What is the role of income in creating health inequalities? Evidence from cross-sectional and longitudinal studies

Inequalities in health have been demonstrated for over 150 years in Britain, and in recent decades have been the focus of increasing policy attention. However in order to tackle inequalities in health there needs to be a clear understanding of their underlying causes. Unfortunately, the existing literature often uses different measures of socioeconomic status (SES) interchangeably; or, pays little attention to how they are measured; is still mainly based on cross-sectional data; and, tends to ‘explain away’ associations by adjusting for numerous confounders. These characteristics are unhelpful in elucidating causal processes and hence identifying mechanisms for reducing inequalities. The set of papers presented here aimed to address these issues by focusing explicitly on the role of income in creating health inequalities in order to develop a better understanding of how policies might potentially use income as a means of reducing the health divide. The first paper based on the General Household Survey (GHS) examined the cross-sectional association between income and health. In comparison to the ‘gold standard’– net equivalent household income - other income measures tended to underestimate the strength of the association at the lower end of the distribution, as did imposing a linear function on it, when non-linear functions performed better statistically. The association was stronger for long-term measures of health status than for recent measures of health state. We also investigated the relative importance of different measures of SES for health, and found that income had a stronger association with all measures of health than did occupational class and education, but a similar association to measures of consumption based on tenure and car ownership. In the second paper, I analysed GHS data to explore the health of a particularly disadvantaged group in the UK – lone parents – and the extent to which low income might be the cause of their health disadvantage. Both lone mothers and fathers, compared to couple parents, had higher risk of ill health, across a range of measures, and income and other material resources accounted for one third to a half of this, depending on the health outcome. I also explored other possible explanations, such as health selection, other social support and length of time as a lone parent, although these analyses were limited by the data available and the cross-sectional nature of the study. The remainder of the papers employed longitudinal data to explore more effectively the relationship between income and health by considering its association over time. The third paper was the first British paper to examine income dynamics and health. Using six years of data from the British Household Panel Survey (BHPS) we found a non-linear association between income and subsequent health, controlling for prior health. Average income, and persistent poverty, across five years were more strongly associated with subsequent overall subjective health and limiting longstanding illness, while current income was more important for recent illness and psychosocial distress (measured by the General Health Questionnaire). Decreases in income were associated with raised health risk, but increases did not lead to reduced health problems. Income volatility (i.e. the size of change irrespective of direction) was also associated with health. Controlling for prior health and measuring income before the health outcome both suggested that the association between income and health may be causal. However, it is important to understand income’s role in broader causal pathways. Paper Four employed both the BHPS and the 1958 birth cohort (National Child Development Study (NCDS)) to examine the role of income in childhood and adulthood for health. Analyses of the NCDS showed that childhood income influenced adult health only indirectly through ‘health capital’ and income potential (education) at age 23. However, in the BHPS, having controlled for earlier health and education and the key social roles – parenting, marriage and employment - that determine income levels, average adult income over five years was still a significant predictor of subsequent health, although childhood SES measures were not. This paper included a policy analysis to assess the effectiveness of policies to reduce health inequalities and found not only that they would have a modest impact, but also that they could worsen inequalities for some groups in society. The final paper examined the inter-relationships in adulthood between a key cause of income change, income and health. Using 10 years of BHPS data, we examined the extent to which financial difficulties mediated the association between employment change and health. There were complex relationships depending on gender and prior circumstances. Moving out of employment into unemployment increased psychosocial distress, while moving back from unemployment to work improved it. Men retiring from non-manual jobs experienced an improvement in their mental health, while the health of those retiring from manual jobs declined. Women leaving work for family roles experienced a decline in their mental health, while moving back to work did not significantly improve it. Financial difficulties mediated these associations, attenuating the effects by approximately 30% for men and 16% for women. This paper demonstrated the complexity of many associations between SES and health over time, and the importance of considering them within appropriate pathways. Overall these papers were among the first to consider the association between income and health over time, and within a lifecourse setting; to investigate issues of income measurement and functional form; and, to compare the relative importance of income with other SES measures. In doing this, I considered how the associations varied by health outcome, by gender and at different life stages, and according to whether or not respondents were from manual or non-manual backgrounds or had pre-existing health conditions. Taken together the papers clearly demonstrate a non-linear relationship between income and health, with the steepest associations at the lower end of the distribution. They show that income is part of the pathway between social roles and health, but that it is not the whole explanation. All of this suggests that low income is an important cause of health inequalities and hence fiscal policies to improve the incomes of the poorest in society are a potential mechanism for reducing the health divide