Michael G. Marmot and Robert Sapolsky
Position in the social hierarchy is closely related to health and risk of disease. The result is a social gradient in health—worse health the lower the social position. The question is why: How does social position affect biological pathways to cause disease? Destitution is bad for health. Inadequate nutrition, lack of sewerage and clean water, and conditions that foster the spread of microorganisms all contribute to illness. The effects on disease of such material deprivation are exacerbated by lack of appropriate preventive and curative medical care. Yet the social gradient in health also exists where everybody is above this level of absolute destitution. The differences are large. In the Scottish city of Glasgow, for example, there is a 28-year difference in life expectancy between the most and least salubrious areas (Hanlon et al., 2006). Similarly, among the poorer neighborhoods of Washington, DC, life expectancy is 16 years shorter than in the wealthier suburbs (Murray et al., 2006).
We propose that psychosocial processes linked to social experience constitute a major reason for the health gradient. An important part of the evidence for this proposition comes from nonhuman primates. There have been polarized views of the relevance of research on animals to understanding humans. Alexander Pope wrote “that the proper study of mankind is man.” By contrast, Darwin said: “He who understands baboons would do more toward human metaphysics than Locke.” One of us has spent his research life studying men and women; the other, baboons. Yet we are more with Darwin than with Pope. This is not to imply that there is a simple read-across from monkeys or apes to man. Humans are not simply baboons
in clothes or killer apes with sophistication, and the differences, as will be discussed, can be highly instructive.
The crucial issue is the insight to be gained by studying variation, both in different human societies and among different primate species. There is a naïve idea that as humans spent 99 percent of their existence as hunter-gatherers on the savannah, the species’ true nature can be understood by studying savannah primates such as baboons. This is quite profoundly wrong. It was not time spent as savannah baboons, but as savannah hominids. There are marked differences among primate species, including Homo sapiens. The study of this variation is illuminating in the way Darwin presumably had in mind. We contend that by studying when and how hierarchies are related to health in other primate species, we understand better how psychosocial factors generate the social gradient in health in humans.
In this review, we consider the ways in which studies of rank and health in nonhuman primates aid in understanding of the health gradient in humans. Specifically, we will (a) review the ways in which the health gradient is grounded in psychosocial factors; (b) consider how this gradient is anything but simple, and the numerous modifiers of it can only be appreciated in the context of psychosocial factors; (c) examine how some similar health gradients exist in nonhuman primates that, importantly, lack the lifestyle risk factors of humans (e.g., smoking, differential health care access); and (d) explore the enormous variability in the primate realm of social status/health relationships. We contend that circumstances under which subordinate animals suffer health disadvantages have their equivalents among humans. The crucial understanding from the nonhuman primate studies is of the links between social circumstances and biology.
VARIED HEALTH GRADIENTS
We seek to understand the causes of the human health gradient. In order to do so, it is helpful to ask two initial questions of the human evidence: Is the health gradient everywhere; if so, even if why it exists can be understood, is it not inevitable?
Taking life expectancy, or all-cause mortality, as the measure of health, the gradient appears to be widespread (Marmot, 2005), being observed in countries rich, poor, and intermediate (Victora et al., 2003; Hurt et al., 2004). A measure such as education can be used with some universality across societies. It is a general finding that the higher the education, the longer the life, although the magnitude of the difference varies, as illustrated in Figure 16-1. Where the data are available, similar gradients are seen with socioeconomic classifications based on occupation, income, parents’ social class, or degree of affluence of area of residence (Marmot, 2004). While there is an undoubted effect of poor health in determining
FIGURE 16-1 Life expectancy) at age 25 by education.
NOTES: ISCED = International Standard Classification of Education; Level 0-2 = pre-primary, primary and lower secondary education; Level 5-6 = tertiary education. Countries: EE = Estonia, HU = Hungary, RO = Romania, BG = Bulgaria, PL = Poland, HR = Croatia, CZ = Czech Republic, SI = Slovenia, PT = Portugal, FI = Finland, DK = Denmark, MT = Malta, NO = Norway, IT = Italy, and SE = Sweden.
SOURCE: Marmot (2013).
socioeconomic position, a wealth of evidence points to the effects of social circumstances linked to status having a powerful influence on health (Smith, 1999; Marmot, 2004).
When we turn attention to specific conditions, this universality of the social gradient does not apply. For example, among women in rich countries, obesity follows the social gradient— low status predicts a high prevalence of obesity. However, the gradient is reversed in low-income countries (Gross National Product under US $745): The more education women have, the higher the prevalence of obesity (Monteiro et al., 2004). It is not a surprise, therefore, to find that Type II diabetes and cardiovascular disease are more common in women of high status in poor countries (Chang et al., 2002). This reversal of the gradient for obesity may result from the fact that in poor countries, the challenge for the poor is to get enough food; in wealthier countries, the challenge for the poor is to get enough healthy food. There may be other influences at play that will be dealt with under “Relevant Variables” below. Yet for the diseases responsible for the bulk of loss of life in such poor countries, there is the familiar social gradient. It is worse to be at the bottom than at the top (Victora et al., 2003).
In richer countries, a social gradient is found for most causes of death, including cardiovascular diseases, several cancers, respiratory and gastro-
intestinal diseases, and accidental and violent deaths. There are exceptions, notable among which have been breast cancer, malignant melanoma, and leukemia (Drever and Whitehead, 1997). Among the specific causes of death that do follow the social gradient, the majority, the steepness of the slope varies: that is, the strength of association between social position and disease can be greater or less. Coronary heart disease (CHD) in richer countries tends to show a fairly large social gradient. In countries, such as those in southern Europe where CHD is a less important contributor to overall mortality than in northern Europe, the social gradient in all-cause mortality is less (Marmot, 2013).
This variability in the social gradient leads in to the answer to our second initial question: Is some version of the social gradient in health inevitable? Variation in the magnitude of the gradient suggests that not taking the health gradient as a given. There is not a simple one-to-one relation between rank in the social hierarchy and health. The relation is contingent on what hierarchy means in a given society. In a low-income country, low place in the hierarchy may mean insufficient dietary calories and high burden of infection, and hence low obesity and low cardiovascular disease, but high mortality in infancy and childhood, high rate of violent deaths, and high rates of tuberculosis.
In richer countries, where “diseases of poverty” are not the major cause of death and the major burden of disease is noncommunicable disease and violent deaths, most people are not suffering from the ravages of destitution. In such circumstances, it is our contention that psychosocial factors play a major role in generating the social gradient in health. But this, too, does not imply inevitability of the social gradient. The meaning of hierarchies varies across societies.
The health implications of this variation are the focus of this paper. For this focus, there will be a heavy emphasis on the fact that there is a relationship between health and social dominance rank in other primate species. Initially, it will appear as if the relationship is straightforward, with low dominance rank associated with poor health. However, as will be reviewed, there is great variability in social structure among the numerous primate species and rank/health relationships vary accordingly. Moreover, recent work has demonstrated that differences in social milieu in different populations of the same primate species produce differences in the rank/health relationships.
Of course, one way the health effect of hierarchies can vary is through different links between position on the hierarchy and health behaviors such as smoking, diet, alcohol and physical activity. These all are important, but results from the two Whitehall studies of British civil servants suggest that these behaviors, and blood pressure and plasma cholesterol levels, account for less than a third of the social gradient in cardiovascular disease
(Marmot et al., 1997; van Rossum et al., 2000). Similarly, while differences in medical care loom large as potential explanations for the link between low status and health in the United States, or in low-income countries with poorly developed health systems, they are less likely as explanations in countries with universal provision of health care such as Finland, Sweden, and the United Kingdom. Instead, both the human and nonhuman studies to be reviewed emphasize the importance of psychosocial factors and the extent to which low rank can be a surrogate for large degrees of psychosocial stress. Differential access to high-quality medical care can potentiate health inequalities, but lack of medical care is not the cause of inequalities in the occurrence of disease.
VARIED HEALTH GRADIENTS IN VARIED PRIMATE SOCIETIES
Baboons, Hierarchies, and Health
A frequent feature of social primates (as well as of other social mammals) is the existence of dominance hierarchies. Regardless of the richness of an ecosystem, resources are finite and are often divided unevenly. Dominance hierarchies formalize and ritualize such inequalities, obviating the need for overt competition for each contested resource. In many primate species, linear hierarchies occur (i.e., A > B > C > D …, rather than circularities, where A > B > C > A), with rankings that remain stable ranging from months to the lifetime of the individual, and where rank shapes quality of life in numerous ways.
Researchers have demonstrated that a primate’s dominance rank is related to its state of health. Much as with the human social gradient literature, it is rank that predicts physiological differences, rather than the reverse. Probably the most detailed study of this issue has been of a population of wild savannah baboons (Papio anubis) living in the Serengeti ecosystem of East Africa. This study is notable for its duration (approximately 30 years of longitudinal data), the range of health-related physiological endpoints examined, and the relatively rare reliance on a wild, rather than captive primate population.
Savannah baboons live in large stable troops of 50–150 individuals. Male baboons, which have been the primary subjects of this study, leave their natal troop at puberty, typically then spending their lives in 1-2 other troops. Males form dominance hierarchies that are fairly stable over time, with individuals typically rising in the hierarchy as they approach early adulthood, and declining beginning in late adulthood. Dominance rank is predictive of access to desirable food items (e.g., kills), preferred resting sites, social grooming, and mating opportunities. Importantly, when the hierarchy is stable, dominant males have considerable control and predict-
ability in their lives, as well as ample outlets for frustration such as social grooming, or displacing aggression onto a lower-ranking individual. In contrast, under such circumstances, social subordination is marked by a disproportionate share of stressors. Typically, attaining high rank involves escalated aggression; savannah baboons have among the highest rates of aggression of any primate and an individual typically rises in rank (and, in particular, achieves alpha status) by decisively winning a key fight. By contrast, maintaining high rank is more dependant upon social intelligence, skill at forming of coalitions, psychological intimidation, and impulse control (i.e., ignoring some provocations). As will be described below in the section “Personality,” males that are particularly successful in this realm tend towards certainly personality styles associated with low basal cortisol levels.
This picture of sustained social stress for subordinates has pathophysiological correlates. In a dominance hierarchy in a typical savannah baboon troop, subordinate males demonstrate, relative to dominant males: (a) high basal levels of cortisol and resting blood pressure, sluggish endocrine and cardiovascular stress-responses, and sluggish recovery of cortisol levels and blood pressure following the end of a stressor; (b) a testicular axis more easily suppressed by stress; (c) suppressed levels of HDL cholesterol; (d) fewer circulating lymphocytes; (e) lower levels of insulin-like growth factor-1; and (f) greater activation of endogenous benzodiazapine signaling (Sapolsky, 1993a; Sapolsky and Share, 1994, 2004; Sapolsky and Spencer, 1997).
A key question, of course, is whether the physiological profile of subordinate animals is sufficiently adverse to actually impact health and lifespan. This is extremely difficult to answer for at least two reasons. First, because rank among male baboons shifts over time, a high-ranking male may have a very different rank a few years earlier or later; thus, the lifelong impact of rank must integrate the shifts in status. Second, because of the completely wild nature of this population, it is rarely possible to determine the cause of death of an individual or even, given the migratory nature of male baboons, whether an individual has actually died or merely emigrated to a different troop. Nonetheless, these rank differences may be meaningful. For example, wounds heal more slowly in subordinate male baboons in these feral populations (Archie et al., 2012), and the basal hypercortisolism of subordinate males is in the range known to adversely impact blood pressure, insulin sensitivity, and immune function (Sapolsky et al., 2000).
While poor health can certainly lead to low social rank, the longitudinal data in these studies demonstrate that the pathophysiological correlates of subordination follow, rather than precede, the establishment of a rank. We argue that, as with the human health gradient, this rank/health link is mostly psychosocial in nature. For one thing, the bulk of the stressors dis-
proportionately experienced by subordinate males are psychosocial, rather than physical. Food is plentiful and easily obtained through foraging, so that, in the absence of a drought, subordination does not come with a nutritional cost. Differential access to food is limited to “luxuries,” particularly kills (which contribute an extremely small percentage of calories for even a dominant male baboon). Furthermore, subordinate animals are predated at the same low rates as are dominant animals. In addition, despite the picture of baboons as being highly aggressive, the vast majority of dominance interactions involve not overt aggression but only threats of aggression or, even more often, psychological intimidation. Moreover, the physiological correlates of rank track more closely with the psychosocial stressors of the rank than with the physical stressors. This is shown in an examination of reversals of the direction of dominance in an interaction among these males, for example, when Male #5 in the hierarchy, typically dominated by Male #4, instead wins a dominance interaction with Male #4. A high rate of such dominance reversals between Males #4 and #5 indicates that the two may soon be switching ranks. In this study, increasing rates of dominance reversals with a male one step below the subject in the hierarchy (i.e., an increasing likelihood of a demotion) were associated with higher basal cortisol levels, whereas increasing rates of dominance reversals with a male one step above the subject (i.e., an increasing likelihood of a promotion) were not. This difference occurred despite the two patterns of dominance interactions (i.e., with the male above, or the male below) involving the same rates of escalated aggression, likelihood of injury, disruption of feeding, and so on (Sapolsky, 1992).
Thus, the pathophysiological price of being a low-ranking male baboon is strikingly similar to that of a low-socioeconomic status (SES) human (Steptoe and Marmot, 1992). This suggests an easy conclusion, namely that these findings regarding baboons are sufficient for extrapolation from the rank/health literature in primates to the social gradient in humans. This would be quite incorrect, and for two important reasons. First, this rank/health profile in baboons occurs only in certain circumstances, and the exceptions are both logical and informative. Second, savannah baboons are not some sort of generic representatives of “social primates” as a whole. Instead, there are more than 150 different species of social primates, each with a distinctive social system. Rank/health relations have been studied in a number of them, and many of the correlates uncovered are dramatically different from those seen in savannah baboons (Abbott et al., 2003). Once again, such variability is logical and potentially quite illuminating for making sense of the health gradient in humans. Thus, while the savannah baboon studies are, arguably, the most detailed in the primate rank/health literature, these other studies are vital and will be given equal weight in the subsequent sections.
In the next section, we will consider some of the key factors that predict the nature of rank/health correlates in different primate species and populations, and how the rules of such variability give insights into the health gradient in humans. Throughout the review of this literature, a particular strength of the primate studies will be apparent. As stated above, we drew the conclusion that neither risk factors associated with lifestyle nor differences in medical care were the major causes of the health gradient in humans. Critically, the great strength of the studies of nonhuman primates is that issues of health care access and lifestyle risk factors (e.g., smoking) can be entirely ruled out. In addition, individuals within the same primate social group will typically have virtually identical levels of activity, and rank-related dietary differences are typically trivial. For example, while access to meat is almost exclusive to high-ranking male baboons, it still constitutes less than 1 percent of their diet.
In what follows, it will be noted that much of the literature on the health gradient in humans has disease, mortality, or life expectancy as the object of study. It is supplemented by smaller scale studies of physiological reactions to stress. The literature on nonhuman primates, on smaller numbers of animals than the human studies, more typically has not disease as an “outcome” but physiological indicators that plausibly indicate risk of disease. Moreover, it should be noted that this nonhuman primate literature has overwhelmingly focused on measurement of cortisol as an endpoint (despite the possibility of measuring additional hormones in the blood, urine, or fecal sample that has been obtained). If one were to choose a single physiological endpoint to serve as a surrogate for the effects of stress on health, measurement of cortisol seems a reasonable choice, given the pathogenic effects of cortisol excess (including inducing insulin resistance, hypertension, immunosuppression, and reproductive impairments). We shall use the term rank/health or social gradient in health in both cases. For shorthand, we shall use the term “primates” to refer to nonhuman primates.
The Relevant Variables
As outlined, hierarchical systems in primates often take the form of ranks potentially shifting over time. Typically, this involves fairly local changes (e.g., Ranks #3 and #4 switch positions), while the overall structure remains the same. Occasionally, instead, there can be periods of dramatic shifts in the hierarchy, with changes up and down the ranks. Commonly this arises due to a major demographic event, such as the death of a key individual. During such a rare time, it is unlikely that, for example, a very
high-ranking individual will switch positions with a very low-ranking one; instead, as before, the changes are quite local. What characterizes the instability, however, is the sheer number and transience of such local changes—e.g., shifts in rankings daily, rather than over, perhaps, a season—and the amounts of aggression and social tension that accompany the instability.
In many Old World primate species (such as savannah baboons), being high-ranking in a stable hierarchy carries with it tremendous psychological advantages. As outlined above, this includes considerable amounts of social control and predictability, and the low basal cortisol levels and low incidence of atherosclerosis typical of dominant individuals at such times are commensurate with these low levels of psychological stress (Sapolsky, 2005).
During periods of hierarchical instability, however, the picture changes markedly, with the bulk of the social tensions, unpredictability, and aggression centered on the higher end of the hierarchy (Sapolsky, 2005). Not surprisingly, then, during times of such instability, it is dominant individuals that display the highest basal cortisol levels along with the highest incidence of atherosclerosis. This dichotomy between stable and unstable hierarchies has been demonstrated in one wild primate population, namely the baboon studies, and in numerous captive ones. The latter cases are particularly striking, in that they are derived from the situation where a social group of captive animals is first formed; the immediate period afterward models an instability in the wild, in that the animals spend an intensely competitive and aggressive period forming a first hierarchy. Over the course of months, stable ranking relationships emerge. Studies of primates of both sexes demonstrate that during the first period of unstable relations, the most severe stress-related pathologies are associated with dominant animals; it is only when ranks stabilize, typically over the course of a year, that the situation reverses (Sapolsky, 1993b, 2005). Importantly, the stressfulness of periods of rank instability may not so much reflect the frequency with which ranks are changing as much as the frequency with which the status quo is being challenged (regardless of whether the challenge results in a shift in rank) (Gesquiere et al., 2011; Sapolsky, 2011).
Is this stable/unstable dichotomy relevant to understanding the health gradient in humans? The greater degree of control and predictability in the lives of high-ranking animals among Old World primates has its counterpart in human societies. In theory, threats to this control and predictability would weaken the health advantage of high status individuals. One such threat is unemployment. Unemployment, like most uncontrollable threats, is more common in low-status individuals than in high. There is evidence from the United Kingdom, however, that when high-status individuals do become unemployed, their mortality rate rises to a higher level than that of the men immediately below them in the hierarchy who do not experience
unemployment (Moser et al., 1987). As in the primates, the high-status individuals are not protected from the health consequences of threats such as unemployment. Unlike in the primates, the turmoil of unemployment does not change the health gradient. In the end, it is the high-status individuals who have more options.
In the former Soviet Union and countries of Central and Eastern Europe, whole societies were in turmoil as communism tottered and finally collapsed, and before a new social order took hold. Mortality rose. In Russia there was evidence that the rise in mortality was more rapid in areas that were marked by social turmoil, as measured by labor market turnover, unemployment, marriage, and divorce rates (Bobak et al., 1998b; Cornia and Paniccia, 2000). As the new societies have formed, the social gradient in mortality increased. In the Czech Republic, a case control study of myocardial infarction in the 1990s showed a clear social gradient, with men with low education having higher risk. Much of this social gradient could be explained by lack of control in the workplace (Bobak et al., 1998a). The importance of lack of control is a direct parallel with the primate literature.
We noted earlier that obesity shows the familiar inverse association with social status in rich countries but high-status people tend more to obesity in poor countries. Clearly obesity is related to calorie imbalance, but why should calorie imbalance show these patterns? One type of explanation is knowledge and opportunities for healthy behavior. In rich countries, high-status people are more likely to have the knowledge, and financial and other resources, to indulge in healthy eating. In poorer countries, being fat may still be a sign of health and the wealth to engage in conspicuous consumption.
A second possibility may involve stress pathways. The evidence points strongly to greater uncertainty for people of low-SES position in rich countries: threats of unemployment and job insecurity, economic shocks due to illness, and other stressful life events. As evidence that stress pathways may be involved in obesity is the finding of a clear inverse gradient in central adiposity—high waist-hip ratio (Brunner et al., 1993)—that has been linked to hypercortisolism (Brunner et al., 2002). It is difficult to imagine a life more stressful than that of a low-income person in a low-income country, but with insufficient calories to eat, such stress is unlikely to result in obesity.
How position in a social hierarchy is attained and how it is maintained are two very different issues. Depending on the species and sex, there are a variety of mechanisms by which animals achieve positions of dominance varying from intense aggressive interactions to inheritance. It has been
observed that low-ranking male baboons that initiate a lot of fights with dominant individuals, and are thereby actively attempting to rise in the hierarchy, have higher basal cortisol levels than rank-matched subordinate individuals that do not initiate many fights (Virgin and Sapolsky, 1997). There is, however, little evidence that the mode by which dominance is achieved affects the rank-health relationship.
In humans there is evidence that rank, however achieved, is related to health. People whose high status is conferred by coming from privileged backgrounds have better health, in general, than those from more humble origins (Davey Smith, 2003). Independent of childhood social class, those in higher status positions in adulthood have better health than those in lower (Marmot et al., 2001).
Among primate species where rank can shift over time, a key question is what is involved for a dominant individual to maintain that high rank. The strategies utilized can be broadly dichotomized. At one extreme, dominance must be regularly reasserted, and in overt ways, for example, frequent aggressive attacks upon subordinates, regardless of whether being explicitly challenged. In such situations, as is seen among the ring-tailed lemurs of Madagascar, it is dominant individuals that have the highest rates of aggression and of dominance interactions. Such high-ranking animals, by definition, win the vast majority of such interactions, that is, they remain dominant. Despite that, in such cases, it is the high-ranking individuals that show the greatest physiological indices of stress (Cavigelli, 1999).
In contrast, at the other extreme are dominance systems in which dominant individuals maintain their status with a minimum of aggression or overt physical subordination of other individuals. In these cases, maintenance is accomplished instead through psychological intimidation, such as repeatedly standing close enough to a subordinate and making direct eye contact so as to disrupt whatever the low-ranking individual was doing and forcing the animal to move away. In such settings, high-ranking individuals have among the lowest rates of overt aggression. In primate social systems where dominant individuals reassert their dominance with minimal overt effort, it is subordinate animals with the greatest physiological indices of stress (Abbott et al., 2003). High-ranking males do not have particularly high testosterone levels, and high rates of aggression among such individuals are often a sign that their position is teetering.
A fine-tuned analysis of rank/physiology correlates in baboons makes this point about rank maintenance with particular subtlety (Gesquiere et al., 2011). This extensive multiyear study of multiple troops of a different wild population of baboons replicated the general finding of elevated basal cortisol levels in subordinate males. The lowest levels were observed, however, in the second-ranking males (in hierarchies consisting of approximately 15 adults). Alpha males, in contrast, had cortisol levels that were elevated
in the range typical of the lowest-ranking cohort; the authors speculated that the elevated levels reflected the metabolic costs of the frequent fighting on the parts of alpha males. Thus, being generically high-ranking (e.g., in the upper-ranking quartile) and being the alpha individual can be qualitatively quite different.
The relevance to human society is clear. In a society where men fought each other physically for status, one would expect a physiological toll on high-status individuals. But that is not how complex stratified societies work. High-ranking individuals maintain their rank, in general, without resort to violence; instead, when they use violence for their ends, they send lower-status individuals to do it on their behalf. Their dominance is maintained, therefore, without the physiological costs that overt physical aggression would entail.
In human societies, the best-known examples of using violence to maintain status are in lower-income urban settings. Daly and Wilson show a link between homicide rates in Chicago and the degree of income inequalities. Their interpretation of this finding is that threats to status under the shadow of income inequality are met with violence (Daly and Wilson, 1988; Wilson and Daly, 1997). There is the obvious health effect of high mortality from violent death in these communities. Such aggressive behavior is also likely to have physiological consequences (Cohen and Nisbett, 1996).
The Vividness and Frequency with Which Subordination Is Experienced
Under certain circumstances, there is a dose-response relationship between the intensity of a physical stressor and the magnitude of the resultant stress-response (for example, between the extent of blood loss in experimental hemorrhages and the extent of increased cortisol secretion [Gann, 1969]). Similarly, there can be a correlation between the intensity of a psychological stressor and the magnitude of the stress-response; for example, between increasing gradations of novelty in an environment and increased corticosterone secretion in a rat (note that corticosterone is the rat equivalent of cortisol) (Levine et al., 1989). This approximation of a dose-response relationship also appears in the rank/health literature. For example, the more often female baboons were harassed by a highly aggressive male who had recently transferred into the troop, the greater the extent of immune suppression (Alberts et al., 1992).
Thus, it is likely that the strength of the rank/health relationship can be modulated by the ease with which individuals can take refuge from their stressors. This is shown most clearly when considering animals of the same species either living ferally or in captivity. In the wild, particularly in ecosystems such as rain forests, a subordinate individual may be able to evade the presence and notice of a dominant individual. In contrast, within the
closer confines of captivity, this option is greatly restricted, potentially increasing the stressfulness of that rank. Commensurate with this idea, social subordination is associated with elevated basal cortisol levels in captive populations of wolves, but not in feral populations (Creel, 2001).
While novelty is stressful for a rat, there is no evidence to suggest anything similar in humans. In a work situation, for example, learning new things is part of human development and may well be associated with better health. What appears to be stressful is not stimulation or novelty, but psychological demands in the absence of control over work (Marmot, 2004).
In humans there has been much interest in Wilkinson’s findings that degree of income inequality is related to overall population health (Wilkinson and Pickett, 2010). Some take issue with these findings on empirical grounds, finding that inequality is correlated with other variables that could account for the association (Deaton and Lubotsky, 2003). Others interpret the findings as meaning that jealousy of those better-off is responsible for the worse health of those lower in status—and reject this interpretation as simplistic (Lynch et al., 2000).
It could be thought that if income inequality is important, then relative position in the hierarchy would be most important for health, rather than absolute socioeconomic level. In support of this position is the famous quote from Karl Marx: “A house may be large or small; as long as the surrounding houses are equally small, it satisfies all social demands for a dwelling. But if a palace rises beside the little house, the little house shrinks to a hovel.”
We do not interpret that as meaning that low relative position is, of itself, the crucial determinant of health. In our view, the human and nonhuman data are consistent: It is not place in the hierarchy per se but what place in the hierarchy means in a given situation. In humans, as the Marx quote implies and as economists from Adam Smith to Amartya Sen have made clear, it is not so much what individuals have that is important but what they can do with what they have (Marmot, 2004). Smith emphasized taking one’s place in public without shame (Smith, 1776/2003), Sen places emphasis on capabilities (Sen, 1992), and Fogel speaks of egalitarianism of spiritual resources (Fogel, 2000). All of this implies that relative position is important for health to the extent that people lower in the hierarchy are disadvantaged with respect to psychosocial factors. Control over life is one crucial factor, both at home and at work, as shown by, for example, path analyses demonstrating these as variables mediating rank/health relations (Bobak et al., 1997; Marmot et al., 1997; Chandola et al.. 2004; Horton. 2004). Another may be dignity linked to fair treatment (Horton, 2004; Kivimaki et al., 2004).
There has been much speculation that if perceived relative position were important for health, it might be worse for a disadvantaged person
to be living in an affluent neighborhood than in a uniformly poor one. While the data are not entirely consistent, studies in the United States and Britain nevertheless show that health of poorer people is even worse when they live in a poor neighborhood (Diez Roux et al., 2001; Stafford and Marmot, 2003).
Primate species differ as to how equally resources (e.g., food, safe resting places) are divided along the lines of rank. Among the South American primates that are cooperative breeders, the hierarchy is considered to be egalitarian, insofar as there is not a particularly steep gradient of resource acquisition as a function of rank. In contrast, Old World primates such as savannah baboons and rhesus macaques have “despotic” hierarchies, in which there is highly unequal resource acquisition. As noted, subordination is associated with hypercortisolism in despotic hierarchical species, but not in egalitarian ones. However, no studies have been able to demonstrate how much the specific issue of resource inequities contributes to this difference.
The issue of resources matters greatly when material resources mean the difference between starvation and adequate nutrition, or between unsanitary conditions and a salubrious environment. Improvement in resources for health provides an explanation for the great improvement in health in the 20th century in the developed countries. At a time when infant mortality in England and Wales, for example, is as low as 6/1000 live births in the bottom social group, compared with 250/1000 a century ago (Rowntree, 1901), resources matter in a different way.
Quoting Sen, we said that in a society where material needs for good health are met, what is important is not so much what one has but what one can do with what one has: capabilities. Adam Smith, father of modern economics, emphasized that the “necessaries” for life include whatever the customs of society render it indecent for “creditable” people to do without. In a society where family background mattered greatly “necessaries” may be less closely linked to income than in a society where prestige, status, and “necessaries” were more strongly linked to current income. In a study in post-communist Hungary, for example, we showed that reported ill-health was more strongly related to lack of luxury goods than it was to lack of goods that might be considered more basic for good health (Pikhart et al., 2003).
An extensive literature examining the building blocks of psychological stress shows that the extent to which psychosocial circumstances can
activate the stress-response reflects, broadly, two halves of an equation: the extent of psychosocial stress to which the individual is exposed, that is, frequency and severity of circumstances of loss of control and predictability, and the coping outlets available to the individual (Levine et al., 1989).
The social coping outlets available to psychosocially stressed primates can be positive or negative in nature. Positive outlets include the stressed animal seeking physical contact with an individual with which it is socially affiliated or initiating a bout of grooming with that individual. Among bonobo chimpanzees, a species distinct from the better-known common chimpanzee, sexual behavior is also used to decrease psychosocial stress (de Waal, 2005); in contrast, sexual behavior increases the risk of further psychosocial stress in most primate species. A small literature suggests that greater amounts of such positive outlets are associated with better health among primates. As one example, among savannah baboons, for the same social rank, basal cortisol levels and rates of positive affiliative behaviors are inversely correlated (with, it should be noted, no direct evidence of causality) (Sapolsky et al., 1997). As another example, in a meta-analysis encompassing all the primate species in which rank/cortisol studies have been carried out, the availability of such positive coping outlets to subordinate individuals was significantly, albeit mildly, predictive of their basal cortisol levels (Abbott et al., 2003).
Findings such as this bring up the key issue of sex differences. There is tremendous variation among primate species as to differences, or lack thereof, between the sexes in social systems. Of most relevance here are the Old World primate species (e.g., savannah baboons and rhesus macaques), in which (a) males control a disproportionate share of resources and females are frequent targets of displacement aggression; and (b) females typically have numerous female relatives in their social group (in contrast to the situation for males). Among females of such species, the stressful consequences of male behavior can be buffered by the social affiliation afforded by other females, with protective effects coming preferentially, but not entirely from the females that are related. For example, the elevated basal cortisol levels reported among socially subordinate female macaques do not occur in a troop with atypically high levels of social affiliation (Gust et al., 1993). As another example, females most able to take advantage of close grooming relationships during periods of being subject to escalated male aggression had the smaller increases in cortisol levels (Engh et al., 2006). Very intriguing is one report that in such circumstances, the cortisol-lowering effects of grooming were a function of the frequency with which an individual groomed other animals, rather than the frequency with which she is groomed back (Shutt et al., 2007).
Thus, an array of positive coping outlets can be physiologically protective among primates and, in stable hierarchies, higher-ranking individuals
typically have far more opportunities for such outlets, particularly social grooming.
Negative social outlets also, in theory, can be protective, and this most often takes the form of displacement aggression (i.e., where an individual, having experienced a stressful psychosocial situation such as losing a dominance interaction, will then aggressively attack an innocent bystander). An ample literature with laboratory rodents shows that such displacement aggression can reduce the pathophysiological impact of an external stressor (Levine et al., 1989). Similarly, in one relevant study, socially subordinate male savannah baboons with a strong tendency to displace aggression (on the few animals to which they could do it) had lower basal cortisol levels than rank matched subordinate males without that tendency (Virgin and Sapolsky, 1997).
Thus, among primates, both positive and negative coping outlets can blunt the health consequences of a stressful social rank. No studies to date have examined the critical issue as to whether the positive or negative outlets are more powerful in modulating the rank/health relationship.
There is a rich literature in humans on the protective effects of social interaction and affiliative behavior. The evidence is strong, particularly at the individual level: Individuals who have multiple social ties have lower mortality rates and better health than those who have fewer social ties (Berkman and Glass, 2000). Social isolation appears to be harmful. The literature is mixed on the degree to which there are main effects—social networks simply associated with lower mortality—or buffering—social networks protective in the context of high levels of other sources of stress (Cohen et al., 2000). Social ties are relevant to the social gradient. There is ample evidence that, with the exception of links with family, most other social ties are less frequent as the social hierarchy is descended (Marmot, 2004).
It is reasonable to speculate that human primates may have evolved to be sociable. Across the ~150 primate species, a significant predictor of the relative size of the cortex is the size of the social group: that is, big cortexes and big social groups go together, with orangs as a striking exception in that they are solitary. In other words, on a certain level, the primate cortex evolved for the purpose of gossip and politics.
Much has been made of the importance of reciprocal altruism in the evolution of social systems in many species, including a great deal of work related to the critical issue of how stable reciprocity is first established. It is plausible that altruism evolved as an adaptive strategy: An animal will expend valuable energy to help another as there is likely to be reward coming the other way in the future. In humans an important part of living in society is indulging in socially generative behavior that will be rewarded (Ridley, 1996). An individual expends effort on behalf of others, and there is an expectation of reward for effort expended. A situation where effort is
not matched by appropriate reward is likely to be stressful. Evidence for this comes from the workplace. People whose work is characterized by imbalance between effort expended and reward received are at increased risk of coronary heart disease (Siegrist and Marmot, 2004), mental illness (Stansfeld et al., 1999), and decrements in functioning (Stansfeld et al., 1998).
These results on effort–reward imbalance are consistent with the insights from the primate studies on the balance between stressors and coping mechanisms. However, humans have many more and more subtle means of coping than are available to other primates. In addition to social affiliative behavior, there is the question of religiosity. An extensive literature suggests considerable health benefits to religiosity, but this conclusion has been confounded by the fact that some forms of religious observance can involve different health behaviors (such as in Mormons and Seventh Day Adventists), and that virtually all forms provide community, independent of religious content (McCullough et al., 2000). However, the interaction between the extent and type of religiosity as a coping mechanism, and the social gradient, remains understudied.
In humans there is much speculation and some evidence that affiliative behavior may also operate at the group level in the form of social capital or collective social efficacy. This is taken up below in the section “Social milieu and Culture.”
The magnitude of the stress-response is not only modulated by the intensity and nature of the stressors, psychosocial or otherwise, and the availability of coping outlets, but also by personality factors that influence whether and how those stressors and coping outlets are perceived. Thus, an anxiety-prone personality can be thought as including a tendency to overestimate the menace intrinsic in a particular stressor; a depression-prone personality can involve a failure to perceive the efficacy of a coping outlet, i.e., a sense of helplessness. The term “personality” is used in a similar way by primatologists and formalizes the fact that individual primates have strong and stable differences in temperament and social reactivity. The extent to which such personality arises from genetic versus environmental factors is as unclear as in humans.
Some of these personality factors can modulate rank/health relationships. For example, in social systems where dominance is associated with better indices of health, this advantage is blunted in high-ranking males that are particularly reactive to novelty, that is, have their ongoing behaviors disrupted, and are least adept at behaviorally distinguishing between neutral and threatening interactions with rivals (Sapolsky and Ray, 1989). Conversely, in social systems where subordination is associated with less
favorable indices of health, low-ranking males who are particularly adept at taking advantage of social coping outlets were spared the hypercortisolism typical of their rank (Virgin and Sapolsky, 1997).
If we generalize from personality to the enduring effects of early life on psychological patterns in humans, there is ample evidence that circumstances in adult life, related to rank, affect individuals differently. Two examples suffice. The first illustrates that stressors may be unequally distributed in the population depending on an individual’s prior characteristics. The second illustrates that the effect of the stressor on health may vary according to the individual’s prior characteristics.
The first can be well illustrated by unemployment. Unemployment rarely strikes at random. A study in Britain has followed a group of people from their birth in 1958 to the present, gathering data as they went through each stage of the lifecourse. Unemployment is more common in people who have experienced unfavorable family circumstances and have less educational attainment (Bartley et al., 1999). Of course, if the population unemployment levels are low, such individuals would not spend as much time out of work. There is, therefore, an interaction between susceptibility of the individual and what is happening in the wider environment.
The second example concerns the genetics of neuropsychiatric disorders. Specifically, it concerns the finding in numerous, but not all studies, of a gene-environment interaction between a particular variant of the serotonin transporter gene and early life stress in increasing the risk of depression (cf., Caspi et al., 2003).
Social Milieu and Culture
As emphasized throughout, there is no single generic primate species but, instead, phylogenetic variability as to social system across the primates. What has been slower to be recognized is that within a single species, there is considerable variation in social behavior and atmosphere from one group to the next. In some cases, the source of the variability is obvious and not particularly interesting; for example, it is to be expected that groups living in harsher environments, and thus forced to spend more time foraging, will spend less time socializing than groups in lusher settings. In other cases, the source of the differences could be genetic. However, considerable social variability emerges as a function of the vagaries of group demographics and personalities of group members. Thus, it is not surprising that the basal hypercortisolism typical of low-ranking female macaques is not observed in troops with lower rates of displacement aggression aimed at such subordinates (Gust et al., 1993).
In some cases, local populational differences in behavior can be transmitted multigenerationally. Such instances, when occurring independent of
ecological or genetic factors, meet the formal definition of “culture,” a term now widely used by primatologists. Most examples of culture in primates are rather narrow (e.g., a particular, regional style of social grooming or of tool use) (McGrew, 2004; Whiten, 2005). However, there have been cases of an entire social milieu being transmitted multigenerationally as a group culture.
The most-studied case involved a troop of savannah baboons where, because of a historical accident arising from proximity to a human settlement, the 50 percent of adult males that were most aggressive and least socially affiliative were all killed (Sapolsky and Share, 2004). This produced a troop social milieu markedly different from that of typical baboon troops, with low levels of displacement aggression and high levels of affiliative behaviors. As new (unrelated) adolescent males migrated into the troop from elsewhere, the novel social culture would be assimilated within about 6 months, and has been transmitted for more than a decade past its founding generation. Furthermore, females in this troop are more willing to chance a spontaneous affiliative gesture with a newly transferred male than is the case in other troops, a pattern that appears to be a driving force for the cultural assimilation of these new males. This female tendency, perhaps implying a greater level of trust in the general benevolence of the social environment, seems a rough approximation of the concept of “social capital” in humans.
Strikingly, this local culture had distinctive physiological correlates. Specifically, the hypercortisolism found among subordinate males in typical baboon troops is not observed in this one. Thus, even within the same primate species, the rank/health relationship will vary as a function of local social milieu.
An enduring question in the study of the social determinants of health is whether the environment in which an individual lives and works has an impact on health over and above the socioeconomic characteristics of the individual. As indicated earlier, the answer appears now fairly clearly to be yes (MacIntyre et al., 2002). In the United States, for example, both individual socioeconomic characteristics and level of deprivation of the area of residence influence risk of coronary heart disease (Diez Roux et al., 2001). In the Whitehall II study of British civil servants, there is evidence that deprivation of the environment of residence has a larger effect on the health of low-status people than of higher (Stafford and Marmot, 2003).
The protective influence of social support on health may have its counterpart at the level of populations. In other words, it is not only that individuals with a high degree of social connectedness have better health, but also societies that are characterized by higher degrees of social cohesion are healthier. In Chicago, neighborhoods with a high degree of collective social efficacy had lower homicide rates than less cohesive areas. It has been
suggested that social capital may provide a link between income inequality and health (Kawachi and Kennedy, 1997). Studies in England show a link between measures of social capital and health. The animal literature here provides little guidance.
The social gradient in health is a pervasive feature of health in populations. To take action to deal with it necessitates understanding of how it comes about. One strategy for doing this is large-scale studies of human populations. Indeed, in recent years, a large body of research has started to address this question. A different, complementary strategy is to study the health gradient in nonhuman primates.
We began this review with the proposition that humans are not simply apes with sophistication. Chimpanzees may be machiavellian but no ape was ever Machiavelli, or Shakespeare, or even the man on the Clapham omnibus. There is no simple read-across from nonhuman primates to people. To go to the other extreme and deny the lessons from studying primates would be perverse. We have put forward the case that variations across primate societies aid understanding of the links between status and health. We take it further. The circumstances under which subordinate animals suffer health disadvantage have their counterpart in human societies.
One clear advantage of studying nonhuman primates is their very nonhumanness. Many of the candidates put forward to explain health inequalities in humans simply are not seen in other species. Baboons don’t smoke, eat fast foods, or have differential access to health care depending on ability to pay.
A stressed primate, however, will have similar physiological responses to those of a stressed human. There is insight to be gained not only in understanding the biological pathways by which social position affects health, but also in understanding the circumstances under which these physiological responses are evoked. They lend credence to our claim that psychosocial factors play a major role in generating the social gradient in health.
Abbott, D., Keverne, E., Bercovitch, F., Shively, C., Mendoza, S., Saltzman, W., Snowdon, C., Ziegler, T., Banjevic, M., Garland, T., and Sapolsky, R. (2003). Are subordinates always stressed? A comparative analysis of rank differences in cortisol levels among primates. Hormones and Behavior, 43, 67-82.
Alberts, S., Sapolsky, R., and Altmann, J. (1992). Behavioral, endocrine, and immunological correlates of immigration by an aggressive male into a natural primate group. Hormones and Behavior, 26, 167-178.
Archie, E., Altmann, J., and Alberts, S. (2012). Social status predicts wound healing in wild baboons. Proceedings of the National Academy of Sciences of the United States of America, 109, 9017-9022.
Bartley, M., Ferrie, J., and Montgomery, S. (1999). Living in a high unemployment economy: Understanding the consequences. In M. Marmot and R. Wilkinson (Eds.), Social Determinants of Health (pp. 78-96). Oxford, UK: Oxford University Press.
Berkman, L., and Glass, T. (2000). Social integration, social networks, social support, and health. In L. Berkman and I. Kawachi (Eds.), Social Epidemiology (pp. 137-173). New York: Oxford University Press.
Bobak, M., Marmot, M., and Skodova, Z. (1997). Psychosocial factors at work and myocardial infarction in the Czech Republic. Acta Cardiology, LII, 136-137.
Bobak, M., Hertzman, C., Skodova, Z., and Marmot, M. (1998a). Association between psychosocial factors at work and non-fatal myocardial infarction in a population based case-control study in Czech men. Epidemiology, 9(1), 43-47.
Bobak, M., Pikhart, H., Hertzman, C., Rose, R., and Marmot, M. (1998b). Socioeconomic factors, perceived control and self-reported health in Russia. A cross-sectional survey. Social Science and Medicine, 47, 269-279.
Brunner, E., Nicholson, A., and Marmot, M. (1993). Trends in central obesity and insulin resistance across employment grades: The WII study. Journal of Epidemiology and Community Health, 47, 404-405.
Brunner, E., Hemingway, H., Walker, B., Page, M., Clarke, P., and Juneja, M. (2002). Adrenocortical, autonomic, and inflammatory causes of the metabolic syndrome. Circulation, 106, 2659-2665.
Caspi, A., Sugden, K., Moffitt, T., Taylor, A., Craig, I., and Harrington, H. (2003). Influence of life stress on depression: Moderation by a polymorphism in the 5-HTT gene. Science, 301, 386-389.
Cavigelli, S. (1999). Behavioural patterns associated with faecal cortisol levels in free-ranging female ring-tailed lemurs, Lemur catta. Animal Behaviour, 57, 935-944.
Chandola, T., Kuper, H., Singh-Manoux, A., Bartley, M., and Marmot, M. (2004). The effect of control at home on CHD events in the Whitehall II study: Gender differences in psychosocial domestic pathways to social inequalities in CHD. Social Science & Medicine, 58, 1501-1509.
Chang, C., Marmot, M., Farley, T., and Poulter, N. (2002). The influence of economic development on the association between education and the risk of acute myocardial infarction and stroke. Journal of Clinical Epidemiology, 8, 741-747.
Cohen, D., and Nisbett, R. (1996). Culture of Honor: The Psychology of Violence in the South. Boulder, CO: Westview Press.
Cohen, S., Gottlieb, B., and Underwood, L. (2000). Social relationships and health. In S. Cohen, L. Underwood, and B. Gottlieb (Eds.), Social Support Measurement and Intervention: A Guide for Health and Social Scientists (pp. 3-24). New York: Oxford University Press.
Cornia, G., and Paniccia, R. (2000). The transition mortality crisis: Evidence, interpretation and policy responses. In G. Cornia and R. Paniccia (Eds.), The Mortality Crisis in Transitional Economies (pp. 3-37). New York: Oxford University Press.
Creel, S. (2001). Social dominance and stress hormones. Trends in Ecology and Evolution, 16, 491-497.
Daly, M., and Wilson, M. (1988). Homicide. Hawthorne, NY: Aldine de Gruyter.
Davey Smith, G. (2003). Health Inequalities: Life Course Approaches. Bristol, UK: The Policy Press.
De Waal, F. (2005). Our Inner Ape. New York: Riverhead/Penguin Books.
Deaton, A., and Lubotsky, D. (2003). Mortality, inequality and race in American cities and states. Social Science & Medicine, 56, 1139-1153.
Diez Roux, A., Merkin, S., Arnett, D., Chambless, L., Massing, M., and Nieto, F. (2001). Neighborhood of residence and incidence of coronary heart disease. New England Journal of Medicine, 345, 99-106.
Drever, F., and Whitehead, M. (1997). Health Inequalities: Decennial Supplement. DS No. 15. London, UK: The Stationery Office, Office for National Statistics.
Engh, A., Beehner, J., Bergman, T., Whitten, P., Hoffmeier, R., and Seyfarth, R. (2006). Female hierarchy instability, male immigration, and infanticide increase glucocorticoid levels in female chacma baboons. Animal Behaviour, 71, 1227-1236.
Fogel, R. (2000). The Fourth Great Awakening and the Future of Egalitarianism. Chicago, IL: University of Chicago Press.
Gann, D. (1969). Parameters of the stimulus initiating the adrenocortical response to hemorrhage. Annals of the New York Academy of Sciences, 156, 740-748.
Gesquiere, L., Learn, N., Simao, M., Onyango, P., and Altmann, J. (2011). Life at the top: Rank and stress in wild male baboons. Science, 333, 357-360.
Gust, D., Gordon, T., Hambright, M., and Wilson, M. (1993). Relationship between social factors and pituitary-adrenocortical activity in female rhesus monkeys (Macaca mulatta). Hormones and Behavior, 27(3), 318-331.
Hanlon, P., Walsh, D., and Whyte, B. (2006). Let Glasgow Flourish. Glasgow, UK: Glasgow Centre for Population Health.
Horton, R. (2004). Rediscovering human dignity. The Lancet, 364, 1081-1085.
Hurt, L., Ronsmans, C., and Saha, S. (2004). Effects of education and other socioeconomic factors on middle age mortality in rural Bangladesh. Journal of Epidemiology and Community Health, 58, 315-320.
Kawachi, I., and Kennedy, B. (1997). Health and social cohesion: Why care about income inequality? British Medical Journal, 314, 1037-1040.
Kivimaki, M., Ferrie, J., Head, J., Shipley, M., Vahtera, J., and Marmot, M. (2004). Organisational justice and change in justice as predictors of employee health: The Whitehall II study. Journal of Epidemiology Community Health, 58, 931-937.
Levine, S., Coe, C., and Wiener, S. (1989). The psychoneuroendocrinology of stress—a psychobiological perspective. In S. Levine and R. Brush (Eds.), Psychoendocrinology (pp. 1-21). New York: Academic Press.
Lynch, J., Davey-Smith, G., Kaplan, G., and House J. (2000). Income inequality and mortality: Importance to health of individual income, psychosocial environment, or material conditions. British Medical Journal, 320, 1200-1204.
MacIntyre, S., Ellaway, A., and Cummins, S. (2002). Place effects on health: How can we conceptualize, operationalize and measure them? Social Science and Medicine, 55, 125-139.
Marmot, M. (2004). Status Syndrome. London, UK: Bloomsbury.
Marmot, M. (2005). Social determinants of health inequalities. The Lancet, 365, 1099-1094.
Marmot, M. (2013). Health Inequalities in the EU. Final Report of a Consortium. Consortium Lead: Sir Michael Marmot. Brussels: European Commission Directorate-General for Health and Consumers. Available: http://ec.europa.eu/health/social_determinants/docs/healthinequalitiesineu_2013_en.pdf [June 2014].
Marmot, M., Bosma, H., Hemingway, H., Brunner, E., and Stansfeld, S. (1997). Contribution of job control and other risk factors to social variations in coronary heart disease incidence. The Lancet, 350, 235-239.
Marmot, M., Shipley, M., Brunner, E., and Hemingway, H. (2001). Relative contribution of early life and adult socioeconomic factors to adult morbidity in the WII study. Journal of Epidemiology and Community Health, 55, 301-307.
McCullough, M., Hoyt, W., Larson, D., Koenig, H., and Thoresen, C. (2000). Religious involvement and mortality: A meta-analytic review. Health Psychology, 19, 211-222.
McGrew, W. (2004). The Cultured Chimpanzee: Reflections on Cultural Primatology. Cambridge, UK: Cambridge University Press.
Monteiro, C., Conde, W., Lu, B., and Popkin, B. (2004). Obesity and inequities in health in the developing world. International Journal of Obstetric and Related Metabolic Disorders, 28, 1181-1186.
Moser, K., Goldblatt, P., Fox, A., and Jones, D. (1987). Unemployment and mortality: Comparison of the 1971 and 1981 longitudinal study census samples. British Medical Journal, 294, 86-90.
Murray, C., Kulkarni, S., Michaud, C., Tomijima, N., Bulzacchelli, M., and Iandiorio, T. (2006). Eight Americas: Investigating mortality disparities across races, counties, and race-counties in the United States. Public Library of Science Medicine, 3, 1513-1525.
Pikhart, H., Bobak, M., Rose, R., and Marmot, M. (2003). Household item ownership and self-rated health: Material and psychosocial explanations. BMC Public Health, 3, 38-41.
Ridley, M. (1996). Origins of Virtue: Human Instincts and the Evolution of Cooperation. London, UK: Penguin Books.
Rowntree, B. (1901). Poverty: A study of town life. In G. Davey Smith, D. Dorling, and M. Shaw (Eds.), Poverty, Inequality and Health in Britain, 1800-2000: A Reader (pp. 97-106). Bristol, UK: The Policy Press.
Sapolsky, R. (1992). Cortisol concentrations and the social significance of rank instability among wild baboons. Psychoneuroendocrinology, 17, 701-709.
Sapolsky, R. (1993a). Endocrinology alfresco: Psychoendocrine studies of wild baboons. Recent Progress in Hormone Research, 48, 437-468.
Sapolsky, R. (1993b). The physiology of dominance in stable versus unstable hierarchies. In W. Mason and S. Mendoza (Eds.), Primate Social Conflict (p. 171). New York: SUNY Press.
Sapolsky, R. (2005). The influence of social hierarchy on primate health. Science, 308, 648-652.
Sapolsky, R. (2011). Sympathy for the CEO. Science, 333, 293-294.
Sapolsky, R., and Ray, J. (1989). Styles of dominance and their physiological correlates among wild baboons. American Journal of Primatology, 18, 1-11.
Sapolsky, R., and Share, L. (1994). Rank-related differences in cardiovascular function among wild baboons: Role of sensitivity to glucocorticoids. American Journal of Primatology, 32, 261-268.
Sapolsky, R., and Share, L. (2004). A pacific culture among wild baboons, its emergence and transmission. Public Library of Science Biology, 2, 106-114.
Sapolsky, R., and Spencer, E. (1997). Social subordinance is associated with suppression of insulin-like growth factor I (IGF-I) in a population of wild primates. American Journal of Physiology, 273, 1346-1351.
Sapolsky, R., Alberts, S., and Altmann, J. (1997). Hypercortisolism associated with social subordinance or social isolation among wild baboons. Archives of General Psychiatry, 54, 1137-1143.
Sapolsky R., Romero, L., and Munck, A. (2000). How do glucocorticoids influence the stress-response? Integrating permissive, suppressive, stimulatory and preparative actions. Endocrine Reviews, 21, 55-72.
Sen, A. (1992). Inequality Reexamined. Oxford, UK: Oxford University Press.
Shutt, K., MacLarnon, A., Heistermann, M., and Semple, S. (2007). Grooming in Barbary macaques: Better to give than to receive? Biology Letters, 3, 231-233.
Siegrist, J., and Marmot, M. (2004). Health inequalities and the psychosocial environment—two scientific challenges. Social Science & Medicine, 58(8), 1463-1473.
Smith, J. (1999). Healthy bodies and thick wallets: The dual relationship between health and socioeconomic status. Journal of Economic Perspectives, 13, 145-166.
Smith, A. (1776/2003). Wealth of Nations. New York: Bantam Classic.
Stafford, M., and Marmot, M. (2003). Neighborhood deprivation and health: Does it affect us all equally? International Journal of Epidemiology, 32, 357-366.
Stansfeld, S., Bosma, H., Hemingway, H., and Marmot, M. (1998). Psychosocial work characteristics and social support as predictors of SF-36 functioning: The Whitehall II study. Psychosomatic Medicine, 60, 247-255.
Stansfeld, S., Fuhrer, R., Shipley, M., and Marmot M. (1999). Work characteristics predict psychiatric disorder: Prospective results from the Whitehall II study. Occupational and Environmental Medicine, 56, 302-307.
Steptoe, A., and Marmot, M. (2002). The role of psychobiological pathways in socio-economic inequalities in cardiovascular disease risk. European Heart Journal, 23, 13-25.
van Rossum, C., Shipley, M., Van de Mheen, H., Grobbee, D., and Marmot, M. (2000). Employment grade differences in cause specific mortality. A 25 year follow up of civil servants from the first Whitehall study. Journal of Epidemiology and Community Health, 54, 178-184.
Victora, C., Wagstaff, A., Schellenberg, J., Gwatkin, D., Claeson, M., and Habicht, J. (2003). Applying an equity lens to child health and mortality: More of the same is not enough. The Lancet, 362, 233-241.
Virgin, C., and Sapolsky, R. (1997). Styles of male social behavior and their endocrine correlates among low-ranking baboons. American Journal of Primatology, 42, 25-31.
Whiten, A. (2005). The second inheritance system of chimpanzees and humans. Nature, 437, 52-55.
Wilkinson, R., and Pickett, K. (2010). The Spirit Level. London, UK: Penguin Books.
Wilson, M., and Daly, M. (1997). Life expectancy, economic inequality, homicide and reproductive timing in Chicago neighbourhoods. British Medical Journal, 314, 1271-1274.