Introduction

Nearly twenty percent of Australians report a disability, a prevalence similar to other developed countries [1,2]. People with disability are one of the most socio-economically disadvantaged groups in society [1]. In Australia, as in most other developed countries, people with disability have lower rates of employment, post-secondary education, income [3], and lower wealth [4]. Disabled people have poorer physical and mental health than those without disability, and many of the conditions they experience are unrelated to their impairment (e.g. diabetes and depression for people with intellectual impairments) [2]. There is some evidence to suggest that the relatively poorer health of people with disability may be partly explained by their disadvantaged living circumstances [5–7]. What is not known is whether the health consequences of acquiring a disability are influenced by access to pre-existing social, cultural or economic resources.

A number of studies have found an increase in psychological distress, depressive symptoms, and a reduction in subjective wellbeing following the acquisition of disability [8–10]. However, not all people who acquire a disability experience deterioration in their mental health and wellbeing [7]. The impacts of negative life events, such as the acquisition of a disability, on mental health may depend on individual resources such as personal resilience [11,12], experience of economic hardship [13], education [9], and social resources. While research on potential modifiers of the relationship between onset of a disability and mental health is scant, an Australian study of 136 young Australians showed that social and economic disadvantage prior to the onset of disability, including low levels of social support, was associated with poorer mental health following acquisition of disability [7]. A US longitudinal study of 478 adults found having above median wealth was associated with smaller decline in subjective wellbeing following acquisition of disability [14].

There is some evidence to show that inequalities in wealth within countries may contribute to health inequalities [11,15–18], particularly in the elderly [15,19]. While studies suggest that the association is independent of income and employment [17,20,21], the extent to which wealth is a causally associated with health is debated [22,23] as wealth may be a cause as well as a consequence of health [21]. Aside from having a direct effect on health, wealth may protect people who experience negative life events from deteriorations in health. Acquiring a disability in adulthood may lead to concerns regarding loss in future earnings because of being unable to work or needing to work fewer hours (potentially in less demanding, lower paid jobs). Wealth may provide a ‘safety net’ alleviating some of these concerns. Evidence from the Household, Income and Labour Dynamics in Australia survey (HILDA), on which this study is based, found that people who acquire a disability in adulthood were more likely to become unemployed or under-employed (i.e. employed in jobs for which they are over-educated and over-skilled) [24,25]. Wealth may also enable access to other resources, such as health and rehabilitative services, which could have mental health benefits.

The aim of this study is to assess whether pre-existing wealth moderates changes in mental health associated with acquisition of disability. The analysis includes nearly 2000 people who acquired a disability (at least two consecutive waves without disability followed by two consecutive waves with disability) in HILDA—a population-based panel survey of approximately 27 000 adults aged 15 years and older.

Methods

Results

There were 2112 persons (15 562 observations) who met our criteria for disability acquisition. There were complete data available for 98% of people, resulting in a final analytic sample of 1977 persons (14 039 observations). The mean number of observations (contributed annual waves of data) per person was 6.8. Further details of sample selection and missing data are presented in the flow diagram (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow diagram of sample selection and missing data (MCS, mental component summary; obs = observations; ppl = people).

https://doi.org/10.1371/journal.pone.0139708.g001

At baseline entry into the analytic sample, nearly a third of the sample were aged 60 years or older; 56% were employed and only one in eight were in the highest income quintile while nearly a third were in the lowest quintile. In terms of wealth, 37% were in the highest wealth tertile and 29% the lowest; 41% were in lowest debt tertile; and assets were evenly spread across the tertiles (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of sample at baseline entry into the analytic sample (n = 1977 persons).

https://doi.org/10.1371/journal.pone.0139708.t001

Table 2 shows the mean of the pooled within-individual MCS scores by disability status for each covariate (see Methods). The MCS score was lower when disability was reported than at baseline (no disability). There were obvious socio-economic gradients with lower mental health in the most disadvantaged groups (unemployed, high debt, low assets and low wealth). The MCS score was positively associated with age and men reported higher mental health scores than women.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Mean of participants’ pooled MCS score in waves reporting disability and no disability and mean differences between the pooled scores, with 95% confidence intervals (n = 1977, 13 518 observations).

https://doi.org/10.1371/journal.pone.0139708.t002

Regression analyses

There was statistical evidence to support the inclusion of a product term between disability acquisition and wealth. In the adjusted models the coefficients for the mid and lowest tertiles of wealth were -0.6 (95% CI -1.5, 0.3, p = 0.164) and -2.2 (95% CI -3.1, -1.2, p = <0.001) respectively. Therefore all results were reported across strata of wealth. While there was deterioration in mental health following acquisition of disability for all wealth tertiles, the difference was the greatest in the lowest tertile. There is more than a 2 point difference in the change in MCS score between the highest (-1.1, 95% CI -1.7, -0.5) and lowest tertiles (-3.3, 95%CI -4.0, -2.5) (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Linear fixed-effects regression coefficients for the difference in MCS score within-persons between waves reporting disability and no disability (n = 1977, observations = 13 518) for wealth tertiles separately, adjusted for age, employment and equivalised household disposable income.

https://doi.org/10.1371/journal.pone.0139708.t003

Results of our main analyses were robust to sensitivity analyses. Exclusion of people with psychological impairments attenuated the results slightly although the wealth gradient was still evident; the reduction in MCS in the lowest wealth tertile was 2.5 (95% CI -3.2, -1.7) compared to 0.6 (95% CI -1.2, -0.1) among people in the highest wealth tertile (S1 File). The results of the complete case analysis yielded almost identical results (S2 File). When wealth was categorised into quintiles, the strongest effect estimates for disability acquisition on MCS score were in the lowest two wealth quintiles Table C in S3 File). Finally, separate analyses of assets and debt demonstrated that, while people in all tertiles of assets experienced a decrease in MCS score after acquisition of disability, the effect estimates were largest in the lowest tertile (-3.0, 95% CI -3.8, -2.3) (S4 File). The magnitude of the difference in MCS score after disability acquisition was similar for the mid and lowest tertiles of debt (mid -2.4, 95% CI -3.1, -1.7; lowest -1.8, 95% CI -2.4, -1.3) (S5 File).

Discussion

We found that acquisition of a disability in adulthood was associated with deterioration in mental health. Those in the lowest wealth tertile experienced the greatest deterioration, with a greater than three point decline in MCS score compared to a 1.1 point decline for those in the highest wealth tertile. There was evidence of a negative additive interaction between wealth and disability with relative excess risk due to being in the lowest wealth tertile of about 2 points.

In this paper we used fixed-effects regression to examine within-person change in mental health—an approach we and others have used in a number of other studies of psychosocial working conditions [27], employment arrangements [38], housing affordability [39], and job quality [40,41] in HILDA. The effect estimates for other exposures have been in the range of 1–2.5 points on the MCS scale. In this paper, people with low wealth prior to disability acquisition had a 3.3 point reduction in their mental health, highlighting the importance of these findings. This reduction is particularly important given that this group also had the poorest mental health at baseline (see Table 2).

Our findings are consistent with those of Smith et al. (2005) who found that having net wealth below the median was associated with a greater reduction in subjective wellbeing following acquisition of disability. However, their study was restricted to five waves of data collected at two year intervals in which only 478 people acquired a disability. Further, disability acquisition was defined as one wave without a disability followed by one wave with disability, and thus it is likely the sample included people with transient disabilities [14].

The relationship between wealth and mental health may be driven by assets, debts or both. In HILDA assets and debts are positively correlated (r = 0.45) as wealthy households can more readily obtain loans [16]. In separate analyses of assets and debt we found that those who had the least assets had the greatest deterioration in mental health, with a similar effect estimate as that found for wealth. For debt, a more modest reduction was found in both the mid and lowest tertiles of debt. What remains to be understood are the relative effects of different forms of assets or debt. The main forms of assets in Australia are housing, equities and superannuation, while housing and credit cards are the biggest sources of debt [16]. We have previously shown in the same dataset that low income private renters have poorer mental health than low income home purchasers [42]. It is possible that having equity in a home is an important buffer when a disability is acquired.

It is possible that the impact of wealth varies according to the type of impairment people acquired. It was not possible to investigate this in this study. Although HILDA does collect information on the limitations people experience from the third wave onwards (e.g. sight problems not corrected by glasses, difficulty gripping things, limited use of legs or feet, difficulty learning and understanding things, mental illness or nervous condition). Using these questions respondents can be classified as having sensory and speech, physical, psychological, intellectual and other impairments. A large proportion of people were classified as having more than one impairment type making it difficult to separate out the effects of the different impairment types. Overall the most common type of impairment was physical (70%) followed by sensory and speech (30%); psychological (17%); intellectual (4%) and other (70%).

This study has a number of strengths. First, it is based on a large population-based longitudinal survey in which over 2000 people acquired disability over 12 years. Second, causally-robust fixed-effects regression models were used to control for time-invariant confounding and important time-varying confounders were adjusted for [34]. Third, we used a comprehensive measure of household wealth and were able to examine assets and debt separately.

There are also potential limitations including the amount of missing wealth data. Results from analyses of data with imputed wealth and complete case analysis were nearly identical. Another potential problem is that wealth may have been measured up to three years before the disability status and mental health. While wealth may be more stable than income, the recent volatility in the value of the major household assets—housing, equities and superannuation [16]–may mean that the accuracy of the wealth variable reduces as the time between the measurement of wealth and disability increases. It is likely that any measurement error due to this would be non-differential. Further limitations include the possibility of dependent misclassification bias which is more likely in studies where there are two or more subjectively reported variables (e.g. disability and mental health). The fixed-effects analysis mitigates against this as any time-invariant influences (e.g., negative affectivity) is controlled for. Attrition bias is another potential limitation, but loss to follow-up in HILDA was low (<10% for most waves) [26]. Additionally, HILDA may under-represent people with more severe disability for whom wealth may be more even more important. Finally, the results may not be transportable to other countries. Australia has a Disability Support Pension for people with disabilities who are unable to work which may buffer against some of the negative mental health effects of acquiring a disability. Research in other countries is required in order to investigate the generalizability of these results. In this analysis we investigated effects of disability acquisition on mental health averaged over all consecutive waves of data in which individuals reported a disability (on average three waves of data reporting disability, but up to 10 waves). It is possible that these effects vary over time and that an average effect would not fully explain the patterns of mental health subsequent to disability onset. However, a longitudinal analysis of disability acquisition and subjective wellbeing using the German Socio-Economic Panel and the British Household Panel Survey found no evidence of adaptation over time, indicating that participants maintained lower levels of wellbeing long after the onset of disability [8]. Future research could further examine whether these mental health effects change over time, using methods such as growth curve models or fixed-effects models with lagged effects.

The results of this study suggest that policies and programs should be developed so as to alleviate the additional mental health burden of acquiring a disability among individuals with low wealth.

Supporting Information

Acknowledgments

This paper uses unit record data from the Household, Income and Labour Dynamics in Australia (HILDA) Survey. The findings and views reported in this paper, however, are those of the author and should not be attributed to either DSS or the Melbourne Institute. We thank Professor Andrew Beer and Dr Emma Baker (University of Adelaide), Dr Shelley Mallet (Brotherhood of St Laurence), Kellie Horton (VicHealth) and Keran Howe (Women with Disabilities Victoria) who are investigators on the project.