calculated using the Japanese employed population in 1995 as the standard population) were 229.7 for men and 107.4 for women, and slightly lower than those of a contemporary community-based cohort that used the comprehensive stroke registration system (268.7 for men and 167.5 for women).25 These findings may indicate that our observations reflect a profile of stroke incidence among community-dwelling Japanese workers. Nevertheless, because underrepresentation of those with access to occupational health care limits the ability to generalize the findings, future replication is needed among representative samples of employed workers.
The Cronbach α coefficient of job control was somewhat low, and our exposure assessment was limited to 1 point in time; both of these aspects probably caused associations toward the null. As with all observational studies, residual and unmeasured confounders remain an alternative explanation for our findings. For example, employment status (full-time vs part-time) and income level were not measured. Furthermore, the prevalence of negative emotions, such as depression, was not ascertained at baseline. However, recent studies indicate that the impact of a tendency toward negative affectivity should not be overstated.26
In conclusion, job strain was associated with incident stroke among Japanese men. Because modification of work structures based on the job demand−control model can be useful for stress reduction, our study has implications regarding the prevention of incident strokes among male workers.
Correspondence: Akizumi Tsutsumi, MD, Occupational Health Training Center, University of Occupational and Environmental Health, 1-1 Iseigaoka, Yahatanishi-ku, Kitakyushu, 807-8555 Japan (tsutsumi@med.uoeh-u.ac.jp).
Accepted for Publication: July 17, 2008.
Author Contributions: All authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Kayaba and Tsutsumi. Acquisition of data: Kayaba, Ishikawa, Tsutsumi, and Kario. Analysis and interpretation of data: Tsutsumi, Kayaba, and Kario. Drafting of the manuscript: Tsutsumi. Critical revision of the manuscript for important intellectual content: Tsutsumi, Kayaba, and Kario. Statistical analysis: Tsutsumi. Obtained funding: Kayaba, Ishikawa, and Tsutsumi. Administrative, technical, and material support: Ishikawa. Study supervision: Kayaba and Ishikawa.
Financial Disclosure: None reported.
Funding/Support: This study was partly supported by a grant-in-aid from the Foundation for the Development of the Community, Tochigi, Japan (Drs Kayaba and Ishikawa), and by a Grant-in-Aid for Scientific Research (C) (Dr Tsutsumi).
1.André-Petersson
LEngström
GHagberg
BJanzon
LSteen
G Adaptive behavior in stressful situations and stroke incidence in hypertensive men.
Stroke 2001;32
(8)
1712- 1720
PubMedGoogle ScholarCrossref 2.Surtees
PGWainwright
NWJLuben
RLWareham
NJBingham
SAKhaw
K-T Adaptation to social adversity is associated with stroke incidence: evidence from the EPIC-Norfolk Prospective Cohort Study.
Stroke 2007;38
(5)
1447- 1453
PubMedGoogle ScholarCrossref 4.Schnall
PLBelkić
KLandsbergis
PBaker
D Occupational Medicine: State of the Art Reviews: The Workplace and Cardiovascular Disease. Vol 15 Philadelphia, PA Hanley & Belfus2000;
5.Karasek
RTheorell
T Healthy Work: Stress, Productivity, and the Reconstruction of Working Life. New York, NY Basic Books1990;
6.Belkić
KLandsbergis
PASchnall
PLBaker
D Is job strain a major source of cardiovascular disease risk?
Scand J Work Environ Health 2004;30
(2)
85- 128
PubMedGoogle ScholarCrossref 7.Kivimäki
MVirtanen
MElovainio
MKouvonen
AVäänänen
AVahtera
J Work stress in the etiology of coronary heart disease: a meta-analysis.
Scand J Work Environ Health 2006;32
(6)
431- 442
Google ScholarCrossref 8.Bourbonnais
RBrisson
CVinet
AVézina
MLower
A Effectiveness of a participative intervention on psychosocial work factors to prevent mental health problems in a hospital setting.
Occup Environ Med 2006;63
(5)
335- 342
PubMedGoogle ScholarCrossref 9.André-Petersson
LEngström
GHagberg
BJanzon
LRosvall
M Social support at work and the risk of myocardial infarction and stroke in women and men.
Soc Sci Med 2007;64
(4)
830- 841
PubMedGoogle ScholarCrossref 10.Kuper
HAdami
HOTheorell
TWeiderpass
E The socioeconomic gradient in the incidence of stroke: a prospective study in middle-aged women in Sweden.
Stroke 2007;38
(1)
27- 33
PubMedGoogle ScholarCrossref 11.Toivanen
SHemström
Ö Is the impact of job control on stroke independent of socioeconomic status? a large-scale study of the Swedish working population.
Stroke 2008;39
(4)
1321- 1323
PubMedGoogle ScholarCrossref 12.Virtanen
SVNotkola
V Socioeconomic inequalities in cardiovascular mortality and the role of work: a register study of Finnish men.
Int J Epidemiol 2002;31
(3)
614- 621
PubMedGoogle ScholarCrossref 13.Ishikawa
SGotoh
TNago
NKayaba
KJichi Medical School (JMS) Cohort Study Group, The Jichi Medical School (JMS) Cohort Study: design, baseline data and standardized mortality ratios.
J Epidemiol 2002;12
(6)
408- 417
PubMedGoogle ScholarCrossref 14.Tsutsumi
AKayaba
KHirokawa
KIshikawa
Sthe Jichi Medical School Cohort Study Group, Psychosocial job characteristics and risk of mortality in a Japanese community-based working population: the Jichi Medical School Cohort Study.
Soc Sci Med 2006;63
(5)
1276- 1288
PubMedGoogle ScholarCrossref 15.Kayaba
KTsutsumi
AGotoh
TIshikawa
SMiura
Y Five-year stability of job characteristics scale scores among a Japanese working population.
J Epidemiol 2005;15
(6)
228- 234
PubMedGoogle ScholarCrossref 16.Adams
HPJBendixen
BHKappelle
LJ
et al. Classification of subtype of acute ischemic stroke: definitions for use in a multicenter clinical trial: TOAST: Trial of Org 10172 in Acute Stroke Treatment.
Stroke 1993;24
(1)
35- 41
PubMedGoogle ScholarCrossref 17.Tsutsumi
AKayaba
KTsutsumi
KIgarashi
M Association between job strain and prevalence of hypertension: a cross sectional analysis in a Japanese working population with a wide range of occupations: the Jichi Medical School Cohort Study [published correction appears in
Occup EnvironMed. 2003;60(2):149].
Occup Environ Med 2001;58
(6)
367- 373
PubMedGoogle ScholarCrossref 19.Reed
DJacobs
DR
JrHayashi
T
et al. A comparison of lesions in small intracerebral arteries among Japanese men in Hawaii and Japan.
Stroke 1994;25
(1)
60- 65
PubMedGoogle ScholarCrossref 20.Woodward
MHuxley
HLam
TH
et al. A comparison of the associations between risk factors and cardiovascular disease in Asia and Australasia.
Eur J Cardiovasc Prev Rehabil 2005;12
(5)
484- 491
PubMedGoogle ScholarCrossref 21.Chandola
TBritton
ABrunner
E
et al. Work stress and coronary heart disease: what are the mechanisms [published online ahead of print January 23, 2008]?
Eur Heart J 2008;29
(5)
640- 648
Google ScholarCrossref 22.Everson
SALynch
JWKaplan
GALakka
TASivenius
JSalonen
JT Stress-induced blood pressure reactivity and incident stroke in middle-aged men.
Stroke 2001;32
(6)
1263- 1270
PubMedGoogle ScholarCrossref 23.von Känel
RMills
PJFainman
CDimsdale
JE Effects of psychological stress and psychiatric disorders on blood coagulation and fibrinolysis: a biobehavioral pathway to coronary artery disease?
Psychosom Med 2001;63
(4)
531- 544
PubMedGoogle ScholarCrossref 24.Japan Ministry of Labour, Rodo-keizai no Bunseki 1989 (Rodo Hakusho) [Analyses on Labour and Economy 1989 (white paper)] Tokyo Japan Ministry of Labour1989;
25.Kita
YOkayama
AUeshima
H
et al. Stroke incidence and case fatality in Shiga, Japan 1989-1993.
Int J Epidemiol 1999;28
(6)
1059- 1065
PubMedGoogle ScholarCrossref 26.Bosma
HPeter
RSiegrist
JMarmot
M Two alternative job stress models and the risk of coronary heart disease.
Am J Public Health 1998;88
(1)
68- 74
PubMedGoogle ScholarCrossref