The associations of household wealth and income with self-rated health – A study on economic advantage in middle-aged Finnish men and women

Abstract

The economic resources available to an individual or a household have been hypothesised to affect health through the direct material effects of living conditions as well as through social comparison and experiences of deprivation. The focus so far has been mainly on current individual or household income, and there is a lack of studies on wealth, a potentially relevant part of household resources. We studied the associations of household wealth and household income with self-rated health, and addressed some theoretical issues related to economic advantage and health. The data were from questionnaire survey of Finnish men and women aged from 45 to 67 years, who were employed by the City of Helsinki from five to seven years before the collection of the data in 2007. We found household wealth to have a strong and consistent association with self-rated health, poor health decreasing with increasing wealth. The relationship was only partly attributable to the association of wealth with employment status, household income, work conditions and health-related behaviour. In contrast, the association of household income with self-rated health was greatly attenuated by taking into account employment status and wealth, and even further attenuated by work conditions. The results suggested a significant contribution of wealth differentials to differences in health status. The insufficiency of current income as the only measure of material welfare was demonstrated. Conditions associated with long-term accumulation of material welfare may be a significant aspect of the causal processes that lead to socioeconomic inequalities in ill health.

Introduction

Studies on inequalities in economic resources as a cause of inequalities in ill health have mostly focused on the effect of income level on mortality and poor overall health status, and there is very little literature on the contribution of household wealth. It is likely, as we discuss in more detail later, that the effects of wealth and income on health are to a large part interconnected, and the mediating mechanisms are likely to be similar. Nevertheless, as wealth is an important part of household economic resources, focusing on income only may be an impartial approach to study the potential effects of household economic position on health. Several previous authors (e.g. Braveman et al., 2005, Ecob and Davey Smith, 1999) have noted the problems of assuming that income data alone can account for the living standard of individuals and households.

Household wealth is likely to reflect long-term living conditions, which is of particular importance given that income data is often available for comparably short time periods only. Data from Britain have showed that there is notable short-term variation in income particularly at the extremes of the income distribution (Jarvis & Jenkins, 1998). Given that illness, disease and death can in most cases plausibly be understood as phenomena caused by an accumulation of influences over the life course rather than momentary risks of becoming ill at a certain point of time, it is justified to assume that the long-term accumulation of economic resources is important. Previous results showing stronger association of long-term than current income with health (Benzeval & Judge, 2001) support this assumption. The distribution of wealth is also more unequal than that of income, indicating that additional information on social inequalities is contained in data on wealth.

Whether economic advantage contributes to inequalities in ill health largely involves the question of whether the standard of living and consumption potential affect health. If income and wealth can be attributed an independent causal contribution to social inequalities in health, the mechanisms mediating this effect are likely to involve consumption potential in some way. Although many other social advantages correlate with wealth and income, the degree to which they can be seen as caused by economic position is questionable. Although income determines current consumption potential more directly, wealth may affect living costs and contributes, in particular, to the acquisition of permanent resources, such as real estate or vehicles.

Three themes are central to explanation of the association of economic position with ill health and mortality. The first of these concerns the degree to which ill health leads to disadvantaged economic position. It is evident that ill health can seriously affect a person’s ability to secure gainful employment, and thus it is mainly a question of the magnitude and scope of such processes in a population compared with the potential causal pathways.

Secondly, the association could reflect the covariation of economic advantage with other aspects of advantaged socioeconomic position, and not be genuinely causal. To be more exact, the issue is whether the differences in consumption potential and in the material standard of living cause the association, and not the fact that income covaries with exposure to detrimental work conditions, for example.

Thirdly, even if it is accepted that economic position has a causal effect on the development of ill health and mortality, there is still the question of why the ability to consume is important to health. Causal explanations about the nature of the interdependence fall into two main categories.

The material explanation is that commodities and services acquired through the use of private economic resources affect health, and that commodities that people with little money can afford to acquire are significantly more detrimental to health than the corresponding commodities people with more money to spend can afford. This hypothesis further requires that people acquire commodities that are beneficial, or in many cases less damaging, to health, if they have the necessary financial resources to do so (although not necessarily consciously in the interest of promoting their health).

The second type of explanation concerns how people see their and others’ positions in society, and the degree to which commodities as signs of status capable of influencing human interaction rather than concrete objects and physical environments are related to mental states that are detrimental to health. This explanation is related to the theoretical notion of relative deprivation, according to which the inability to acquire what is considered to belong to the good life and thus the inability to lead a good life relative to social norms, and the lack of social status attached to such acquisitions, cause chronic mental distress (Wilkinson, 1999). It could also be said that the distress is caused not by the person’s interpretation as such, but by other people’s behaviour towards someone who lacks the signs of status. According to stress theory it could be assumed that distress caused by low status may lead to somatic disease and ultimately even death (Marmot, 2005).

Both types of explanation assign causal importance to income level, but in the case of the latter this is not necessarily only to do with consumption potential. Both income as such and the commodities it makes available may be relevant.

One empirical approach to testing whether limited consumption potential causes ill health and mortality is to use different measures of income. Household income net of taxes and income transfers and adjusted for household size is the closest operationalisation of the degree of freedom in consumption choice, and could be assumed to be the best estimate of causal effects related to consumption. Nordic studies have found weaker associations with ill health and mortality for household disposable income than for gross household or individual income (Fritzell et al., 2004, Martikainen et al., 2009; Rahkonen, Arber, Lahelma, Martikainen & Silventoinen, 2000), which suggests that a significant proportion of the observed inequalities in ill health measured in terms of individual gross income are the result of factors other than the causal effect of consumption (Martikainen et al., 2009).

Another discussion concerns the form of the association of income with mortality and ill health. There is a widely held view that a curvilinear association with diminishing health gains relative to increasing income is more compatible with the causal effect of standard of living than a linear association (see e.g. Der, 2001), although the grounds for this assumption are often not clearly indicated. It is not unproblematic to claim that differences in material living conditions are less important above the average income level than below it given the lack of data on actual living conditions. Nevertheless, empirical results from the U.S. have confirmed a curvilinear association with mortality (Backlund, Sorlie, & Johnson, 1996), the effect of income levelling off at the higher end of the range. Similar curvilinearity has also been found for the association of income with self-reported morbidity measures in Swedish and British populations (Blaxter, 1990, Ecob and Davey Smith, 1999, Fritzell et al., 2004) although different measures of health applied to the same data showed both curvilinear and linear associations (Der, Macintyre, Ford, Hunt, & West, 1999). In a study comparing seven Western European countries, Mackenbach et al. (2005) found a clearly less steep effect on self-rated health in the highest income quartile than in mid-range incomes in them all, but the results for the lowest income quartile varied. Mortality data from Finland covering 1991–1996 showed a mainly linear association, with only weak signs of a levelling off at high incomes (Martikainen, Mäkelä, Koskinen, & Valkonen, 2001), whereas recent Finnish data on mortality during the period from 1998 to 2004 indicated curvilinear associations before adjustment for other aspects of socioeconomic position, but only weak associations after adjustment (Martikainen et al., 2009).

There have been attempts to separate objective purchasing power and relative income level (Åberg Yngwe, Fritzell, Burström, & Lundberg, 2005). However, the problem seems to be that when income measured in absolute units is accounted for the remaining effect of the distance of an individual from the population average income potentially measures for the effects of all other dimensions of socioeconomic position. Such effects are not necessarily causally mediated by ‘relative deprivation stress’. Pathways could include the effects of social class mediated through work conditions, and the potential effects of education mediated through behavioural patterns.

Some authors have argued that area-level income inequalities could be used as an indirect test of the contribution of social comparison to inequalities in health (Kawachi, Subramanian, & Almeida-Filho, 2002). Studies comparing the effects of household income and income inequalities on ill health and mortality across different states in the U.S. have reported some contributions of income inequalities to morbidity and mortality in excess of individual income (Kennedy et al., 1998, Lochner et al., 2001). It is nevertheless questionable whether fractionalising the effect between individual and area-level measurements of income can reveal very much about the underlying mechanisms of the effect of income level on ill health when no other information on living conditions is available. Either measurement could relate to both material effects and social comparison.

Household wealth has been addressed in previous studies mainly using proxies such as house ownership or access to car. Several British studies (e.g. Blane et al., 1997, Macintyre et al., 1998), as well as a recent Finnish study (Laaksonen, Martikainen, Nihtilä, Rahkonen, & Lahelma, 2008) have found associations of home ownership with illness and mortality in excess of income, and these results may be interpreted as pointing to the importance of accumulated wealth as a material resource. Studies directly focusing on wealth and ill health are rare, however. Results from the Whitehall studies on middle-aged British civil servants have indicated that wealth is associated with lower prevalence of poor self-rated health and clinical conditions in excess of the effect of income (Martikainen et al., 2003, Perel et al., 2006). Other recent studies have also reported associations of household wealth with stroke (Avendano & Glymour, 2008) and with psychological distress (Carter, Blakely, Collings, Gunasekara, & Richardson, 2008).

Further studies on the potential effect of wealth on illness could provide valuable evidence to strengthen the general inference about the contribution of economic resources and the standard of living to inequalities in ill health. Thus our objective was to study the effects of household wealth as well as household income on less-than-good self-rated health. The study was cross-sectional, but in taking into consideration potential reverse causation we included employment status as a covariate in the analyses. We also included factors that might covary with economic resources but are not caused by them in an attempt to counter the confounding from other aspects of socioeconomic position. We did not adjust for social class or education because it seemed more justified to adjust for the conditions that are hypothesised to mediate their effects. Therefore we included measures of work conditions in the analysis. Health-related behavioural patterns were also considered, although it is much less clear whether they are causally independent of economic resources. However, testing for the contribution of health-related behaviour could help in identifying the potential mechanisms determining the effects of economic resources on health.