Decomposing the causes of the SES-health gradient with biometrical modeling
Garrison, Sarah Mason
:
2016-11-21
Abstract
The consistent relationship between socioeconomic status (SES) and health has been widely covered in both the popular media and in scientific journals. These articles, both popular and scientific, draw strongly upon conventional explanations, that physical health inequalities are caused by material disadvantage directly (e.g., access to medical care; Hummer, Rogers, & Eberstein, 1998) or indirectly (e.g., chronic environmental stress; Baum, Garofalo, & Yali, 1999; McEwen & Stellar, 1993). Such explanations account for differences between those who have resources and those who do not, but they do not account for the finely stratified health differences that exist across the entire range of SES (see Gottfredson, 2004, for discussion of additional limitations and contradictions).
Recent theories to explain the SES-health gradient have grappled with limitations and contradictions of early models, and have implicated very different pathways to explain the gradient. For example, articles in differential epidemiology have argued that individual differences, such as personality and cognitive ability, are the ‘fundamental cause’ of the
gradient, acting through genetic sources (Arden et al., 2016; Marioni et al., 2014). Alternatively, variants of the allostatic load theory (McEwen & Stellar, 1993), like the Risky
Family model (Repetti, Taylor, & Seeman, 2002) implicate the early home-environment. Surprisingly, very little research has applied behavior genetic modeling to understanding
the sources of the SES-health gradient.
The purpose of this paper is to narrow the scope of fundamental causes, by explicitly untangling the sources of variance associated with the gradient. Specifically, we decompose the gradient into genetic (a²), shared-environmental (c²), and non-shared environmental (e²) pathways, using data from the National Longitudinal Survey of Youth 1979. Monozygotic-twin, dizygotic-twin, full-sibling, half-sibling, and cousin pairs from recently validated kinship links are used in the current study (n= 4018 pairs; J. L. Rodgers et al., 2016).
Our findings have decomposed the relationship between SES and health into its biometrical components. Our results differed by health construct. Mental health’s relationship with SES is primarily explained by the shared-environment, which is consistent with the Risky Family model of the gradient. The relation of physical health relationship with SES is primarily explained by genetic effects.