Full length articleHealth inequalities in the South African elderly: The importance of the measure of social-economic status
Introduction
There is strong evidence for the existence of important gradients in health outcomes by socio-economic status in most countries (Meara et al., 2008, Schalick, 2000, Van Doorslaer, 1997, Wagstaff, 2000). These inequalities have been remarkably persistent in the face of policy actions that try to reduce inequality (Costa-Font and Hernández-Quevedo, 2012, Font et al., 2011, Victora, 2000) and there has been a call for improved policies to address and reduce health inequalities (Evans, 2001, Carr, 2004, Organization, 2000). In principle, health inequality could be defined as the differences in health across people (Gakidou et al., 2000) in much the same way as income inequality is defined as the differences in income across people. However, in social epidemiology, health inequality is usually defined in terms of differences in health across different socio-economic groups, that is, the gradient in health with socioeconomic status (Berkman et al., 2014).
This raises the issue of how to measure socio-economic status and a variety of approaches have been developed in the literature. A natural approach is to use household income as a metric. However measuring household income in surveys from developing countries is difficult. As a consequence, studies concerning child health have relied on a wealth index based on housing characteristics and ownership of consumer durables (Filmer and Pritchett, 2001). The use of this proxy measure has been based on the argument that the wealth index is highly correlated with household income per capita and that child health gradients are similar with both socioeconomic measures. Subsequent studies have confirmed that while the relationship between households ranked by quintiles of wealth and consumption measures is imperfect, child health gradients are similar in both approaches (Wagstaff and Watanabe, 2003). However more recent studies show that gradients in health care utilization may differ using the two approaches (Lindelow, 2006) and that the exact composition of the assets used to construct the index matters (Howe et al., 2008, Montgomery, 2000).
The wealth index approach to measuring socioeconomic status has been mainly used in child health studies. However, recent studies have explored adult health gradients using a household wealth index and found either no, or only a very small gradient, in many countries (Debpuur, 2010, Hirve, 2010, Kyobutungi et al., 2010, Mwanyangala, 2010, Ng, 2010, Razzaque, 2010, Van Minh, 2010, Xavier Gómez-Olivé, 2010). This conclusion has potentially important consequences for how we think about the health of the elderly and health policy. However, unlike the case of child health there has been little evaluation of how the health gradient using the wealth index compares to gradients in household income or consumption for adult and elderly health.
To address this issue, we use data from the first wave of the Health and Aging in Africa: A longitudinal Study of an INDEPTH Community in South Africa (HAALSI) that provides data on health outcomes, household consumption, and the household wealth index, to compare health gradients using different measures of socioeconomic status. We model our approach closely on previous work on this population by Gomez-Olive et al. (Xavier Gómez-Olivé, 2010) to provide comparability to our results. We construct three different summary measures of health and disability status; each based on a different collection of health variables. We measure socioeconomic status using household consumption adjusted for household size. There is an issue that consumption per capita may not be a good measure of household socioeconomic status if some consumption goods are shared within the family, and there are economies of scale in household consumption. Wagstaff and Watanabe (Wagstaff and Watanabe, 2003) suggest equivalent consumption, defined as consumption divided by the square root of household size, as a better indicator of household wellbeing. We find much stronger health gradients in equivalent consumption than consumption per capita.
While there is evidence of a mortality gradient with the wealth index (Kabudula, 2017), previous work in Agincourt (Xavier Gómez-Olivé, 2010) finds a shallow adult health gradient in the wealth index when not adjusting for the other covariates, and no gradient when adjusting for covariates. From a policy perspective the adjusted gradient is more important. Some of the unadjusted health gradient may be due to the correlation of socioeconomic status with exogenous personal characteristics that affect health, for example, sex, age, marital status, and national origin, which are unlikely to be affected by policies. In our analysis we also adjust the gradient for education status. There may be very long run policies to reduce health inequality in the elderly by equalizing educational opportunities; by controlling for education we rule this out and focus on the potential effect of policies that address inequality in the current generation of elderly whose education levels can be considered fixed. This is the appropriate adjustment to find the potential impact of polices, such as pensions social grants, that redistribute income and consumption to the elderly (Exworthy et al., 2003, Marmot, 2002, Marmot, 2012).
Our result undermines the use of the wealth index alone as a proxy for household consumption when studying health gradients in adult and elderly health. We find much steeper gradients in health in equivalent consumption, than in the wealth index, suggesting the potential for a much larger health impact for policies that redistribute income. Further, our results emphasize the need for studies that collect detailed household consumption data, as well as recording the asset holdings needed for the wealth index.
Section snippets
Methods
In this study we use data on the elderly from the first wave of the Health and Aging in Africa: A longitudinal Study of an INDEPTH Community in South Africa (HAALSI) study. HAALSI is a sister study to the Health and Retirement Study that collected data on 5059 respondents-a 85.9% response rate- aged 40 and older living permanently in the Agincourt Health and Demographic Surveillance Site (DSS). This interdisciplinary survey collected data on household economic conditions, demographics,
Results
Our approach is to examine the gradient of health in socioeconomic status conditional on demographic characteristics given by sex, age, education level, marital status, and country of birth. To allow comparison to previous work in Agincourt (Xavier Gómez-Olivé, 2010), we follow the same approach and use a logistic regression to evaluate the association between quintiles of socioeconomic status and being in the top two quintiles of health. In our models, like those in the literature, we control
Conclusion
The objective of this paper was to evaluate whether the choice of welfare measure makes a difference in the estimation of health gradients for the elderly in South Africa. Using data from an ageing survey in Agincourt, we show that health gradients in three different health indices are much steeper in equivalent consumption compared to the wealth index. Our results finding a shallow gradient in the wealth index are in line with several SAGE studies in recent years (Debpuur, 2010, Hirve, 2010,
Conflicts of interest
Dr. Riumallo–Herl, Professor Canning, and Dr. Kabudula report grants from National Institute of Aging, during the conduct of the study.
Acknowledgments
This work was supported by the National Institute of Aging at the National Institute of Health (1P01AG041710-01A1, HAALSI – Health and Aging in Africa: Longitudinal Studies of INDEPTH Communities). The Agincourt HDSS was supported by the Welcome Trust, UK (058893/Z/99/A, 069683/Z/02/Z, 085477/Z/08/Z and085477/B/08/Z), the University of the Witwatersrand and South African Medical Research Council.
References (41)
Income-related inequalities in health: some international comparisons
J. Health Econ.
(1997)- et al.
Measuring inequalities in health: what do we know? What do we need to know?
Health Policy
(2012) Explaining trends in inequities: evidence from Brazilian child health studies
The Lancet
(2000)Socioeconomic differences in mortality in the antiretroviral therapy era in Agincourt, rural South Africa, 2001–13: a population surveillance analysis
The Lancet Global Health
(2017)WHO European review of social determinants of health and the health divide
The Lancet
(2012)- et al.
Cut-point shift and index shift in self-reported health
J. Health Econ.
(2004) - et al.
The gap gets bigger: changes in mortality and life expectancy, by education, 1981–2000
Health Aff.
(2008) The widening gap in death rates among income groups in the United States from 1967 to 1986
Int. J. Health Serv.
(2000)Socioeconomic inequalities in child mortality: comparisons across nine developing countries
Bull. World Health Organ.
(2000)- et al.
Persistence despite action? Measuring the patterns of health inequality in England (1997–2007)
Health Policy
(2011)