Greater equality and better healthBMJ 2009; 339 doi: https://doi.org/10.1136/bmj.b4320 (Published 11 November 2009) Cite this as: BMJ 2009;339:b4320
- Kate E Pickett, professor of epidemiology1,
- Richard G Wilkinson, professor emeritus of social epidemiology2
- 1Department of Health Sciences, University of York, Heslington, York YO10 5DD
- 2Division of Epidemiology and Public Health, University of Nottingham Medical School, Nottingham NG7 2UH
In 1996, when discussing studies of income inequality and health, the editor of the BMJ wrote, “The big idea is that what matters in determining mortality and health in a society is less the overall wealth of that society and more how evenly wealth is distributed. The more equally wealth is distributed the better the health of that society.”1
Since then, the “big idea” has been widely tested. More than 200 peer reviewed studies of the association have been published. Income inequality has been variously associated with lower life expectancy, higher rates of infant and child mortality, shorter height, poor self reported health, low birth weight, AIDS, depression, mental illness, and obesity. In the linked meta-analysis (doi:10.1136/bmj.b4471), Kondo and colleagues further assess the association between income inequality and health.2
Perhaps because of the deep political implications of a causal relation between better health of the population and narrower differences between incomes, interpretations of the evidence have come to different conclusions.3 4 5 6 The controversial question is not whether more equal societies really do have better health, but why they do and whether it is an effect of inequality itself. We know that lower incomes and poorer conditions are related to worse health, but can the association between smaller differences in incomes and better population health be explained on that basis alone, or does inequality have broader psychosocial effects on population health?
Two rival explanations for the link between inequality and health have been proposed—“compositional” and “contextual” explanations. Compositional explanations suggest that more unequal societies have worse health simply because they have more poor people. Redistribution of income from rich to poor would be expected to improve average health if the poor spent the extra money on things that benefit health, such as better food and warmer housing, whereas reducing luxury expenditure among the rich had little effect on their health. Such explanations also imply that the relation between inequality and population health has little to do with inequality itself—for example, with social comparisons or hierarchy—but is merely a result of how individual incomes affect health.
In contrast, contextual explanations are more closely related to the idea that inequality is somehow divisive and socially corrosive. What we now know about the importance to health of psychosocial factors—including social status, friendship, social capital, and sense of control—makes contextual explanations increasingly plausible.7
Compositional effects can be separated from contextual effects by using multilevel statistical models, in which the societal association between inequality and health is adjusted for the health effects of the individual incomes of the population. Kondo and colleagues report a meta-analysis of multilevel studies including about 60 million people.2 They find that even after taking account of the effects of individual socioeconomic characteristics, health is better in more equal societies.
Because we are dealing with whole populations, the attributable risk of the contextual effects is large, even though the effect size is fairly small. For every 0.05 increase in the Gini coefficient of income inequality, mortality increased 7.8%, with an estimated 1.5 million excess deaths each year in 30 countries in the Organisation for Economic Cooperation and Development. But policy makers should not be misled. The distinction between compositional and contextual effects is merely a distinction between causal pathways. If income differences were made smaller, health would improve in two ways—from increasing the relative incomes of the poor, and from the wider contextual benefits of greater equality.
Multilevel modelling is likely to produce conservative estimates of the size of the contextual effects of inequality on health. The idea that individual socioeconomic characteristics affect health purely through material pathways, regardless of the wider social context, has become less plausible over the years. As Marmot and others have argued,8 an important part of why individual income is related to health probably reflects its role as a marker of social status. Similarly, the association between education and health probably reflects the importance not only of knowledge, but of the individual’s educational status relative to other people.9 Rather than a simple dichotomy between material effects of individual income unaffected by context, and context dependent psychosocial influences, psychosocial processes are probably important in both spheres.
When the empirical evidence of the effects of inequality was confined to health, it was reasonable to think that we should not assume that inequality had any psychosocial effects before we eliminated other possible explanations. But since then the evidence base has grown. It is now clear that unequal societies have an increased prevalence of a host of social problems, including violence, bullying, teenage births, higher rates of imprisonment, low educational performance, reduced social mobility, low levels of trust, and longer working hours.10 Insofar as these are behavioural outcomes, they provide strong evidence that psychosocial processes are associated with inequality.
The benefits of greater equality tend to be largest among the poor but seem to extend to almost everyone.10 A more equal society might improve most people’s quality of life. Rather than merely paying lip service to creating a “classless society,” it is a task for politicians and policy experts to repair our “broken society” by undoing the widening of inequalities that has taken place since the 1970s.
Cite this as: BMJ 2009;339:b4320
Competing interests: None declared.
Provenance and peer review: Commissioned; not externally peer reviewed.