The impact of employee level and work stress on mental health and GP service use: an analysis of a sample of Australian government employees

The PATH Through Life Project is a longitudinal study of individuals living in the community with participants being drawn from three age groups: 20–24, 40–44 and 60–64 years. Those in the age group of interest for this study were aged from 40 to 44 years on 1 January 2000 and drawn from the Australian Electoral Rolls for Canberra in the Australian Capital Territory and adjacent town of Queanbeyan in New South Wales. Enrolment on these rolls is compulsory for all Australians aged 18 and over. Potential participants were drawn from a 10-year age range, the minimum range then released for research purposes by the Australian Electoral Commission. The number of potential participants found, and in the required age range, was 3919, of whom 2530 participated in the survey, giving a response rate of 64.4%.

Canberra is the national capital of Australia and many Australian Government entities are based in the ACT, including both houses of parliament, and the 16 major agencies that currently comprise the Australian Public Service. In this study, 806 respondents aged between 40 and 44 reported that they worked in office-based government administrative positions, developing and implementing government policy. As well as providing information on labour force status and the type of position held, respondents who worked in government positions were specifically asked to provide details on the level of the position they occupied. Five mutually exclusive employee categories were formed by grouping together those whose levels of employment were broadly comparable as follows. Employees who occupied positions at Australian Public Service (APS) Levels One to Four were grouped together as Junior employees and those at both the APS Levels 5 and 6 were classified as belonging to the Intermediate category of employees. Employees in the next two classifications of the APS (Executive Levels 1 and 2) were allocated to separate groups, Senior 1 and Senior 2. While those in Executive Level 1 positions develop policy and implement government programs, those at the Executive 2 level are primarily managers with direct responsibility for managing a number of employees from APS Level 1 to Executive Level 1 [12]. Finally, all respondents in the Senior Executive Service of the government were allocated to one category, Executive. The number of participants in each of these five categories of employees is given in Table 1, together with a short description of the positions included in each of those categories.

Survey participants completed a questionnaire that included socio-demographic characteristics and measures of physical and mental health, and well-being. Participants in the workforce were asked 22 questions relating to their work situation. These questions matched those used in the Whitehall II study [7, 13]. Nine questions related to job control, and reflected the amount of authority the worker has over decision-making [9]. Four questions concerned the manageability of job demands; the extent to which the worker is faced with difficult time and workload pressures and conflicting demands. Finally six questions addressed skill discretion and related to the variety of tasks to be done and the breadth of skills needed to undertake those tasks. For each of these 19 questions, respondents could answer: often, sometimes, rarely or never. Responses were given values of 1 to 4 with the highest score allocated to the less stressful work circumstances: those in which the individual had more job control, more manageable job demands, and higher levels of skill discretion. Participants were also asked the number of hours they usually worked per week and their assessment of their employment security and future employment opportunities. Answers to the last two questions used four point Likert-type scales and, again, were coded to give higher scores to those who reported that they had a more secure position or could obtain another job relatively easily. The mean of these two scores was used as an overall measure of job security.

Socio-demographic measures used in these analyses included sex, age, years of education, level of household responsibilities, and experience of any of six life events during the previous six months. Since each of these factors has the potential to modify an individual's mental health independently of their work stress, we adjusted for these in our final analyses. Scores for level of household responsibility were drawn from participants' responses to questions concerning the extent to which, in their household, they were responsible for four areas: household tasks, childcare, financial management, and providing money. Comparable scores for participants who did not have children in their households were then derived by calculating the mean of their measures for household tasks, financial management and providing money, and adding this to their total household responsibility score. Scores for these measures could range from zero to 16 with higher sores representing more household responsibilities.

Health measures obtained from participants and used in these analyses included: scores on Goldberg's depression and anxiety scales [14] and state measures of positive and negative affect using the Positive and Negative Affect Scales (PANAS) [15]. Measures of self-rated health, mental and physical health were taken from respondents' answers to the Medical Outcomes Study 12-item Short-Form Health Survey (SF-12) [16]. The first of these is measured by a single question in which participants rate their health as excellent, very good, good, fair or poor with higher scores indicating poorer self-rated health. Records on participants' visits to general practitioners (GPs) were also obtained. In Australia, the costs of most health care visits made to medical practitioners by Australians with citizenship or residency status are subsidised, either partly or totally, through the Australian Government funded universal health insurance scheme, Medicare. Information on the number of such visits is collected by the Health Insurance Commission. These data are used for administrative purposes and identify general practitioner and specialist services, but not the health problems addressed during each visit. While these records cover most visits made to general practitioners, they will not include a small number of services, paid for by patients but not claimed against Medicare. All participants were asked if they would consent to the researchers being provided information on the number of visits they made to general practitioners for specific periods before and after their interview. 709 (88.0%) of the 806 participants consented to this request and information on the number of GP visits they made during the six months preceding and the six months following their PATH interview was obtained from the Health Insurance Commission.

Analyses of variance (ANOVAs) were first undertaken to examine the extent to which sociodemographic measures and work stress measures changed with level of employment. Similar analyses, conducted separately for men and women, then compared mean mental health measures across the five employee levels. Finally, regression analyses were used to examine the contribution of employee category and work attributes in explaining participants' health and health service measures, whilst controlling for the following possible modifying factors: participant's age, years of education, level of household responsibilities and life events experienced in the past six months. For these analyses, categorical variables identifying each of the five government employee categories were created and the first four of these included in the regression equations, taking the most senior category, Executive, as reference group. After initial testing indicated that two dependent variables, negative affect and use of GP services, were not normally distributed and more closely fitted the negative binomial and Poisson distributions respectively, analyses of these two measures used negative binomial and Poisson regressions respectively. Strength of associations between dependent variables and predictor variables were measured using R ² for linear regressions and Incidence Rate Ratios when the Poisson or negative binomial regression model was used. Incidence Rate Ratios (IRRs) are interpreted in a similar manner to odds ratios and represent the expected change in the dependent variable in response to one unit change in the predictor variable. The contribution of employee level and work stress measures in explaining variation in health measures was also obtained. This contribution was measured using change in R² for linear regressions and the change in the Chi-square estimate of the fit of model for the negative binomial and the Poisson regression analyses. A final analysis examined the impact of employee category and work stressors on use of GP services, taking into account demographic, lifestyle and health measures. Analyses were undertaken using SPSS 11.5 and STATA 7 [17].

Our first hypothesis, that government employees working at lower levels would report higher levels of work-related stress, is supported in the main by our research. Overall, participants at more senior levels reported that they had more control over aspects of their work, greater opportunities to do interesting work using a range of skills, more job security but also that they were subject to higher job demands. These results closely match those reported by Marmot and colleagues in their 1991 study of over 10,000 British civil servants aged between 35 and 55 [7].

We had hypothesized that those in lower level positions would have poorer physical and mental health compared with more senior staff. None of our results supports this hypothesis. Although both male and female participants in the higher grades reported better well-being as measured by higher positive affect, we found no mental health benefit associated with having a more senior position. Furthermore, for men, we found this result to be reversed when the analyses controlled for work stress factors. Men in lower level positions reported higher levels of positive affect and better mental health, as measured by the SF-12. Women's mental health and negative affect were not affected by their employee level both when this factor was considered alone and when work stress factors were taken into account. We were unable to replicate the finding by Marmot and colleagues that those in higher positions had significantly better physical health compared with their subordinates. This difference in our results might be explained by our having a smaller sample. However, it also indicates that while those working at lower levels are more likely to experience some types of work stressors, working at those lower levels, per se, is not automatically associated with poorer mental or physical health for Australian government employees. While these findings are unusual, they do align with some previous research undertaken in the UK showing that those working in lower grades had better mental health [18]. One possible interpretation of our results is that men and women of this age group who have continued to work at lower employee levels may be pursuing satisfying goals in other areas of their lives, for example, family, outside business, recreational pursuits. Of course, confirmation of such an interpretation would require more detailed information from participants concerning their working arrangements and life priorities.

We found government employee level affected men's use of GP services but the direction of this effect was the reverse of that hypothesised. After taking into account education, other personal measures and levels of work stress, men at lower levels of the public service used fewer GP services than their superiors. This aligns with the previous finding that men at lower level positions also had better mental health [18] and suggests that further research is needed on the health drawbacks for men of their rising through levels of government employment. We also found that women in middle-management levels were less likely to have obtained GP services compared with their more senior counterparts.

Work stress factors experienced at all levels of employment played a more significant role in affecting participants' health and well-being. For male and female participants, those whose job demands were more manageable had significantly better mental health. This measure of work stress was particularly important for women, for whom manageability of demands was significantly associated with all health measures. Correspondingly, level of skill discretion had a greater impact on men's mental health. This last finding concurs with a recent Australian study of government employees which reported that, for men, having interesting work was an important reason for their staying in the government sector whereas women saw their relative job security as the advantage of this form of employment [19].

Since previous research has linked job insecurity and poorer mental health [20, 21], we expected similar findings in this current study. Women with less job security had less positive and more negative affect. Men with less job security, on the other hand, scored significantly worse on all six mental health measures. This finding, with that of the Australian study reported above, suggests that, while women place value on security, this and other work attributes may play a less important role in affecting their overall mental health, compared with their male counterparts. Previous studies have also reported sex differences concerning the mental health consequences of job insecurity [22]. Job security has also previously been found to affect the physical health of men more than women [23, 24]. However, the one measure of physical health used in this study was not directly affected by job insecurity for men or women.

A number of work stress factors contributed significantly to explaining men's and women's use of GP care. Women who used more services had less job control while men were more likely to visit a GP if they reported less skill discretion, job control or felt more insecure about their job. Research reported in the 1980's also found that factory workers with higher job insecurity used more health services [25]. This finding in our study of government employees suggests that job insecurity has an important effect on men's, but not women's, self-assessment of their levels of health and well-being.

Women who worked longer hours used slightly more GP services and also had more manageable job demands. Men who worked fewer hours, however, obtained more GP care. These findings indicate that time spent at work is unlikely to have deterred women from addressing their health care needs. For men, on the other hand, working fewer hours per week might have given them needed, additional opportunity to set aside time for a GP visit. In Australia, many GP surgeries are open during standard working hours and offer only limited access outside of these times [26].

This study is limited by having access only to cross-sectional data, since the direction of relationships between psychological stress and work stressors cannot be confirmed. It may be that those with fewer symptoms of depression or anxiety, for example, also have more positive views about the current or potential benefits of their positions. A number of issues raised here may become clearer when data from the next wave of the PATH project are collected.