Pro-rich health inequalities, which favor groups who are advantaged either with respect to education or wealth, are consistent and statistically significant in each of the domains of health and in health overall. There was a descending gradient in the prevalence of poor health, moving from the poorest wealth quintile to richest, and moving from the lowest to the highest educated groups.
In the combined countries data set, in all the domains and in overall health, the education-related inequality in poor health was higher than the wealth-related inequality in poor health. Adjusting for the effects of sex, urban/rural area, marital status and education or wealth as other possible confounders on top of age and country of residence attenuated our measured inequalities, but all of the inequalities remained statistically significant. These findings are consistent with the literature reporting positive associations between SES (measured by education, income and wealth) and SRH [8
Reports of health status incorporate complex combinations of an individual's assessment about their health and health conditions [6
]. Considered alone, summary measures of overall health mask differences in the domains of health that may be important to know about in order to target specific interventions. This study takes the additional step of assessing socioeconomic inequalities in the individual domains of health and comparing these measures with inequalities in overall health measured separately. The findings contribute to current evidence of health inequalities by reporting statistically significant inequalities in poor health separately within the widely used domains of health developed by the WHO.
In the combined countries data set and for both country income groups, inequalities were highest for self-care, cognition, vision and mobility. It is possible that correlations between the domains to some extent influence these results, for example persons reporting poor self-care may also report poor mobility and poor cognition. Sadana et al. [17
] assessed correlations between six WHO health domains (affect, cognition, mobility, pain, self-care and usual activities) and overall health from 66 surveys carried out by the WHO and showed relatively high correlations between self-care and mobility and self-care and cognition (Rho: 0.76 and 0.59, respectively).
Associations between SES and aspects of SRH [22
] are widely documented. The ageing of populations has increased interest in cognitive function as a public health issue. There is now a mounting body of evidence that low SES, measured by education, occupation, income, and ownership of financial assets, predicts decline in cognitive function in older adults [37
Internationally there is a recognized need to address the "social determinants of health" [4
]. Yet taking action to reduce inequalities and inequities in health within countries requires understanding of how social and economic factors are associated with all of the key components of health, as well as health overall. This is the first study of its kind to provide evidence of associations between education and wealth specifically in all eight domains of health at a pooled multi-country level, and also by comparing groups of lower- and higher-income countries. The analysis by country income groups provides additional insights into the patterning of relative socioeconomic inequalities in the domains of health. Although, on average, health is better in the higher-income countries, the distribution of individual health states in accordance with educational rank as well as wealth rank is more unequal in this group.
The large multi-country dataset allowed us to assess socioeconomic inequalities in the health domains and in overall health. We have ensured comparability of data between countries by using the WHS. Inequalities in health are measured here by SRH. Self-rated health instruments are applicable within many cultural, demographic and socioeconomic settings [10
] and are strong predictors of health care utilization, morbidity and mortality outcomes [12
This work is based on the domains of health developed by the WHO. They provide a consistent validated way of describing and comparing population health within and between countries [2
Higher SES is associated with better living standards, the most direct (and popular) measures of which are income and consumption expenditure. However measuring income and consumption expenditure can be problematic. In high-income countries, consumption expenditure patterns are very complex and income can be a better measure, and in developing countries, where formal employment is less common and home production widespread, consumption expenditure can be a more accurate measure of living standards [48
]. This study uses a method to estimate household wealth that is based on the premise that wealthier households are more likely to own a given set of assets. In addition to being consistent with a broader definition of poverty, this approach provides a way of drawing international comparisons when analyzing health inequalities [28
It is important to consider evidence of socioeconomic inequalities when developing interventions that target specific domains (e.g. mobility, pain and cognition), otherwise interventions may widen inequalities in health. The results from this study are of relevance to public health policy-makers and others because they identify inequalities in the individual health domains.
The data are cross sectional and so can only describe associations between socioeconomic factors and health. The results show that health inequalities according to education and wealth exist after adjusting for age, at a point in time, but they do not explain casual relationships or health change over time. There is a need for research that examines ways in which socioeconomic factors mediate changes in health domains as well as overall health.
The results of this study are based on self-reported information about health. Future studies need to incorporate health examinations and biomarkers within household surveys in order to improve the validity of self-reported health states and to detect and correct systematic reporting biases.
The countries were not probabilistically selected and therefore not necessarily representative of the world or of similar groups of countries (e.g. defined by geography or income). The use of pooled data masks possibly important variation between countries. For example, in order to tailor intervention programs aimed at improving health it is important to understand SRH within local context and culture and these aspects are not captured in our data [44
]. Although it was not the purpose of this study to examine inter-country variation, we did include a "country" variable to control for any potential confounding effects related to individual countries.
There is evidence that relative to advantaged groups, disadvantaged groups may fail to perceive and report the presence of illness or health deficits which may result in misleading assessments of population health [9
]. However if such bias exists in our study, then it is likely that the results underestimate the true size of education-and wealth-related inequalities in the domains of health and health overall.
Lastly, we used RII which is a relative measure of inequality that is adjusted for variation in average prevalence of poor health across health domains. As such, information on the absolute size of differences in prevalence of poor health between socioeconomic groups is not reflected in RII [50
]. To address this we have provided prevalence estimates of poor health by wealth quintile and educational level.
Deconstructing inequalities in health in different domains, separately by wealth and education in lower- and higher-income country groups, paints a much more nuanced picture than would be otherwise visible. Major differences between the country groups highlight the complex interaction between health, a country's level of economic development and individual socioeconomic status.