Article Text

Job strain among blue-collar and white-collar employees as a determinant of total mortality: a 28-year population-based follow-up
  1. Mikaela B von Bonsdorff1,
  2. Jorma Seitsamo2,
  3. Monika E von Bonsdorff1,2,
  4. Juhani Ilmarinen2,
  5. Clas-Håkan Nygård3,
  6. Taina Rantanen1
  1. 1Gerontology Research Centre, Department of Health Sciences, University of Jyväskylä, Jyvaskyla, Finland
  2. 2Finnish Institute of Occupational Health, Helsinki, Finland
  3. 3School of Health Sciences, University of Tampere, Tampere, Finland
  1. Correspondence to Dr Mikaela von Bonsdorff; mikaela.vonbonsdorff{at}jyu.fi

Abstract

Objectives To investigate the effect of job demand, job control and job strain on total mortality among white-collar and blue-collar employees working in the public sector.

Design 28-year prospective population-based follow-up.

Setting Several municipals in Finland.

Participants 5731 public sector employees from the Finnish Longitudinal Study on Municipal Employees Study aged 44–58 years at baseline.

Outcomes Total mortality from 1981 to 2009 among individuals with complete data on job strain in midlife, categorised according to job demand and job control: high job strain (high job demands and low job control), active job (high job demand and high job control), passive job (low job demand and low job control) and low job strain (low job demand and high job control).

Results 1836 persons died during the follow-up. Low job control among men increased (age-adjusted HR 1.26, 95% CI 1.12 to 1.42) and high job demand among women decreased the risk for total mortality HR 0.82 (95% CI 0.71 to 0.95). Adjustment for occupational group, lifestyle and health factors attenuated the association for men. In the analyses stratified by occupational group, high job strain increased the risk of mortality among white-collar men (HR 1.52, 95% CI 1.09 to 2.13) and passive job among blue-collar men (HR 1.28, 95% CI 1.05 to 1.47) compared with men with low job strain. Adjustment for lifestyle and health factors attenuated the risks. Among white-collar women having an active job decreased the risk for mortality (HR 0.78, 95% CI 0.60 to 1.00).

Conclusion The impact of job strain on mortality was different according to gender and occupational group among middle-aged public sector employees.

This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: http://creativecommons.org/licenses/by-nc/2.0/ and http://creativecommons.org/licenses/by-nc/2.0/legalcode.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Article summary

Article focus

  • High job strain and its components, high job demand and low job control, predict cardiovascular and total mortality.

  • Although lower socioeconomic position is a risk factor for premature total mortality, few studies have explored the effect of job strain on mortality within socioeconomic groups and the ones that exist, report conflicting findings.

Key messages

  • In a population-based cohort of middle-aged public sector employees, low job control among men increased and high job demand among women decreased the risk of mortality during a 28-year follow-up.

  • High job strain increased the risk of mortality among white-collar men and passive job among blue-collar men compared with men with low job strain.

  • Active job among white-collar women decreased the risk for mortality compared with those with low job strain.

Strengths and limitations of this study

  • A major strength was the representative large sample of public sector employees working both in white-collar and blue-collar professions and the long follow-up time on mortality collected from the national mortality register.

  • A limitation is the self-reported job strain, however, high correlations between subjective and expert ratings on work conditions have been reported. The assessment of job strain was measured at a single time point in midlife which might imperfectly reflect long-term job strain, however, the municipal employees in our cohort had stable work histories indicating stability probably also for job strain during their earlier working life.

Introduction

A constant stressful situation, such as job strain, is associated with an increased prevalence of disease and accelerated ageing process.1 Epidemiological studies have shown that high job strain particularly among men is associated with the risk of cardiovascular morbidity and mortality.2–9 However, notable negative findings have also been reported and findings among women are inconsistent.10 11

Job demand and job control differs according to occupational class.5 12 13 A lack of job control has been shown to be more frequently present in blue-collar professions and high job demand among white-collar employees.13 14 The effect of occupation on the relation between job strain and mortality has in most studies been accounted for by adjusting the analyses for at least one of the common socioeconomic indicators, occupational class, education and income. However, the effect of job strain on cardiovascular morbidity and mortality has been less studied within socioeconomic groups and the ones that exist report conflicting findings.8 14 15 Furthermore, few studies have investigated the effect of job strain in midlife on total mortality with follow-up extending to old age.

We defined job strain according to the model described by Karasek.16 In the model, high job strain, indicated by high job demands and low job control, is considered to be the most detrimental type of job strain in terms of subsequent health. Other job strain categories are active job (high job demand and high job control), passive job (low job demand and low job control) and low job strain (low job demand and high job control). Based on earlier research, we hypothesised that the association between high job strain and total mortality was more pronounced among the white-collar employees because of overall differences in the work. In this study, we examine the effects of job demand, job control and job strain on total mortality among middle-aged white- and blue-collar public sector employees in a representative cohort with a 28-year follow-up.

Methods

Study design and population

The Finnish Longitudinal Study on Municipal Employees (FLAME) was established in 1981 by the Finnish Institute of Occupational Health.17 The 6257 (44.7% men) baseline participants were aged 44–58 years. Participants were randomly selected municipal employees who were working in all municipal professions in Finland. Data have been collected in 1981, 1985, 1992, 1997 and 2009 with questionnaires on various work-related factors and health and lifestyle. At baseline, data were available for 5731 (91.6%) participants on job strain. Compared with those with no information on job strain at baseline (n=526), the analytical sample (n=5731) of this study did not differ statistically according to gender (p=0.35), but they were younger and more frequently white-collar employees (p values <0.001). The Ethical Committee of the Finnish Institute of Occupational Health has approved the Finnish Longitudinal Study of Ageing Municipal Employees.

Job strain

We created four job strain categories according to the job demand–control model described by Karasek.16 18 Job demand was assessed with five questions (Cronbach's α=0.73) dealing with pressures related to the job: work pace fast and time schedule tight, responsibility, conflicting demands regarding work tasks and responsibility, pressure and interference with the job by the supervisor, and pressure of failing or doing errors on the job. Answering alternatives were ‘not at all=0’, ‘little=1’, ‘somewhat=2’ or ‘a lot=3’. A summary score ranging between 0 and 15 was calculated for job demand with high scores indicating high job demand. Job control was assessed with 10 questions (Cronbach's α=0.86) requesting participants to indicate to what extent (‘not at all=0’, ‘little=1’, ‘somewhat=2’ or ‘enough=3’) they were able to have guidance in the job, influence the work environment, participate in planning the work, get a promotion, get further training to maintain and develop professional abilities, chance to use ones abilities and talents, learn new things and develop oneself, get recognition and respect, work with co/workers, and to see the meaning of the work. A summary score ranging between 0 and 30 was calculated for job control, with higher scores indicating a high level of control over work done. There was a significant interaction between job demand and control on total mortality (p=0.005). The job demand and job control variables were dichotomised at the median, and four quadrants of job strain were constructed: passive job (low job demands and low job control), low-strain (low job demands and high job control), high-strain (high job demands and low job control) and active jobs (high job demands and high job control).

Mortality data

Data on total mortality were obtained from the Finnish National Population Register. The study population was followed up for mortality between 1 January 1981 and 31 July 2009. Survival time was calculated as the number of days between 1 January 1981 and death or end of the follow-up, whichever happened first.

Covariates

The analyses were adjusted with covariates known to modify the association between job strain and mortality.3 At baseline, in 1981, age, smoking history (current or former smoker vs never smoked), alcohol consumption (at least once a week vs less) and physical activity during previous year (vigorous physical activity at least once a week vs less) were assessed. Socioeconomic status was controlled with occupational group defined as the participants' position of employment at baseline. The 133 different identified occupational titles were clustered into 13 occupations based on job analysis at the work places.19 These were further collapsed and divided into white-collar (eg, teachers, doctors, registered nurses, managers) and blue-collar (eg, cleaners, plumbers, construction workers) according to objective assessments of the job characteristics. The self-reported physician-diagnosed or -treated cardiovascular diseases (eg, hypertension and angina pectoris), metabolic disorders (eg, diabetes and obesity) and cancer (malignant tumours at any site) were included in the models to control for the employees' health.

Statistical analyses

Baseline characteristics of the participants according to job strain categories were compared using χ2 tests for categorical variables and analysis of variance for continuous variables. The interaction between gender and job strain on total mortality was significant (p=0.002), thus analyses were stratified by gender. Proportional hazards assumption was tested to identify variables over time to see whether the associations were moderated by the time elapsed between the survey and death.20 Cox regression models were used to analyse first the independent effects of job demand and job control and then their combined effect (job strain) on mortality for men and women. The Cox regression analyses were adjusted for age, then occupational group and finally also for lifestyle factors such as smoking, alcohol intake and physical activity and health factors such as prevalent cardiovascular disease, metabolic disorders and cancer. The employees with low job strain served as the reference group. There was no significant interaction between occupational group and job strain on total mortality (p>0.299) so we stratified analyses by occupational group to see whether the effect of job strain on mortality differed for the occupational groups. Mortality rates per 1000 person-years and 95% CIs according to job strain for white-collar and blue-collar employees were calculated from 1 January 1981 to the date of death or the end of follow-up 31 July 2009 whichever occurred first. In Cox regression models, the base model was adjusted for age and the fully adjusted model for smoking, alcohol intake, physical activity and prevalent cardiovascular disease, metabolic disorders and cancer. Analyses were carried out with SPSS V.15.0.1 software (SPSS Inc).

For those missing at most two items of five for job demands (3.3%) or five items of 10 for job control (5.0%), missing values were imputed with a median value calculated from the other items included in the scale. In sensitivity analyses, we excluded the first year of follow-up in the Cox regression models to avoid reverse causation, but, as this hardly affected the HRs, all participants were included in the analyses.

Results

Mean age of the participants at baseline was 50.5 (SD 3.6) and mean length of follow-up was 25.3 (range 0.04–28.6) years. During the follow-up, 1836 (32.0%) of the 5731 baseline participants died. Mean age at death was 72.1 (SD 7.9) years for women and 69.7 (SD 8.0) years for men.

Table 1 presents the baseline distribution of socio-demographic, lifestyle and health factors for the four job strain groups for men and women. Among men, more than 80% of those with passive or high job strain were blue-collar employees, while half of those with active or low job strain worked in blue-collar professions. Men with passive jobs were the oldest. The group of men with low job strain had smoked less, consumed less alcohol, was more frequently physically active and had the lowest number of cardiovascular disease, metabolic disorders and cancer compared with the other job strain groups. Sixty per cent of women with passive or high-strain jobs were blue-collar employees, while a third of those with active and low-strain jobs were working in blue-collar professions. The groups with active jobs or low job strain were more frequently physically active than the other job strain groups. The group of women with low job strain had the lowest number of cardiovascular disease and metabolic conditions compared with the others.

Table 1

Baseline characteristics according to job strain in midlife in the FLAME Study

Mortality rates for men and women according to job strain for white-collar and blue-collar employees are presented in table 2. For men in white-collar work, total mortality was lowest for those with passive jobs and low job strain and highest for those with high job strain, but the differences were not statistically significant (log-rank test p=0.13). For men in blue-collar work, the lowest mortality rate 17.1 per 1000 person-years was observed among those with low job strain and the highest 22.5 per 1000 person-years for those in passive jobs (log-rank test p=0.033). Among women in white-collar work, the mortality rate was highest for those with low job strain (log-rank test p=0.079) and among women in blue-collar work for those in passive jobs (log-rank test p=0.029).

Table 2

Mortality rates per 1000 person-years and 95% CIs according to occupational group and job strain in midlife among men and women in the 28-year follow-up of the FLAME Study

Table 3 shows the gender-stratified individual effects of job demand, job control and finally job strain (ie, combined job demand and job control) on total mortality. Among men, high job demand increased the risk of total mortality in the follow-up, but the risk was not statistically significant, HR 1.10 (95% CI 0.98 to 1.28). Adjusting for occupational group and lifestyle and health factors further attenuated the association. Job control was associated with mortality in men (age-adjusted HR 1.26, 95% CI 1.12 to 1.42). Adjusting for covariates decreased the strength of the association, but it remained statistically significant. For the effect of job strain, compared with men with low job strain, those with passive, active and high job strain had a statistically significantly increased risk for death, but adjustment for lifestyle and health factors attenuated the association. Among women, high job demand decreased the risk for mortality during the follow-up compared with those with low job demands and adjusting for covariates had little effect on the association (HR 0.82, 95% CI 0.71 to 0.95), see table 3. Job control was not associated with mortality among women. For the effect of job strain, compared with women with low job strain, those with active job had a statistically significantly increased risk for death (HR 0.77, 95% CI 0.63 to 0.95).

Table 3

HRs and 95% CIs for mortality for individual effects of job demand, job control and job strain in the 28-year follow-up of the FLAME Study

Table 4 shows the HRs for total mortality for white-collar and blue-collar work according to job strain. For men in white-collar work, high job strain was a significant predictor of total mortality compared with those with low job strain (HR 1.52, 95% CI 1.09 to 2.13). For men in blue-collar work, passive job was associated with an increased risk of death compared with those with low job strain (HR 1.28, 95% CI 1.05 to 1.56). These observed associations were attenuated by adjustment for lifestyle and health factors. Active job among white-collar female employees decreased the risk of mortality during the follow-up compared with those with low job strain (HR 0.77, 95% CI 0.59 to 1.00). Job strain was not associated with total mortality among women in blue-collar work. Including older employees has been shown to decrease the relation between job strain and cardiovascular disease21 which is why we conducted analyses stratified by age at 44–50 and 51–58 years (data not shown). In the analyses confined to younger men in white-collar work, high job strain was a significant predictor of mortality after adjustment for lifestyle and health factors (HR 2.26, 95% CI 1.16 to 4.42). Among the younger men in blue-collar work, passive job increased the risk of total mortality (HR 1.42, 95% CI 1.00 to 2.02). Among younger white-collar women in active jobs, the risk of mortality was lower than for those with low job strain (HR 0.59, 95% CI 0.36 to 0.96). Among the older men and women, these associations were less evident.

Table 4

HRs and 95% CIs for mortality according to job strain in midlife stratified by occupational group among men and women in the 28-year follow-up of the FLAME Study

Discussion

This prospective 28-year follow-up showed that low job control among men increased and high job demand among women decreased the risk of total mortality among middle-aged employees working in the public sector. Stratifying the analyses by occupational class showed that high job strain increased the risk of mortality among men in white-collar work, while passive job increased the risk among men in blue-collar work compared with the ones with low job strain. Active job decreased the risk of mortality among white-collar women. Our findings support in part the existing body of knowledge and showcase new findings on the effect of job strain on total mortality in different occupational groups.

Low job control was associated with total mortality among men parallelling earlier findings on the effect of job control on coronary heart disease and mortality in various cohorts.2 4 22 Our findings concerning the effect of job strain within occupational groups on the risk of mortality among men support the findings of Kuper and Marmot8 who found that high job strain increased the risk of coronary heart disease especially among administrative workers. Unlike Johnson et al,14 we did not find an increased risk for mortality among those with high job strain compared with the ones with low job strain for men working in blue-collar professions. This might, however, in part be due to the different measure of job strain as our analyses did not include job support.

We found the effects of job strain on total mortality to be more evident in the analyses confined to the younger employees, aged 44–50 years at baseline. One plausible explanation for the observed association might be the overall effect of an increased stress level such as high job strain which may lead to cumulative wear and tear potentially suppressing immune function over time, increasing general susceptibility, reducing systemic regulation and increasing broad disease risk and decline in health.23 This is probably more evident among the younger employees while the healthy worker survivor effect might dilute the association between job strain and mortality in the older employees because the ones who are healthier are more likely to participate in the study.21 24 In addition, in a long follow-up, other factors influencing mortality may dilute the association between job strain and mortality more among the older workers particularly towards the end of the follow-up. Although we did not measure the duration of job strain, we know that the participants had a relatively stable job history. Among male employees, more than 69% had worked in one or two professions during their work career and more than 70% had held the current job position for more than 10 years (data not shown). This gives us reason to believe that the duration of job strain exposure had been long.

Passive job has been shown to predict mortality in the US Panel Study of Income Dynamics cohort.22 In our study, among blue-collar men, the mortality rate for those with passive job was higher compared with those with low job strain. A potential explanation for this is that those blue-collar men with health problems might have had to quit working in high-strain and active jobs and transfer to a passive job with low job demand and control already before this study began as stated in the healthy worker survivor effect.25 Blue-collar men who have lower education and income level are more likely to have been exposed to an unhealthy lifestyle, poorer living conditions and a physically strenuous work environment than white-collars.26 This increases the risk of chronic illnesses such as diabetes and cardiovascular disease and disability, which, in turn, increase the risk for death.27 28 It is plausible that this negative exposure has been highest among those blue-collar men who worked in passive jobs with low job demand and low control, which typically outline the lowest professions.

The effect of job strain and its components job demand and control on coronary heart disease and mortality have been less studied in women and the findings have been inconsistent.4 9–11 We found a protective effect of high job demand compared with low demand on total mortality among women working in the public sector. The finding of Eaker et al10 on job demand and total mortality is similar but not statistically significant (adjusted RR 0.96, 95% CI 0.91 to 1.01). We found in white-collar women with active jobs the mortality risk to be lower compared with the women with low-strain jobs. The finding might reflect socioeconomic differences in mortality,27 29 while white-collar women in active jobs with high job demands and control represent the highest socioeconomic gradient known to have lowest premature mortality rates. However, there is little information on the effect of job strain on health within the occupational classes and these associations need to be investigated in different populations.

Study strengths and weaknesses

The strengths of our study include the representative large sample of municipal employees working both in white-collar and blue-collar professions. In addition, the follow-up on mortality was long and the data were collected from the national mortality register. Some limitations of the study should be recognised. We did not have data available on job support to investigate the effect of job strain and low social support at the work place, which have been shown to correlate highly with CVD mortality.2 Using another job stress models, such as the effort-reward balance model,30 in which imbalance between personal efforts and rewards has been shown to predict coronary heart disease31 would have further strengthened the analyses. The data on job strain were self-reported which might yield possible reporting bias; however, high correlations between subjective and expert ratings on work conditions have been reported.24 32 The assessment of job strain was measured at a single time point in midlife which might imperfectly reflect long-term job strain and result in an underestimation of the association between job strain and mortality.33 However, the municipal employees in our cohort had stable work histories thus indicating stability possibly also for job strain during their earlier working life. Our study cohort included persons who worked in the public sector, which should be considered when generalising the results to the entire working population in the Western countries. Non-response in this study was relatively small (8%) and those who did not participate were older. According to the healthy worker survivor effect, those who are older and potentially less healthy more frequently fail to participate in the study. This might on the one hand cause underestimation of the effects of job strain on mortality but on the other hand including older employees might dilute these effects while older participants have a higher prevalence of other age-related health problems that are not related to job strain.21

Conclusions

Our evidence suggests that job strain was associated with mortality in middle-aged public sector employees but that it differed according to gender and occupational class. The effect of job strain on total mortality in different occupational groups is not well known, which warrants further investigations into these differences.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • To cite: von Bonsdorff MB, Seitsamo J, von Bonsdorff ME, et al. Job strain among blue-collar and white-collar employees as a determinant of total mortality: a 28-year population-based follow-up. BMJ Open 2012;2:e000860. doi:10.1136/bmjopen-2012-000860

  • Contributors JI, C-HN, JS and TR contributed substantially to conception and design of the FLAME Study. JS, MEvB and MBvB contributed to the data collection. MBvB and TR interpreted the data. MBvB analysed the data and drafted the manuscript. All the authors critically revised the article and gave final approval of the version to be published.

  • Funding FLAME was supported by the Finnish Institute of Occupational Health and The Local Governments Pension Institution. Mikaela von Bonsdorff was supported by Academy of Finland grant no. 132597, University Alliance Finland, University of Jyväskylä, Yrjö Jahnsson Foundation; Monika von Bonsdorff was supported by the University Alliance Finland, University of Jyväskylä; C-HN was supported by the Signe and Ane Gyllenberg Foundation.

  • Patient consent Obtained.

  • Ethics approval Ethics approval was provided by Ethical Committee of the Finnish Institute of Occupational Health.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement No additional data are available.