Objectives Recent research from industrial employees suggests the components of job control might be differently associated with mortality; high skill discretion with lower but high decision authority with higher mortality. This observation has not been confirmed in other cohorts.
Methods The purpose of this study is to further examine the association of skill discretion and decision authority with all-cause and cause-specific mortality in an independent cohort of 60 202 public sector employees from the Finnish Public Sector study by stratifying analyses by sex and socioeconomic status.
Results High skill discretion and high decision authority were associated with lower all-cause mortality rates in white-collar women. By contrast, high decision authority was associated with higher all-cause mortality rates in blue-collar women. No robust association between skill discretion, decision authority and mortality was observed among men. There were no robust associations with cause-specific mortality rates.
Conclusions These results suggest that the associations between components of job control and mortality are mixed and may vary depending on sex and socioeconomic status.
Statistics from Altmetric.com
What this paper adds
High job control has been viewed as beneficial for employee health.
A recent study found that a component of job control, high decision authority was, in fact, associated with higher mortality among forest industry employees.
This study aims to replicate the findings of job control components and mortality from the forest industry study among a cohort of 60 000 public sector employees.
The results show that high decision authority at work can be associated with higher or lower all-cause mortality depending on the socioeconomic status and gender of the employees.
The concept of job control, also known as decision latitude, is considered to be a key element in the effects of psychosocial stress at work on health.1 Low levels of job control have been linked to an increased risk of myocardial infarction and coronary death.2 However, job control is a multifactor concept (ref. 1, p. 58), the two most prominent components being decision authority (ie, the degree to which an employee can decide on the amount, tempo and method of his/her work) and skill discretion (opportunities for variable work in which the employee can use his/her competencies and learn new things). In a recent analysis of industrial employees, we found that the subcomponents of job control might not be uniformly associated with disease endpoints, possibly explaining the fact that the observed associations with job control measured as a single score have been modest.3 We found that low skill discretion was associated with increased mortality, but surprisingly, high rather than low decision authority at work was associated with an increased risk for all-cause, cardiovascular and alcohol-related mortality. It has been suggested that chance, the organisational or societal context, or the non-standard job control measure used in that study may have affected the findings.4 ,5 However, evidence substantiating these speculations or replications of the findings are not available.
The aim of this study is to further examine the associations of the components of job control, that is, skill discretion and decision authority, with all-cause and cause-specific mortality. Using a sample of over 60 000 public sector employees and data based on the standard validated scales of job control components,6 ,7 we examined whether the results from industrial employees can be replicated in this much larger cohort with a different work context. Additionally, we examined whether these associations varied by sex or socioeconomic status (SES), and how each individual item in the job control scale is associated with mortality.
The data were derived from the Finnish Public Sector (FPS) study, an on-going prospective cohort study of employees in the municipal services of 10 Finnish towns and 21 public hospitals.6 The ethics committee of the Hospital District of Helsinki and Uusimaa approved the study.
The eligible population from the register cohort of FPS (n=151 618) included those who had been employed for a minimum of 6 months at the participating organisations between 1991 and 2005. Employers’ records have been used to identify the eligible employees for a nested survey cohort to whom questionnaire surveys have been repeated every 4 years since 2000. In total, 66 418 participants responded at least once to surveys in 2000 (response rate 67%) and 2004 (response rate 65%). In this study, we used data from the first year when the survey response of an employee was available, that is, either 2000 or 2004. The analyses are based on this pooled data from the two starting points. We excluded employees who had been admitted to hospital for cardiovascular, cancer, alcohol-related or for mental health reasons according to the Finnish Hospital Discharge Register,8 at any time during the 10 years prior to the survey (n=6216). Thus, the final analytic cohort included 60 202 employees (47 927 women and 12 275 men, mean age 43.3 years, range 17–69 years). Using the unique national identification number that each permanent inhabitant in Finland receives, the participants were linked to the Statistics Finland National Death Registry9 which provided comprehensive data on dates and causes of mortality.
Components of job control
Job control was measured with the Job Content Questionnaire.7 Skill discretion (Cronbach's α 0.81) was measured by 6 items (‘My job requires that I learn new things’, ‘My job involves a lot of repetitive work’, ‘My job requires me to be creative’, ‘My job requires a high level of skill’, ‘I get to do a variety of things on my job’, ‘I have an opportunity to develop my own special abilities’); and decision authority (α 0.77) was measured by 3 items (‘My job allows me to make a lot of decisions on my own’, ‘I have a lot to say about what happens in my job’, ‘On my job I am given little freedom to decide how I do my work’). All items had a 5-point response scale. As there are no validated clinical cut-off points for these scales, and in order to keep consistency with the analyses in the forest industry study, we divided both summary scales into tertiles.
Ascertainment of all-cause and cause-specific mortality
Mortality data from 2000 to 2010 were obtained from the National Death Registry maintained by Statistics Finland. The database provides virtually complete population mortality data.9 We obtained the dates and causes of death (from death certificates) of all the participants. Diagnoses were based on the International Classification of Diseases (ICD) versions 9 or 10. Separate analyses were conducted for deaths due to cardiovascular diseases (ICD-9 codes 390–459, ICD-10 I00–I99) cancer (ICD-9 codes 140–208, ICD-10 C00–C97), alcohol-related causes (ICD-9 codes 291, 3050, 303, 3575, 4255, 5353, 5710–5713, 5770D-5770F, 7607A, 7795A, E851, E860, ICD-10 F10, G312, G4051, G621, G721, K929 I426, K70, K860, O354, P043, X45) and external causes (ICD-9 codes E800–E858 or E860–E990, ICD-10 V01–X44 or X46–Y89).
The covariates obtained from the employers’ administrative records were age, sex and occupational title. Occupational titles were used as an indicator of individual level SES. We classified individuals into two groups: white-collar workers (eg, physicians, teachers, registered nurses), and blue-collar workers (eg, cleaners, maintenance workers) based on the International Standard Classification of Occupations.10
To identify participants with diabetes or hypertension at baseline, we used data for entitlement to drug reimbursement due to chronic diabetes or hypertension from 1962 until the survey; this data was retrieved from the Drug Reimbursement Register maintained by the Social Insurance Institution.11 The national sickness insurance scheme covers the entire population, regardless of age or occupational title, and provides reimbursement for all filled prescriptions. Additional covariates were smoking status, (current smoker vs non-smoker), leisure time inactivity (<2 metabolic-equivalent task hours per day, ie, approximately 30 min brisk walking a day), obesity (Body Mass Index ≥30 vs <30), and heavy alcohol consumption (≥24 drinks/week for men, ≥16 drinks/week for women), which all were derived from survey responses.12 Time pressure at work was measured using three items from the Job Content Questionnaire: ‘My job requires working very fast’, ‘My job requires working very hard’, ‘I have enough time to get the work done’ (α 0.78).
We examined the associations between the components of job control and mortality using Cox proportional-hazards models. For each participant, we calculated person-days of the follow-up from the first baseline measurement in 2000 or 2004 to the end of the follow-up period (31 December 2010) or death, whichever came first. The time-dependent interaction terms between each predictor and logarithm of the follow-up period were non-significant, confirming that the proportional hazards assumption was justified (all p values >0.70). The HRs and their 95% CIs for categorical independent variables provided risk estimates. We conducted the analysis in four stages. First, we examined separately the crude effect of skill discretion and decision authority on mortality. Next, in Model 1, job control variables were entered separately and adjusted for age, sex, SES and physical health (hypertension, diabetes). In Model 2, skill discretion and decision authority were entered together adjusting for age, sex, SES, and physical health (hypertension, diabetes). In Model 3, job control variables were entered together and adjusted for age, sex, SES, physical health (hypertension, diabetes), time pressure, smoking, alcohol consumption, physical activity and obesity. Finally, we carried out the analyses in subgroups (men and women, white-collar and blue-collar employees) with Model 1 adjustments.
All the analyses were performed with maximum data, which resulted in some variation in the number of participants in different comparisons. The only exception was the multivariate models, which involved only the participants with no data missing for any of the predictors. The analyses were conducted using the PHREG procedure in the SAS V.9.2 statistical software package.13 The forest plot figure was produced with Meta Data Viewer software.14
Table 1 presents the descriptive statistics of the cohort and the psychosocial variables and their associations with skill discretion and decision authority. Of the employees, 82% were white-collar employees and 18% blue-collars. During the mean follow-up of 8.8 years, SD 1.7 years, 696 subjects (249 men and 447 women) had died. The average time from the survey to death was 5.6 years, SD 2.6 years.
Table 2 presents the number of cases for all-cause and cause-specific mortality and HRs for baseline covariates. In the models adjusted for sociodemographic factors, older age, male sex, blue-collar SES, prevalent diabetes, prevalent hypertension, smoking, high alcohol consumption and low physical activity were all associated with all-cause mortality.
Associations of the components of job control with all-cause and cause-specific mortality
Table 3 shows how skill discretion and decision authority are associated with all-cause and cause-specific mortality. For consistency with our earlier study of industrial employees,3 Model 1 includes the same adjustments as the main analysis in that study. High skill discretion was associated with lower all-cause mortality in the unadjusted model, but not when adjusted for demographics and physical health (Model 1). Decision authority was not associated with increased all-cause mortality in the unadjusted or adjusted models. There were no associations in the adjusted models in the analysis for cause-specific mortality.
We repeated the analyses for all-cause mortality separately for men and women and according to SES with Model 1 adjustments (figure 1). High skill discretion was associated with decreased mortality risk among white-collar men and women in combination, but when stratified by sex, the association remained only for white-collar women. High decision authority was associated with lower mortality among white-collar women. By contrast, high decision authority was associated with higher mortality among blue-collar women. As a sensitivity analysis, we repeated this analysis with all supervisors excluded in a subsample (the 10-town cohort, n=33 933) where information on supervisor status was available. The magnitude of the associations remained similar, but with the reduced sample size the confidence limits became wider, and only high decision authority for white-collar women remained statistically significant (HR 0.63, 95% CI 0.72 to 0.95), and high skill discretion for white-collar women (HR 0.70, 95% CI 0.48 to 1.01) and high decision authority for blue-collar women (HR 1.69, 95% 0.77 to 3.69) did not.
Further analyses examined all-cause mortality associations with each individual item of the job control scale in the subgroups where associations were found for the subscales. We fitted the 5-point scale as a continuous variable (per 1 point increase) with adjustments for age and physical health. For white-collar women, three of nine items showed an association with lower mortality risk: two from the skill discretion
‘My job requires me to be creative’ (HR 0.91, 95% CI 0.82 to 1.00),
‘My job involves a lot of repetitive work’—reverse coding (HR 0.86, 95% CI 0.78 to 0.95),
and one from the decision authority subscale
‘In my job, I am given little of freedom to decide how I do my work’—reverse coding (HR 0.90, 95% CI 0.81 to 0.99).
For blue-collar women, two items were associated with increased mortality: one from the skill discretion subscale
‘My job requires a high level of skill’ (HR 1.29, 95% CI 1.04 to 1.57).
and one from the decision authority subscale
‘I have a lot to say about what happens in my job’ (HR 1.32, 95% CI 1.09 to 1.61).
We examined the associations of the two components of job control, decision authority and skill discretion, in relation to mortality in a large cohort of public sector employees, and found no associations with all-cause or cause-specific mortality in the total sample of men and women. However, subgroup analyses suggested that high skill discretion and high decision authority were associated with lower all-cause mortality in white-collar women, whereas high decision authority was associated with higher mortality in blue-collar women.
We have previously reported data from a forest industry cohort where employees were mainly men, showing high decision authority to be a risk factor for mortality, particularly cardiovascular and alcohol related,3 as well as for hospitalisation due to mental disorders.15 In the present study, we sought to replicate the findings in a cohort of female-dominated public sector employees. The results were only partially replicated; we found no evidence to support an overall differential effect of skill discretion and decision authority on mortality in the total cohort. However, the findings were similar to those among industrial employees in a subgroup of blue-collar women; high decision authority, instead of low, was associated with increased mortality. In white-collar women, the results were in line with the theory, showing an association of high decision authority and high skill discretion with lower mortality. In the forest industry study, the effect of decision authority on mortality was of similar magnitude for blue-collar and white-collar employees (HRs 1.35 vs 1.31), but statistically significant only for the blue-collar employees. In the forest industry study, high decision authority was a risk factor for men and women, but the effect was larger for women (HRs 1.24 vs 1.67). Skill discretion did not show any significant association in the forest industry study in the subgroups. A further stratification into blue/white-collar men and women was not made in the forest industry study analyses.
There have also been previous findings that higher levels of job control and combination of high job demands and high job control, an active job, is associated with cardiovascular disease outcomes16 ,17 and all-cause mortality18 in women. It could be argued that there might be a gender difference in the effects of job control components on mortality. Higher job control is generally associated with higher occupational status,19 ,20 and in some studies occupational title has been used as a proxy measure for job control.21 There is more variation between job control levels for women than in men between occupational classes, that is, low-level men are closer to high-level men in their job control than low-level women to high-level women.20 But given that we found an elevated, although smaller, risk also for high decision authority men in the earlier forest industry study,13 it is more likely that these findings in women reflect more the types of jobs or organisation practises in which women work than individual characteristics of women as such. The job market remains quite segregated into male and female occupations,22 and it can be that in this study the blue-collar work occupations which were female-dominated, for example, cleaners, have other organisational features, for example, lack of resources or management styles, that make higher decision authority burdensome for the employees. It is thus possible that the effects in blue-collar women are related to an ‘unofficial’ supervisor status, where diligent employees are expected, or themselves feel they are obliged, to carry extra responsibility and support the whole team. For white-collar women individual decision authority may be more easily incorporated into the overall management structure. Excluding supervisors from the analyses did not affect the magnitude of the associations found, but this analyses was weakened by the loss of half of the study sample. Future studies should look more closely to the formal and informal status, and objectively and subjectively evaluated job descriptions. It has been noted that there is a discrepancy between the subjectively reported and external assessments of working conditions between men and women; women in ‘active’ jobs with high perceived demands and control have more externally evaluated hindrances and less externally evaluated influence than men in active jobs.23 Also, the finding that mortality risk was higher among blue-collar women who reported that their job required a high level of skills could indicate that blue-collar employees may still not receive adequate training and resources to do their job properly. Women in white-collar jobs may have more resources and access to training if they need it.
There are some limitations to our study. Despite the large sample size, the number of outcome events in some of the subgroups was fairly small. Supervisor status could be taken into account in only half the study sample limiting statistical power in subgroup analyses. We cannot exclude residual confounding by ill health and the effect of other unmeasured risk factors. For example, socioeconomic circumstances before the study entry and their changes during the follow-up could have also affected the observed associations. Also, important cardiovascular risk factors, such as cholesterol levels, could not be adjusted for.
Despite these limitations, the findings of the present study provide further justification of revising the concept of job control from a uniform concept of control into its subcomponents of decision authority and skill discretion. Stratified analyses would also add information regarding the diversity of the health effects of these work characteristics. Our study on public sector employees, along with a previous study on industrial employees, suggests that treating job control as a single, uniform concept might not be justifiable because the associations of its components with mortality can be opposite, depending on the subgroup, and thus cancel each other out in analyses of the global concept of job strain. Future studies should analyse which organisational or individual features could explain why job control or its components seem to be harmful for other occupational groups and beneficial for others.
Contributors All the authors participated in designing the study, interpreting the data, and writing and critically reviewing the article. JP performed the statistical analyses. MJ wrote the first draft.
Funding This work was supported by the Academy of Finland (grant no. 264944).
Competing interests None.
Ethics approval The Ethics Committee of the Hospital District of Helsinki and Uusimaa.
Provenance and peer review Not commissioned; externally peer reviewed.
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.