Mel Bartley thanks the Centre for Longitudinal Studies, Institute of Education, for the use of the 1958 birth cohort data and the United Kingdom Data Archive and Economic and Social Data Service for making those data available. None of the organizations that provided access to data had responsibility for the analysis or interpretation of these data.

This book owes a great deal to past and present colleagues, in particular Amanda Sacker, Meena Kumari, Pekka Martikainen, Archana Sing-Manoux, Paul Clarke, Mai Stafford and Eric Brunner, Updating the British official data on health inequality would not have been possible without the help of Peter Goldblatt.

Many of those who studied health inequality in the 1970s and 1980s saw it as a chance to improve our understanding of disease and our ability to prevent ill health in the whole population, regardless of social background. Social differences in premature mortality from almost all causes were so large; it seemed that if we could understand the ways in which socio-economic adversity ‘gets under the skin’ to produce disease, this might lead to major improvements in population health and ways of making medical care more effective. What is the point, we asked, of treating someone’s bronchitis in hospital and then sending them back to a damp, cold house to become ill again? The existence of health inequalities was seen as proof that diseases were indeed preventable by changes to the environment. Since 1948 the British National Health Service offered medical care free at the time of use to all citizens, but by the 1970s it was clear that free health care was not reducing the size of health inequalities; in fact, they increased (see chapter 2). This was all the more encouragement to regard prevention as better than cure. After all, even free operations are not a pleasant experience; most people would prefer not to become ill in the first place. But more surprising was that health inequalities continued to increase even after 40 years of a welfare state that in theory prevented the worst extremes of poverty and that had actually succeeded in reducing income inequality.

The Black Report of 1980 was the first attempt to drill down into the statistics and identify what it was about social class (the measure of social inequality used in British official statistics) that produced these large differences in risk. The Black Report put forward four possible models of explanation: selection, artefact, material and behavioural-cultural. In the early 2000s, the artefact explanation had been completely discredited and the selection explanation was little considered outside of economics. Accordingly, the first edition of this book concentrated on the material and behavioural-cultural models and tried to assess how well each of them fitted the existing evidence. It also described and assessed three additional factors that had been increasingly investigated in research between 1980 and 2000: psycho-social stress at work, social isolation and life-course effects. In this second edition, this evidence will be updated, but consideration will also be given to some of the more recent thinking that has arisen, particularly what the evidence has to say about the importance of individual characteristics for health inequality.

Between 1997 and 2001, when the ideas behind this book were being gestated and written down, the United Kingdom had a ‘New Labour’ (moderate social democratic) government that was in many ways determined to reduce health inequality and was prepared to do so by addressing at least some of the social and economic factors (the so-called ‘upstream factors’) believed to be involved, such as the absence of a minimum wage, low pensions, widespread educational failure and child poverty. The successor to the Black Report, the Acheson Report, set out a large number of rather precise recommendations (Acheson 1998), and a government plan published in 2002 described the ways in which many of these would be met (Department of Health 2002). As pointed out by Mackenbach (Mackenbach 2010a, 2010b), the British programme was ‘by far the best resourced of all the Western European strategies to reduce health inequalities which started during the decade’. It should have been a golden age during which the research paid off in terms of real-world reductions in health disparities between social groups, and overall improvements in population health.

So it has been surprising and dismaying to see the verdict on the past 10 years of policy initiatives: that they have not succeeded in reducing health inequality in the United Kingdom (Department of Health 2009; Mackenbach 2010a; Law, Parkin and Lewis 2012). At least, not as far as we can tell. It is not easy to trace the changes in health inequality since 1997 because the availability of data that can be used to evaluate the success or failure of new policies has changed. This has made it quite hard to compare the situation now with that in 1997 and 2014 (let alone any further back). When I decided it was time to update Health Inequality, the first thing to do was to revise the tables and figures. This seemed like it would be a simple task. But it did not turn out that way. For a start, the evaluations proposed in the 2002 report ‘Tackling Health Inequalities’ only required a comparison of mortality by geographical areas, not by social class as had been done every 10 years since 1921 in England and Wales. Social class gaps were only assessed in infant mortality. It appears that the degree of class inequality in infant mortality did begin to fall quite a lot (Bambra 2012), though it may have risen again after 2010. The gaps in adult life expectancy between the most and least deprived areas, at least in England, have remained stubbornly unchanged, or even slightly increased (Law et al. 2012; Vizard and Obolenskaya 2013). But that is not the same as being able to extend the 70-year old analysis of social class differences in premature death that gave rise to the whole issue of health inequality in the first place.

Should we be surprised that policy measures left the health gap between more and less advantaged citizens untouched? Many people do not think so (Howden-Chapman 2010; Mackenbach 2010a, 2010c). When, after its election in 1997, the New Labour government, so as to design policies, asked the experts what were the causes of health inequality, the answers were all too vague.

Part of the motivation for the first edition of Health Inequality was to set out the different explanations that were being offered, and how they were being tested, in such a way as to encourage a wider group of people to participate in the search for explanations and hence for solutions. But this is not what has happened. Health inequality research seems to have become rather an arcane topic beset with mysterious numbers. The influential think tank the Kings Fund published blogs pointing out the retreat from serious consideration (Buck 2014). There seems to have been a reduction in the gathering and use of evidence of changes in the differences in health between more and less socially advantaged groups. Kat Smith’s study, which is one of the recommended readings to accompany this chapter, goes ‘Beyond evidence-based policy’ (Smith 2013) to give an instructive account of the politics of research in this area. Once it was shown that policy initiatives did not seem to be effective, what took over was an increase in philosophical arguments about the injustice of health inequality, rather than detailed attention as to why policy changes had had such little effect. No one, for example, went out and asked citizens in the most deprived areas of Britain (termed ‘spearhead’ areas for policy purposes) why the energetic attempts to discourage smoking in these areas had had so little influence.

Another view of the political paralysis, however, is that we can now return to the more serious science involved in understanding health inequality. There have been major advances in related areas of research that will, I am sure, have considerable effects on our understanding. There are now several large international data sets, designed to make the experiences of people in different countries comparable. One of these, the Survey of Health, Ageing and Retirement in Europe (SHARE), is an international study of ageing which focuses on health. Although SHARE does not contain data on all phases of the life course, participants complete a life-grid in which they report on major life changes after young adulthood. Another is EU-SILC, the European Study of Income and Living Conditions, which collects far fewer data on health but is beginning to be used in interesting ways. Outside of SHARE, there now exists a collection of comparable ageing studies in the United Kingdom (the English Longitudinal Study of Ageing: ELSA), the United States (the Health and Retirement Study: HRS), the Korean longitudinal ageing study (KLOSE) and the Japanese longitudinal study of older people (J-STAR), with other nations joining as time passes. The first hugely influential comparative studies using these data to compare health in older people in the United Kingdom, the United States and Europe showed large health disadvantages in the United States compared to English citizens, and similar US health deficits in older people compared to European nations (Banks et al. 2006; Avendano et al. 2009). However, these studies have not as yet been used to throw light on inequalities in health between social groups within nations.

How should a second edition of this book respond to these changes? One possibility would be to stick to the initial basic structure of the first edition in term of the models that attempt to explain health inequality. Is the classification of approaches into material, behavioural, psycho-social and life-course models still useful when we try to understand more recent trends? It is interesting that, while the amount of commentary on national trends in health equality has declined, there are still fairly lively debates about the relative importance of risky health behaviour (Dunn 2010; Stringhini et al. 2010, 2011; Nandi, Glymour and Subramanian 2014) and psycho-social factors such as work stress, effort−reward imbalance and work−life balance (Kivimäki et al. 2012; Landsbergis, Dobson and Schnall 2013; Theorell 2014). There has also been a strong revival in the ‘indirect selection’ model in the shape of a series of papers showing strong associations between childhood cognitive abilities and adult health (Batty et al. 2007a, 2007b). Not surprisingly, genetic research has gone from strength to strength as the ability to analyse the human genome has improved due to technical advances, although as yet there is little evidence of any association of genetic variants with a measure of social position such as social class or income.

The most important advance in social epidemiology since 2004, however, has been the rapid increase in understanding the life course (Power and Hertzman 1997; Wadsworth 1997; Hertzman et al. 2001; Case, Fertig and Paxson 2005). This is now possible because of the maturing of the British birth cohorts, representative samples of citizens born in 1946, 1958, 1970 and 2000 (see chapter 10). Detailed health measures are now also available for all participants in the British Household Panel Study and its far larger (40,000 households) successor, Understanding Society. So the content of the different aetiological models needs to be revised in the light of what has happened in life-course research. No one thinks any more that cross-sectional (relating health at one time point to social circumstances at the same time point) research on work stress, poor housing or even smoking can enlighten us any further or help in the design of more effective policies. This is partly because there are now so many new and extended data sets available that trace what happens to people either from birth or over many years of adulthood and into old age. When we observe the health of people in a certain social class, we now know that we are observing the accumulated consequences of their material, emotional and cultural histories. ‘Selection’ arguments always were life-course explanations. But we are now in a far better position to investigate what went on before; for example, poor school performance led to a risky job in adulthood, or low self-efficacy resulted in addiction to tobacco (Sweeting et al. 2016). This new capability in health research has been enriched by an important innovation, a focus on life-course justice. Researchers in France and the United Kingdom are beginning to ask: Can we regard differences in health behaviours as a matter of personal choice? Or, if such behaviours are seen in life-course studies to be strongly linked to disadvantage in early life, should inequalities in smoking, exercise and diet also be regarded as unjust (Jusot, Tubeuf and Trannoy 2013)?

In the first edition, a whole chapter was devoted to the work of Wilkinson and others on the relationship of health to income inequality at the macro-level of whole nations or other geopolitical units. But at this level, income inequality is no longer the only variable of interest to health inequality researchers. In the years since 2002, a lot of research has been carried out into health and health inequality in different welfare regimes. It took more than a decade for the path-breaking work of Esping-Andersen (1990) to penetrate social epidemiology. But once it had been realized that it was highly plausible to link the commodification of labour power with poor health and the de-commodification with better health, this became a welcome task. The revised chapter on the macro-level determinants of health (chapter 7) will therefore need to put this work alongside that of Wilkinson and Pickett’s powerful analysis of overall income inequality (Wilkinson and Pickett 2009). To somewhat anticipate the contents of this chapter, the weight of evidence shows that more generous welfare regimes do not have lower levels of health inequality. This has been one reason for the revival of interest in the importance of selection by personal characteristics (Mackenbach 2012).

Despite, or perhaps because of, the failure to show that what looked like promising policies, or policy changes, are linked to lower health inequality, there is now a rising concern with the more philosophical issues involved. A leading researcher, author and proponent of policy change is Michael Marmot, whose Whitehall II study has figured prominently in the research literature for more than 30 years. Whitehall II findings figured prominently in the first edition of Health Inequality. The recent monitoring exercises of the successes and failures of British policies have mostly been carried out by Marmot’s group. At the same time, some of the Whitehall II research, particularly into the importance of psycho-social work conditions and the relative importance of behavioural versus material and psycho-social factors, has come under attack. In fact, these have been some of the most controversial contributions to the research base since 2004. However, increasingly Marmot has been on record as emphasizing the philosophical and ethical over the scientific importance of health inequality. Health inequality, he and his colleagues argue, is fundamentally a matter of social justice (Marmot, Allen and Goldblatt 2010). In his view, we will know when we have socially just policies because at this point social inequalities in health will start to decline. This is, for some people, a long way from an initial interest in the phenomenon because of what it might be able to tell us about the ways in which disease arises and thereby how to prevent disease. It is as if we have more or less given up on understanding health inequality well enough to design appropriate policies. Rather, policies must be changed for the sake of fairness, and a decline in health disparities will act as evidence that new policies are really fair. To take an example, 40 years of anti-smoking advice has failed to reduce inequalities in premature death from smoking-related disease. This has to be one of the most reliable bits of evidence ever derived from a public health intervention. The response − that when we start to see a reduction in the differences in smoking between social classes we will know that policies (economic and social) have become more just − is actually rather appealing, certainly more so than the prospect of endlessly repeated anti-smoking advice.

Some of the most exciting advances in research are now combining philosophical concepts of freedom and fairness with a life-course approach to take a more fruitful perspective (Jusot et al. 2013) and to revive the area as a ‘progressive research programme’. I am going to take the view that this interdisciplinary programme is in the process of revolutionizing our work. However, problems do remain, as will become evident in the ‘applied’ chapters of the book on ethnicity and gender. It may be that applying a social justice approach over the whole of the life course will do more to counter health disparities in the longer term. But we also know that population health can respond very quickly to policy changes that occur during adulthood. A life-course approach must not become a reason to ignore adult social conditions at work and at home (Edwards, Gillies and Horsley 2014).

It is not the purpose of this book to set out a general theory of health inequality. Rather, it aims to provide readers with the wherewithal to understand and evaluate arguments and explanations put forward by other researchers (and possibly to think up their own explanations). However, the researcher should admit her own biases and prejudices, and of course I do have my own hunches about why health inequality exists and is so persistent in the face of rising living standards and improving general levels of health. These hunches centre around the problems involved in the negotiation of personal identity (Bunton and Burrows 1995; Langman 1998; Howard 2000). The prospect of young people deserting apparently safe homes in western nations to risk their lives fighting to establish a caliphate in the Middle East is an extreme example of how the struggle for identity can tragically shorten life.

Identities are shifting and have highly variable sources, making generalization hazardous (Giddens 1991). But it is possible that the ‘psycho-social’ and the ‘behavioural’ processes thought to be involved in health inequality, and even the importance of money, have their roots to some extent in problems of identity (Elstad 2010). Where an individual’s sense of identity is assured by the stable occupation of ‘central life roles’ (Siegrist 2000a), such as worker, partner or parent, and/or by acceptance in a reasonably stable community, the cost of identity management is lower. But these sources of stability may be accompanied by extensive constraints on factors such as occupational choice or sexual behaviour or orientation. Where there is no stable community of acceptance, identity may depend more on outward symbolic display, which needs to be constantly repeated and updated. Individuals gain little sense of inherent worth just by ‘being themselves’. Large cities are places where the shackles of traditional role performance may be gladly shaken off but where making and retaining alternative sets of social ties may present other problems. Such circumstances may increase the importance of high numbers of material goods, and goods of the culturally symbolic kind. Living in a fragmented society tends to be expensive. The advent of social media may well have exacerbated these problems. In fact, one could surmise that one reason for the relentless rise of health inequality at a time of rising overall living standards and falling income inequality (Bartley 2012) is to be found in the increase in communication media. The years following the widespread introduction of television in homes in the 1950s were at the origin of the puzzle of rising health inequality.

Identity may be supported or threatened by any of the major forms of inequality. High prestige protects against adverse events of shorter duration. For example, living in a run-down student flat for a few years does not threaten someone from a secure middle-class family who knows they are sacrificing income at the present time in order to secure a satisfying, high-status long-term career. Security and autonomy at the workplace support a stable sense of self, even for people without high income or status. And income? As indicated above, this may act as a means to an end in terms of the prestige and security it can buy. In other cases, the purchase of consumer goods may act as a substitute for other sources of stable identity. The implication of this is that money for conspicuous consumption may be more important when autonomy, security or other sources of support for identity are not present.

As societies modernize, institutional constraints exerted by religion, family, caste or clan loosen, and freedom for many (in particular, for everyone who is not an adult male of the ‘majority’ or dominant cultural or ethnic group) increases. But under these conditions the maintenance of stable identity may become more problematic. In order to maintain social acceptance, the individual may need to devote more resources to identity-sustaining relationships. Identity maintenance for those in dominant positions has traditionally depended on being able to exert superior power and has been provided by women for everyone else. If traditional forms of dominance are breaking down and women’s time is more occupied with paid employment and the ‘public sphere’, this changes. Along with greater individual freedom comes the necessity for every individual, male and female, and, increasingly, older children as well as adults to do their own work in creating and maintaining the relationships that contribute to a sense of creditable identity. There is an ever-greater premium on the capacity for internal self-regulation and the management of one’s own identity and the creation of one’s own community.

Here again, a life-course approach offers useful sources of understanding. Those individuals whose experiences in their family of origin have given them an inner sense of security and the ability to form and maintain relationships may revel in the freedom of a less constrained community, even if social ties are looser. In Beck’s words:

The ability to choose and maintain one’s own social relations is not an ability everyone has by nature. It is . . . a learned ability which depends on special social and family backgrounds. The reflexive conduct of life, the planning of one’s own biography and social relations, gives rise to a new inequality: the inequality of dealing with insecurity and reflexivity. (Beck 1992: 98)

Those with fewer inner resources may adopt images from the mass media as a substitute source of identity. For these individuals, periods with relatively little money can be catastrophic for the sense of self. If self-esteem depends on possession of symbolic goods, these will be given priority over, for example, nutrition and warmth. And the cost of imitating cultural, mass-media images is what will determine in part how much health will suffer during financially insecure periods.

One way of understanding the dynamic that produces health inequality may be the vital importance of social participation to identity and the effect of social inequality on the costs of social participation. But wider social inequality has another effect. This is the influence of the experience of inequality on the consumption of mood-altering substances. This means not only drugs such as heroin and cocaine, or even alcohol. In the early 1980s, Cameron and Jones termed alcohol and tobacco ‘drugs of solace’ (Cameron and Jones 1985). They argued that, while doctors and health educators see alcohol and tobacco as a ‘problem’, for those who use them they are in fact a solution. To these, I would add a wide range of ‘comfort foods’, all of which are consumed to dull the experiences of uncertainty and isolation. The importance of the struggle for a stable identity in the causation of health inequality is shown by the age-patterned nature of many of these influences. Very large numbers of adolescents become obsessed with expensive consumer items, and take up smoking and the use of recreational drugs. It is a period when the struggle for identity is at its most acute and the outcome is uncertain. As they grow older, adolescents from the more socially (and perhaps emotionally) favoured backgrounds tend to moderate their attachment to these (Giesinger et al. 2014) − they find other sources of identity in work and in stable adult relationships. Difficulty in establishing adult identity can have a number of different sources: disturbed parental relationships and an inability to find work which is sufficiently well paid to make independent life possible are two of the most common. For those with very disturbed emotional backgrounds, even great riches may not be sufficient to feed an unsuccessful quest for identity, as is frequently seen in media ‘stars’ and in members of rich and aristocratic families who engage in self-destructive forms of behaviour.

In what follows, we shall see whether existing research on health inequality can be understood with the help of a model of accumulated biological, psychological and social advantages and disadvantages within the contexts of different national and local economies and cultures. There are numerous potential combinations of circumstances that individuals may pass through in their life course, all of which may contribute to healthy life expectancy many years into the future. The challenge for research is to arrive at adequate measures of these, and adequate methods for putting them together in causal models that are useful for policy discussion.