Background Unemployment and economic inactivity are associated with worse mental health in the general population, but there is limited understanding of whether these relationships are different for those persons with mental or physical disabilities. The aim of this study was to assess whether there were differences in mental health by labour force status among persons with and without disabilities.
Method Over eight annual waves of the Household, Income and Labour Dynamics in Australia (HILDA) survey, a total of 2379 people with disabilities and 11 417 people without disabilities were identified. Mental health using the Mental Component Summary (MCS) from the Short Form 36 was modelled as a function of labour force status using fixed-effects regression models to control for time invariant confounding. Differences between those with and without disabilities were assessed by including an interaction term in regression models.
Results After finding evidence of effect modification, regression models were stratified by disability status. After adjustment, unemployment and economic inactivity were associated with a −1.85 (95% CI −2.96 to −0.73, p=0.001) and −2.66 (95% CI −3.46 to −1.86, p<0.001) reduction in scores of the MCS among those with a disability. For those without a disability, there were smaller declines associated with unemployment (−0.57, 95% CI −1.02 to −0.12, p=0.013) and economic inactivity (−0.34, 95% CI −0.64 to 0.05, p=0.022).
Conclusions These results suggest a greater reduction in mental health for those persons with disabilities who were unemployed or economically inactive than those who were employed. This highlights the value of employment for people with disabilities.
- MENTAL HEALTH
Statistics from Altmetric.com
Disabilities are broadly described as health conditions or impairments that can restrict participation in activities such as work and employment.1 ,2 Eighteen per cent of Australians report a disability, which is similar to other OECD (Organisation for Economic Co-operation and Development) countries where the prevalence is between 10% and 23%.3 Compared with the general population, those with disabilities are less likely to participate in the workforce and face a range of other economic and social disadvantages such as low income and poor education.3–5 People with disabilities are likely to have much poorer health than people without disabilities6 and there is some evidence from longitudinal studies in Australia and the UK that disadvantaged circumstances are major contributors to the poorer health status among those with disabilities.5 ,7
There are numerous health and economic reasons for encouraging employment of those with disabilities.3 First, there is a large amount of research demonstrating the positive influence of employment on mental health and well-being.8–10 Evidence from qualitative studies indicates that the psychological benefits of employment also extend to those with disabilities.11 Evidence also demonstrates the detrimental impact of unemployment12 or being economically inactive on mental health,13–16 although there is a lack of research on the health effects of being without paid work on people with disabilities. There is also a clear economic incentive to promoting labour market participation among those with a disability as this may ease reliance on government support.3
There are a number of qualitative studies on the mental health effects of different labour force states for those with disabilities,11 and more specifically among those with serious psychiatric disabilities.17–19 While useful, these studies are unable to provide information on these relationships across a large nationally representative sample of people. Quantitative evidence about the association between unemployment and economic inactivity on the health of people with disabilities is important in the context of employment policy development. With this in mind, the aim of this study was to investigate the relationship between labour force status and mental health among those with disabilities, and to assess whether this relationship was different for those without disabilities. To do this, the study draws on information from a nationally representative cohort of Australians and uses fixed-effects regression models to control for time-invariant individual, demographic and socioeconomic factors.20
The key questions investigated in this study were: (1) What is the association between mental health and unemployment and economic inactivity (compared with being in employment)? (2) Is this relationship different for people with disabilities compared with people without?
The Household, Income and Labour Dynamics in Australia (HILDA) survey is a longitudinal, nationally representative study of Australian households established in 2001. It collects detailed information annually from over 13 000 individuals within over 7000 households.21 The response rate to wave 1 was 66%.21 The survey covers a range of dimensions including social, demographic, health and economic conditions using a combination of face-to-face interviews with trained interviewers and a self-completion questionnaire. Although data are collected on each member of the household, interviews are only conducted with those older than 15 years of age.
The initial wave of the survey began with a large national probability sample of Australian households occupying private dwellings.21 Interviews were sought in later waves with all persons in sample households who turned 15 years of age. Additional persons have been added to the sample as a result of changes in household composition. For example, if a household member left his or her original household (eg, children left home, or a couple separated), he/she formed an entire new household including all persons living with the original sample member. Inclusion of these new households is the main way in which the HILDA survey maintains sample representativeness. The response rates for the HILDA survey are above 90% for respondents who have continued in the survey and above 70% for new respondents being invited into the study.21
Definition of disability
The measure of disability used in the HILDA survey was based on the definition used in the International Classification of Functioning, Disability and Health.2 Disability was defined based on reporting over two consecutive waves, ascertained from the following survey question “…do you have any long-term health condition, impairment or disability that restricts you in your everyday activities, and has lasted or is likely to last, for six months or more?” While this question was being asked, specific examples of long-term conditions were shown on a card, such as limited use of fingers or arms, or problems with eyesight that could not be corrected with glasses or contact lenses, which were used to categorise reported disabilities into 17 long-term conditions.22 These questions were introduced into the HILDA questionnaire in 2003 and asked at every subsequent wave.
The Mental Component Summary (MCS) of the Short Form 36 (SF-36) measure was used as the primary outcome measure. The MCS score is comprised of four scales: mental health, role emotional, vitality and social functioning, which are derived from 14 questions.23 The SF-36 is a widely used self-completion measure of health status; it has been validated for use in the Australian population, and to detect within-person change over time.23 The SF-36 in the HILDA survey has been shown to be psychometrically sound, with good internal consistency, discriminant validity and high reliability.23 The mean score on the MCS in HILDA was approximately 49.8, with an SD of 10.3, a minimum of 4.4 and a maximum of 73.9.
The key exposure of interest was labour force status. This was classified as employed, unemployed (those who remain in the labour force and are actively seeking employment) or economically inactive (‘not in the labour force’ or NILF). Those who were economically inactive could have been out of the labour force for various reasons including ill health, retirement or looking after children.24 As HILDA is an annual survey, the measurement of mental health may follow the start of a spell without work by up to 12 months.
Socioeconomic factors that were controlled for included age (under 34, 35–54, 55 years and over), educational attainment (high school not completed, high school completed, certificate/diploma, bachelor degree or above), equivalised household income and household structure (couple or lone adult residing with dependants, couple without dependants, lone person without dependants and a group or multiple person household).
To be eligible for inclusion in the sample, participants had to have data available on labour force status and the MCS (figure 1). Following this, information was required on disability status. The final sample in the disability analysis consisted of persons who reported at least two consecutive waves of disability and reported a disability at the time of measurement of exposure variables. The non-disability sample consisted of people who experienced no disability through the period of the HILDA survey. The sample used for analysis was also required to have data available on all variables used in the regression models.
All annual waves between 2003 and 2010 were included in these analyses. Longitudinal linear fixed-effects regression models were used to estimate the association between labour force status and the MCS score within individuals. Coefficients generated from these models describe differences in the MCS score associated with being in a state of unemployment or economic inactivity compared with that individual's mean MCS score during their time in employment. These models provide an indication of within-person effects, where each individual acts as their own control and estimates are not confounded by time-invariant personal, demographic and environmental factors.20 Fixed-effects models are particularly useful where time-invariant confounding is likely to cause bias in causal estimates.20 For example, in the current paper, mental health may be affected by within-person factors such as personality. We controlled for time-variant factors by including a number of relevant covariates into the fixed-effects models (see below). Robust or sandwich estimators of variance were calculated to account for clustering of observations within individuals.
First, we tested the hypothesis that the association between labour force status and mental health differed by the presence of disability using an interaction term in fixed-effects models, and examined this using the likelihood ratio test and inspection of significance values of interaction terms in the model. The interaction term represented disability (yes/no) and labour force status (employed, unemployed, economically inactive).
Second, based on the fact that we found evidence of significant interactions, results were then stratified by two groups: (1) persons who had two consecutive waves of disability and reported disability at the time of measurement of exposure variables and outcome; and (2) those who reported no disability throughout the entire period of the HILDA survey. Fixed-effects regression models were used to examine mental health when unemployed or economically inactive compared with when employed within those with and without disabilities separately. Analyses were performed using STATA, V.12.1.25
As a sensitivity test, we re-ran the analysis after removing those who reported psychological disabilities because they may be more likely than people with other impairment types to suffer generally worse mental health, which may lead to changes in labour force status (ie, the healthy worker effect).26 A further examination was undertaken to assess whether the interaction between disability and employment status was confounded by age differences between those with and without disabilities. This was tested by including an interaction term between age and labour force status in a fixed-effects regression model while simultaneously controlling for the disability-labour force status interaction. We also examined the robustness of our results by examining the association between employment status and the MCS among those with three reported waves of disability, rather than two reported waves.
There were 2379 (observations 9901) persons who reported disabilities over two consecutive waves included in the sample. The mean number of observations for participants in this group was 4.1. Just over half of the participants in this sample reported multiple disabilities (51.2%). There were 11 417 (observations 52 491) persons who did not report any disability who were included in the sample. There were a mean number of 4.6 observations for all persons included in this sample. As can be seen in table 1, there were a number of demographic differences between those with and without disabilities. There were also notable differences in mental health among those with and without disability (table 2).
Regression results testing for effect modification by presence of disability
Results of the fixed-effects model (table 3) showed significant interaction terms for the disability-labour force status variables. The result of the likelihood ratio test was also significant (x2 (2)=96.52, p<0.001). This suggests that the association between employment status and the MCS differed depending on the presence of disabilities (ie, that disability was an effect modifier of the association between employment status and mental health).
Stratified models examining differences within those with and without disabilities
The fixed-effects regression models were then stratified into a sample of people with disabilities and a sample of people with no disabilities to assess MCS comparisons in the association between unemployment and economic inactivity compared with employment.
Results of the bivariate and multivariate regression analyses for those with disabilities can be seen in table 4. Results for the multivariate analysis (also in table 4) adjusting for other predictors suggest that those who experienced unemployment had a −1.85 (95% CI −2.96 to −0.73, p=0.001) point lower score than when they were employed, while those who experienced economic inactivity had a −2.66 (95% CI −3.46 to −1.86, p<0.001) point lower score on the MCS than when they were employed.
Among people without disabilities, in bivariate analyses, unemployment was associated with a −0.52 (95% CI −0.97 to −0.02, p=0.025) point lower score on the MCS (table 5) compared with when they were in the reference category of employment. Being economically inactive was associated with a −0.27 point lower score on the MCS (95% CI −0.56 to 0.02, p=0.067). After adjustment for other variables, the effect of being in unemployment increased slightly and there was a greater decline in MCS scores (−0.57, 95% CI −1.02 to −0.12, p=0.013). The association between being in the economically inactive category and MCS attenuated slightly and was associated with a −0.34 point difference (95% CI −0.64 to 0.05, p=0.022) compared with employment (table 5).
When those with long-term mental health problems (psychological disabilities) were removed from the disability analysis, unemployment was associated with a −1.85 point lower score on the MCS (95% CI −2.96 to −0.74, p=0.001) compared with being employed after adjustment). Those who were in the economically inactive category had a −2.51 point difference in the MCS (95% CI −3.35 to −1.68, p<0.001) compared with when they were employed.
The definition of disability was then altered to include in the sample only those persons who reported three consecutive waves of disability (n=1657). Results suggest that the association with economic inactivity stayed the same (−2.65, 95% CI −3.50 to −1.79, p<0.001), while the association with unemployment decreased slightly (−1.77, 95% CI −2.98 to −0.56, p=0.004) among those with at least three waves of disability.
When the age-labour force status interaction was included in the model with a disability-labour force status, interaction results suggested that both terms were significant (see online supplementary table S1). This indicates that disability was still a significant effect modifier of the relationship between labour force status and mental health after considering the possible effects of age.
A missing value analysis showed a minimal amount of missing information for all variables (under 10%). The outcome variable (MCS) had the most missing information (13%). We conducted further analysis to investigate missingness in the outcome variable by labour force and disability status. Our results suggest that those persons who failed to provide an MCS score were slightly more likely to be economically inactive or unemployed. For example, 21.2% of those with missing MCS information were in the economically inactive category (compared with 19.2% of those with MCS information). Further, 5.9% of those with missing MCS information were unemployed compared with 4.2% of unemployed persons who reported MCS information. There were also slight differences based on disability, with those reporting disabilities being more likely to have missing MCS information than those without (17.7% vs 15.5%).
The main results of this paper suggest that the presence of a disability influences the magnitude of the relationship between labour force status and mental health. Among those people with disabilities, there were larger mental health differences between being employed versus being unemployed or economically inactive than were observed for those without disabilities.
Previous research suggests that persons with chronic conditions and disabilities are much more vulnerable to job loss and less likely to enter paid employment than those with good health.26 As shown in a study from Iceland, those with disabilities may be particularly at risk of unemployment during times of economic recession.27 People with disabilities are also more likely to experience a number of barriers that prevent them regaining work;28–30 hence, among persons with disabilities who have experienced paid work, the loss or lack of work may have a greater negative psychological impact than among those without disabilities. These barriers include perceived negative employer attitudes towards hiring workers with disabilities, absence of or unreliable transportation to and from job sites, and fear of lacking the skills to compete successfully within the labour market.28 Other problems connected to job seeking include concerns about finding a job to match a person's skills, difficulties in balancing competing health issues and family responsibilities, as well as fear of losing government support benefits.30 These problems may then compound the adverse psychological effects and mental health effects associated with unemployment in the general population, which are believed to be brought about by a number of factors, such as loss of work role identity, financial stress and worry about effects on family.31 Greater difference in mental health among those with disabilities when unemployed or economically inactive is also likely to be related to their generally greater economic and social disadvantage3–5 compared to those without disabilities. Our results could also be explained by work having a greater positive influence among persons with disabilities as it may counter the negative impacts of social isolation and promote self-efficacy, self-esteem and a sense of acceptance by society. To the extent that this is the case, the loss of or lack of paid work could be felt more strongly among those with disabilities.
In the present study, the adverse influence of economic inactivity was also shown to be worse among those with disabilities than those without. The fact that the economically inactive category for those persons with a disability was associated with the largest reduction in mental health may reflect their exit from work as a consequence of a pre-existing disabling health condition.7 ,32 ,33 It is also worth considering that people who reported themselves as economically inactive may have actually been ‘frustrated job seekers’ who had given up looking for work (as, in Australia, the definition of unemployment is based on looking for work within a 4-week period).34
The results of this paper emphasise the importance of employment for those with disabilities as persons who were employed had significantly better mental health than when they were unemployed or economically inactive. These findings align with a recent qualitative review in the area, which reported that for most persons with a disability, work was a meaningful and important source of identity and self-efficacy and provided feelings of normality, financial support and socialisation.11 Further, qualitative research on the value of occupational engagement as a source of meaning following a serious and debilitating illness such as cancer suggests that work represents a way for persons to regain a sense of control and normalcy in their lives.35
We note several aspects of this study that limit the extent to which we can generalise results. First of all, it is likely that the HILDA survey is biased towards those with relatively minor disabilities. As there is no consistent indication of how long a health condition or impairment needs to be present before it can be considered a disability, we based our definition on reporting in two consecutive waves in order to assess relatively permanent disabilities. Sensitivity tests indicate that changing the sample definition does not change the direction of results. It is also worth noting that our definition of disability is based on a variety of health conditions and impairments. We were unable to examine these separately due to sample size restrictions in relation to different labour force states. Last, although the magnitude of the differences between those with and without missing data was not substantial, there is a potential that this could have biased results. In particular, based on the above, we would speculate that the patterns of missingness in the MCS would have led to an underestimate in the relationship between labour force status and mental health.
At the same time, our study also has a number of strengths including its large sample size and longitudinal design with eight waves of data, which meant that multiple time-variant exposures could be assessed against a commonly used measure of mental health. We also used a fixed-effects regression approach, which controlled for many of the time-invariant sources of confounding that may have otherwise affected the results. We adjusted for important time-variant factors by including them in our analytic models.
We also note that the paper makes no distinction between the directionality of the relationship between disability, labour force status and mental health. There is evidence to suggest that those with poor health26 and a mental health problem36 are less likely to gain employment than those with better health, but also that the loss of a job leads to a decline in mental12 ,36 and physical health.37 Our analysis examined within-person changes in mental health in relation to different labour force states after controlling for a range of individual and demographic factors. Regardless of the cause, results suggest that people have better mental health when they are working than when they are without work. This is an important message from a policy perspective as it highlights the beneficial role of employment for those with disabilities.
Future quantitative research could investigate the experiences of persons with disabilities when becoming unemployed and the range of factors that discourage those with disabilities from seeking work. Research also needs to examine what factors predict positive employment outcomes among those with reported disabilities who obtain employment. Aside from having its own merit as a research topic, this future research would be invaluable in further informing national disability employment policies.
In conclusion, we found that unemployment and economic inactivity are associated with worse mental health than being in employment. Our results suggest that the magnitude of the effect sizes was greater among those with disabilities than among those without disabilities. These findings suggest that policies or programmes aimed at improving employment opportunities and retention of people with disabilities may also have the potential to influence mental health as well as providing work. In Australia, there are dedicated government services for job seekers with disabilities (eg, the ‘Disability Employment Services’). An evaluation published in 2012 suggests that these services have facilitated access to employment among job seekers with disabilities, but acknowledges that more work is needed to improve disability employment systems.38 Further, there is little information on the mental health effects of employment. Based on the results of the current paper, we would suggest that new policy development would benefit from further qualitative and quantitative research into the experience of paid employment and what it means to people with disabilities, as well as further research into the beneficial impacts of paid work on the health and well-being of people with disabilities.
What is already known on this subject
A vast amount of research suggests that people who are unemployed or economically inactive have worse mental health than those who are employed. There is little understanding of whether the relationship between employment status and mental health differs for those with and without a reported long-term disability. Evidence about this topic is important for guiding policy development in relation to employment for people with disabilities
What this study adds
The negative impact of unemployment and economic inactivity on mental health is greater in magnitude among those people with disabilities compared to those without. Among those people with disabilities, there were larger mental health differences between being employed versus being unemployed or economically inactive than were observed for those without disabilities. These findings suggest that policies or programmes aimed at improving employment opportunities for people with disabilities may have the potential to positively influence mental health as well as providing work.
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:
- Data supplement 1 - Online supplement
Contributors AM conceived the article, downloaded the data, conducted the analysis, retrieved all references and wrote the initial draft. AMK, ADL, RB and ZA helped with the interpretation of results and drafts of the paper. All authors made substantial contributions to the final draft.
Funding The HILDA Project was initiated and is funded by the Australian Government Department of Social Services (DSS) and is managed by the Melbourne Institute of Applied Economic and Social Research (Melbourne Institute). This work was supported by the Australian National Health and Medical Research Council (NHMRC) through project (grant #375196) and postdoctoral research fellow (to AM) support (NHMRC Capacity-Building grant #546248), and was supported by Centre grant funding (#2010-0509/1) from the Victorian Health Promotion Foundation, Melbourne.
Competing interests None.
Ethics approval Approval for the use of HILDA data was provided by the Government Department of Social Services.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement This paper uses unit record data from the Household, Income and Labour Dynamics in Australia (HILDA) Survey. The HILDA Project was initiated and is funded by the Australian Government Department of Social Services (DSS) and is managed by the Melbourne Institute of Applied Economic and Social Research (Melbourne Institute).
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.