Principles of Social Psychiatry

FUNDAMENTAL SOCIAL CAUSES OF HEALTH INEQUALITIES 183 80 77 75 74 70 70 65 61 60 55 55 50 47 45 40 Figure 14.2 US life expectancy at birth 1900 1920 1940 1960 1980 2000 1900–2000 [4] diseases, age-adjusted death rates from influenza and as heart disease, influenza and, since 1990, all cancers pneumonia dropped from 48 per 100 000 in 1950 to 24 combined. It seems clear that whatever is driving per 100 000 in 2000. improvements in population health, that factor is not What is driving these dramatic improvements in coterminous with the primary factors that have been health? Clearly, some powerful processes are at work, put forward to explain health inequalities by SES which are having a remarkably strong impact on and race. population health. Whatever these processes are – let So what is causing the dramatic improvements in us call them ‘X’ – should not X also have relevance for population health – what is X? Of course, it is not one explaining inequalities in health by SES and race? single thing but many different things. For example, Certainly, explanations of health inequalities should recent declines in age-adjusted rates of lung cancer not ignore this factor. From this vantage point, let us are probably influenced by the lagged effects of return to the centrepieces of previous explanations for declining smoking rates in previous decades. The health inequalities and ask whether these factors rapid decline in HIV/AIDS mortality in the United might be the same ones causing the dramatic improve- States is probably related to antiretroviral drugs ments in population health we have observed. Let us developed in the late 1990s. The precipitous decline begin with the first social selection explanation, in in mortality due to Hodgkin disease since the 1960s is which genes affect health, which in turn affects SES. probably due to the development of chemotherapy Clearly, disease-related genes cannot have changed so treatments that are able to cure the disease. The list of rapidly and in such a uniformly positive direction as to factors that have contributed to improvements in have created the enormous improvements in popula- population health is a long one, including screening tion health documented above. Next consider the for disease, public health efforts to increase the second social selection explanation, in which genes consumption of fruits and vegetables, promote influence factors like intelligence or conscientious- exercise and eradicate smoking, pollution control, ness, which influence both health and SES. Again, one flu shots, seat belts, cholesterol drugs, angioplasty, would have to posit enormous and implausible gains and so on. It seems clear that improvement in popula- in these traits over the last 50–100 years to explain the tion health is not due to just one thing and that it is observed improvements in health. Turning to the likely to be due to different factors for different mainstay of social causation explanations, we encoun- diseases. Just as clearly, the confluence of these many ter similar problems. One would have to posit that the factors has had an enormously positive impact on prevalence of stress or adversity has declined drama- population health. Over the last century, human tically over the last several decades and that this beings have dramatically increased their capacity to decline had a powerful impact on illnesses as diverse control disease and death.

184 SOCIAL DETERMINANTS 14.3 THE CORE PROPOSITION: CONTROLLING DISEASE AND CREATING INEQUALITIES We can now state our core proposition. It is our connections. According to this principle, whatever enormously expanded capacity to control disease and the health differences between advantaged and dis- death in combination with existing social and eco- advantaged groups might have been before a health- nomic inequalities that create health inequalities by enhancing discovery, the uneven distribution of new race and SES. It does so because of a very basic knowledge and technology results in a powerful principle. As we develop the ability to control disease social shaping of health inequalities. and death, the benefits of this new-found capacity are This explanation for health inequalities is a core not distributed equally throughout the population, but component of the theory offundamental social causes. are instead harnessed more securely by individuals Because our proposition derives from that theory, and groups who are less likely to be exposed to we briefly develop a broader rationale for the theory discrimination and who have greater access to knowl- and then turn to evidence that bears on our core edge, money, power, prestige and beneficial social proposition. 14.4 FUNDAMENTAL SOCIAL CAUSES OF HEALTH INEQUALITIES The theory of fundamental social causes begins with living conditions were the principal risk factors. the graded relationship regularly observed between People of lower SES were exposed to these condi- indicators of socioeconomic status, on the one hand, tions to a greater degree and had much higher and health and mortality, on the other [6–9]. Clearly, mortality rates as a consequence [10,11]. Subse- biological mechanisms must be involved in this quently, however, sanitation was greatly improved, association. Just as clearly, other mechanisms invol- water systems were made safe, housing conditions ving behaviours and environmental exposures must dramatically improved and effective vaccines were also be present: disease does not flow directly from developed, with the result that death from cholera, income, educational or occupational statuses into the diphtheria, measles, smallpox or tuberculosis is now body. Despite the necessary role played by these rare in the United States and Western Europe. We mechanisms, the effect of SES on health and mor- would have expected the association between SES tality cannot be understood by focusing only on the and mortality to disappear, because the mechanisms mechanisms that happen to link them at any parti- linking them were virtually eliminated or blocked. cular time. But it did not disappear. Why not? New risk factors Imagine a causal model with SES as the distal (such as chemical pollutants) arose, new knowledge factor that is linked to death by more proximal risk about risk factors (such as smoking) emerged and factors. If the proximal risk factors are eliminated, we new treatment technologies (such as medicines that would expect the SES–mortality association to dis- reduce cholesterol) were developed, and those pos- appear. On the contrary, there have been several sessing the most resources were best able to avoid the notable instances in which major proximal risk fac- new risks and take advantage of the new protective tors were eliminated but SES inequalities in mortality factors, resulting in the emergence of a socioeco- persisted. Consider Europe and the United States in nomic gradient in these factors. The list of circum- the nineteenth century, when diseases such as cho- stances that are shaped by SES-related resources is lera, diphtheria, measles, smallpox and tuberculosis very long and is not limited to the standard beha- were the major causes of death, and poor sanitation, vioural risk factors (e.g. smoking, exercise, diet) contaminated water, and substandard and crowded typically measured in risk factor epidemiology. For

FUNDAMENTAL SOCIAL CAUSES OF HEALTH INEQUALITIES 185 example, Lutfey and Freese [12] use an ethnographic According to the fundamental cause idea, this approach to study the management of diabetes; they dynamic reproduction of the association between SES show how the organization of clinics (single provider and disease occurs because the flexible nature of versus rotating providers), physicians’ expectations resources of knowledge, money, power, prestige and of patients’ capacity to use the newest diabetes- beneficial social connections allows the association control techniques, access to insurance and many to be reproduced in widely varying circumstances. other circumstances result, on average, in an advan- Flexible resources are important in at least two ways. taged circumstance for those who are rich in socio- First, they directly shape individual health behaviours economic resources. As Lutfey and Freese argue, it is by influencing whether individuals are aware of, have when circumstances like these are reproduced across access to, can afford and are supported in their efforts many situations (e.g. situations related to health to engage in health-enhancing behaviours. Second, behaviours, preventive health care and the full range resources shape access to broad contexts such as of existing diseases) that the robust association neighbourhoods, social networks and occupations that between SES-related resources and health emerges. vary in their associated profiles of risk and protective As new discoveries are made, our ability to control factors. For example, white-collar jobs are less dan- disease advances, new items will be added to the list gerous and more likely to include health care benefits of health-enhancing circumstances and, our theory than blue-collar jobs. Low-income housing is more says, those who command more resources will, on likely to be located in neighbourhoods whose limited average, be better able to access and benefit from the power and political organization make them vulner- new knowledge we gain. In this way, the association able to noise, pollution and noxious social conditions, between SES and disease is reproduced dynamically including targeted advertising of health-harming pro- through a complex and evolving set of intervening ducts such as tobacco and alcohol. Thus, the processes mechanisms that change over time and vary from implied by the fundamental cause perspective operate place to place. at both individual and contextual levels. 14.5 EMPIRICAL EVIDENCE BEARING ON THE THEORY We argue that SES inequalities in mortality arise 14.5.1 SES associations with more and because people of higher SES use flexible resources less preventable causes of death to avoid risks and adopt protective strategies. It fol- lows that SES–mortality gradients should be dimin- Phelan et al. [9] tested this prediction using the ished when people cannot use their resources in this National Longitudinal Mortality Study (NLMS) and way. Phelan et al. [9] constructed a test of the funda- ratings they developed of the preventability of death mental cause explanation by identifying a situation in from specific causes. The NLMS [3,13] is a large which it is difficult to use resources to prolong life – American prospective study that uses combined sam- when even the richest or most powerful person on ples of selected Current Population Surveys that are earth cannot use resources to escape death. This is the then linked to the National Death Index to determine case when we consider death from diseases that we do occurrences and causes of death in a follow-up period not yet know how to prevent or treat. If the utilization of approximately nine years. Reliable ratings (intra- of resources is critical in prolonging life, then, when class correlation of 0.85) of the preventability of death resources associated with higher status are useless, were made by two physician-epidemiologists. Causes high SES should confer little advantage, and the were categorized into high-preventability and low- usually robust SES–mortality association should be preventability groups. Common high-preventability reduced. If these expectations were disconfirmed, it causes included cerebrovascular diseases, chronic would pose a serious challenge to the theory. obstructive pulmonary disease, ischaemic heart

186 SOCIAL DETERMINANTS disease, malignant neoplasm of the trachea, bronchus table include heart disease, Hodgkin disease, lung and lung, and pneumonia and influenza; common low- cancer and colon cancer. In contrast, if death remains preventability causes included arrhythmias and ma- difficult or impossible to prevent, as it is for brain and lignant neoplasms of the pancreas, female breast and ovarian cancer, inequalities should not change sub- prostate. Gradients according to SES indicators of stantially with time. The Phelan et al. [9] test did not education and income were then examined separately provide evidence on changes in inequalities over time. for high- and low-preventability causes. Consistent Here we report evidence bearing on these predictions. with predictions derived from the fundamental cause First consider mortality trends for diseasesfrom which explanation, Phelan et al. found that the SES–mor- death remains difficult to prevent. Figure 14.3 presents tality association was significantly stronger for highly age-adjusted brain cancer mortality rates per 100 000 preventable causes of death than for less preventable men in the United States between 1950 and 1999. causes of death. For example, for individuals between Consistent with the idea that we have not yet learned 45 and 64 years of age, the relative risk of death how to prevent death from this disease, the death rates associated with having an eight-grade education as are lower for black than for white groups, and the compared with more than a college education is 2.00 difference between the groups remains relatively con- for high-preventability causes as opposed to 1.21 for stant through time, as do the modestly lower rates for low-preventability causes. Similarly, in the same age black compared with white. group, the relative risk of death associated with an Consider next some diseases for which great strides income less than $5000 compared with more than in prevention and treatment have been made. A dra- $50 000 is 2.81 for high-preventability causes and maticexampleisheartdisease.AsshowninFigure14.4, 1.86 for low-preventability causes. two patterns are obvious. First, consistent with the idea that we have made great strides in our capacity to prevent death from this disease, age-adjusted rates 14.5.2 Time trends in socioeconomic and declined rapidly between 1950 and 2000 in the US. racial inequalities in mortality Second, whereas rates of death for black and white populations were quite similar in 1950, inequalities If our core proposition is correct, inequalities by SES favouring white over black populations have emerged and race should emerge when new health-enhancing since that time. In addition, death rates from heart information or technology is developed. Diseases for disease are now much higher among those with lower which death has become dramatically more preven- levels of education. Taken together, this evidence is 6 5.47 5.26 5 4.81 4.97 5.17 5.04 4.5 4 4.18 3.86 White Males 3 3.21 2.91 3.1 3.08 2.84 2 Black Males Figure 14.3 Brain cancer – age-adjusted death rates per 100 000 1950–1999 (males) (From 1 Cancer Mortality Maps and Graphs website, National Cancer Institute: http://cancer/gov/ 0 atlasplus/) 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89 90-94 95-99

FUNDAMENTAL SOCIAL CAUSES OF HEALTH INEQUALITIES 187 600 586.7 559 584.8 550 548.3 512 500 492.2 450 455.3 Black 409.4 400 391.5 350 White 317 324.8 300 250 253.4 Figure 14.4 Heart disease – age-adjusted 200 death rates per 100 000 1950–2000 [14] 1950 1960 1970 1980 1990 2000 consistent with the idea that when we develop the were most important in defining that component were wherewithal to prevent or treat a disease, groups that median family income, family poverty rate and per- are richer in resources and that are less likely to experi- centage of the population with more than 12 years of ence discrimination (such as white individuals and education. Singh, Miller and Hankey [16] examined people with higher levels of education) benefit more the stability of this measure using data from the 1970 fully from the new-found capacity to control disease. and 1980 Censuses and found it to be quite stable. Monitoring mortality rates by SES in the United Using quintiles of the socioeconomic scores, Singh States is more problematic than by race, because no et al. [16] observed dramatic evidence of changing indicators of SES were identified on US death certi- associations between SES and two major cancer kill- ficates until recently. To achieve some assessment of ers: lung cancer and colon cancer. Figure 14.5 shows SES trends over time, Singh et al. [15] employed age-adjusted lung cancer mortality rates for three principal components analysis to develop socioeco- quintiles (highest, middle and lowest) for men aged nomic scores from 1990 Census data for every county 25–64 years. Rates were substantially higher in the in the United States. Scores were based on 11 variables highest SES counties in 1950. However, once evi- measuring county-level SES. Of all the 11 variables dence about the harmful effects of smoking emerged loaded on one component, the three variables that in the late 1950s and early 1960s, mortality in the 90 80 70 60 50 High SES 40 Middle SES 30 20 Low SES Figure 14.5 Age-adjusted lung 10 cancer mortality (men 25–64) 0 1950–1998 by SES of county of 50 56 62 68 74 80 86 92 98 residence [16]

188 SOCIAL DETERMINANTS 25 High SES 20 15 Middle SES 10 Low SES 5 Figure 14.6 Age-adjustedcoloncancermortality (men 25–64) 1950–1998 by SES of county of 0 residence [16] 50 56 62 68 74 80 86 92 98 highest-SES counties began to flatten out, while rates to take account of so-called ‘competing risks’. In this in the poorest counties continued to rise. By 1998, the alternative explanation, inequalities emerge when association between county-level SES and lung can- black people or individuals from lower SES areas cer mortality had completely reversed, such that begin to die at higher rates of diseases like heart the highest rates are now in the lowest-SES counties. disease or colon cancer because they survive long The situation for women is not as dramatic, but a more enough to do so. However, our focus on relatively modest version of the same reversal seems to be early death (before 65 years) should minimize this underway: until the late 1980s, women from high- problem, and the effect of such competing risks would SES counties had the highest rates of lung cancer have to be extremely large to create the observed mortality, but by 1998 they had the lowest rates. inequalities. Finally, this potential limitation does not Figure 14.6 shows age-adjusted mortality rates apply to our earlier test using the NLMS, as those data from colon cancer in men (aged 25–64 years) in three were analysed using a competing risks model [9]. quintiles of county-level SES (highest, middle and A second potential limitation of the time-trend data lowest). As the figure shows, men in low-SES counties is the possibility of diagnostic change over time by had dramatically lower rates of colon cancer mortality race and SES. It is possible that in an earlier era black in 1950. However, as our ability to prevent death from people and individuals of low SES were less likely to this disease increased in the ensuing decades, men in receive accurate cause-of-death designations, produ- high-SES counties experienced a dramatic decline in cing lower rates in these groups for diseases like heart mortality rates, while men in low-SES counties did disease and colon cancer. However, by this reasoning, not, such that, currently, overall mortality rates from we would also expect to find inequalities emerging for colon cancer are modestly higher in low-SES counties diseases like brain cancer and ovarian cancer, but we than in high-SES counties. A similar pattern occurred do not. Thus it is not evident how a diagnostic-bias for women. explanation would account for these differences. Sec- Two potential limitations apply to the trend data ond, our previous test using the NLMS examines a reported in Figures 14.3 to 14.6. The first is the period of just 9 years and is not nearly as subject to this possibility that in examining specific diseases we fail form of bias. 14.6 FUNDAMENTAL CAUSES AND PSYCHIATRIC ILLNESS As with physical illnesses, there are strong inverse These patterns were brought to light in classic works in gradients between SES and many psychiatric illnesses. psychiatric epidemiology, such as Hollingshead and

FUNDAMENTAL SOCIAL CAUSES OF HEALTH INEQUALITIES 189 Redlich’s [17] Social Class and Mental Illness, Srole There is evidence that both selection and stress play and colleagues’ [18] Midtown Manhattan Study and some causal role in producing the inverse association Leighton’s [19] Stirling County Study. In more recent between SES and psychiatric illness. Regarding research that employs very different methodologies selection, in Dohrenwend et al.’s [24] large-scale and definitions of psychiatric illness, the basic patterns epidemiological study in Israel designed to test social remain. For example, Kessler et al. [20] found mono- selection and social causation explanations for the tonic associations between educational attainment and association between SES and specific psychiatric several psychiatric disorders in the US. Odds ratios for disorders, the authors concluded that the evidence people with fewer than 12 years of education as was more consistent with a social selection explana- compared with people with 16 or more years of tion in the case of schizophrenia. education are 1.79 for any affective disorder, 2.10 for However, with the exception of schizophrenia, any substance abuse and 3.79 for having three or more evidence indicates that social causation processes are disorders of any kind. Therefore, the socioeconomic of greater significance than are social selection pro- distributions of physical and psychiatric disorders, cesses in explaining the connection between SES and broadly speaking, are similar – those with lower SES psychiatric illness [25]. Within the general social have more disorder. causation approach, the social stress explanation has However, when we consider historical trends in been especially prominent [26]. Much of the argument prevalence, we find a striking contrast between psy- for the role of stress in producing social inequalities in chiatric and physical disorders. Although age- psychiatric illness rests on findings of differences in adjusted mortality rates have declined dramatically, rates of illness between socially advantaged and dis- and people are now experiencing many more years of advantaged groups, and thus does not provide direct life free from physical disability, the same sort of evidence that social stress is the explanation for those sea change has not occurred for psychiatric illness. differences [27]. However, there also exists direct Indeed, the best evidence from true prevalence studies evidence pointing to the importance of stress in ele- finds current rates to be very high [20,21]. Moreover, vating psychiatric problems in lower socioeconomic evidence about secular changes, while subject to groups. For example, Turner, Wheaton and Lloyd [28] debate, finds rates of depression that are either higher found that exposure to recent stressful life events, in more recent cohorts or that do not differ from the lifetime trauma and chronic stress explained about older cohorts that preceded them [22,23]. Similarly, 30% of the association between occupational prestige Kessler et al. [21] found that the 12-month prevalence and major depressive disorder in a representative of psychiatric disorders overall did not change sig- sample of Toronto residents. nificantly in the US between 1990–1992 and What are the implications of this pattern offindings – 2001–2003. significant socioeconomic inequalities in psychiatric Two general implications of these patterns are illness coupled with an overall rate of illness that has important for our purpose here. First, in the case of not declined – from the perspective of fundamental mortality from physical illness, we concluded that the cause theory? The historical data suggest that whatever significant improvements in population health in the is causing any SES or race inequalities that exist at the last century were incompatible with both genetic current time is something other than our capacity to selection explanations and social stress explanations, decrease the prevalence of these illnesses and the mal- because it would be implausible to posit the kind of distribution of that capacity in the population. If our changes in these factors that would have been neces- capacity to prevent or treat effectively had increased, sary to explain the reductions in mortality. However, overall rates of disorder should have declined. the historical pattern for psychiatric illness is compa- This conclusion is not surprising if we think about tible with both selection and social stress explana- the knowledge we have accrued regarding the preven- tions. These historical data cannot tell us the extent to tion of psychiatric illness. We believe most clinicians which either of these explanations is actually valid, and researchers would agree that we have garnered but it leaves them both open as viable explanations. little specific knowledge about the measures one can

190 SOCIAL DETERMINANTS take to avert a psychiatric illness. From this perspec- view as a structural aspect of stigma, also surely tive, we would not expect to have found declines in the decreases utilization of effective treatments. Some incidence or lifetime prevalence of most psychiatric evidence exists that perceptions of mental illness illnesses. stigma are associated with withdrawal from treatment However, the persistence of high rates of psychia- once begun [32,33]. tric disorder is somewhat more surprising if we think Whateverthereasonsthatratesofpsychiatricillness about advances in the development of effective treat- have failed to decline, we believe it is only a matter of ments. We have made tremendous strides in the time before this situation changes. Unless the massive treatment of psychiatric illnesses with both psy- nationalinvestmentinneurobiologyandgeneticsfails, chotherapy and medication. These treatments would we will gain the capacity to influence the onset and not be expected to reduce incidence or lifetime pre- course of common but serious psychiatric illnesses, valence of disorder because their benefits can only be and given that large inequalities in access to mental garnered after an illness develops. However, the health treatment according to SES, race and ethnicity development of effective treatments should reduce exist at the current time [17,34,35], we can almost current or short-term prevalence, such as 12-month certainly expect that any new knowledge about how to prevalence, because effective treatment should reduce prevent or cure psychiatric illnesses will be maldis- individuals’ symptoms such that they no longer meet tributed by these same factors in the time ahead. In this diagnostic criteria for the disorder for which they were way, social inequalities will become an even more treated. Yet the data do not indicate a reduction in important determinant of disease distribution in an era 12-month prevalence [21]. Why has the availability of of successful prevention and treatment. They will add effective treatments not led to a significant decline in to or combine with the factors that currently produce these rates of psychiatric illness? associationsbetweenSESandthe psychiatric illnesses. We can only speculate on the reasons, which may Some readers may doubt the ultimate success of include a tendency for individuals to stop psychother- neurobiological and genetic approaches in reducing apy or medication when symptoms subside or to the prevalence of psychiatric illness – it is possible that terminate medications because of unpleasant side our enormous investment in these approaches will fail. effects. Also undoubtedly an important factor is that Ifso,socialfactors willnotbeascendantforthereasons only a minority of people with psychiatric disorders we have speculated about here. Instead, the situation receive treatment in the first place. For example, in the shall remain much as it is: robust associations between National Comorbidity Study Replication, a large SES and many psychiatric disorders will be explained nationally representative US survey of psychiatric through processes of social stress or social selection illness and treatment, only 41.1% of people who met across the life course. The importance of social status criteria for a psychiatric disorder in the previous 12 has been with us since the inception of psychiatric months received any treatment during that period, and epidemiology, it is with us today and, if the reasoning only 32.7% of the 41.1% received ‘minimally ade- represented herein is apt, it will become an even more quate treatment’ [29]. Many factors surely contribute prominent cause in the time ahead. to this huge discrepancy between need and treatment. We point to the stigma that still surrounds psychiatric illness as a prominent factor [30]. Receiving treatment REFERENCES for a psychiatric illness formalizes and solidifies a psychiatric illness label. Medication for depression, 1. Antonovsky, A. (1967) Social class, life expectancy and the most common psychiatric disorder, is presented overall mortality. Milbank Memorial Fund Quarterly, quite negatively in the mass media, often as a self- 45, 31–73. indulgent crutch [31], and people who take such 2. Kunst, A. E., Feikje, G., Mackenbach, J. P. and EU medication are also portrayed negatively. Nonparity Working Group on Socioeconomic Inequalities in between psychiatric and physical illness in terms of Health (1998) Occupational class and cause specific mortality in middle aged men in 11 European countries: health insurance coverage, a state of affairs that we

FUNDAMENTAL SOCIAL CAUSES OF HEALTH INEQUALITIES 191 comparison of population based studies. British Med- Health of Americans, National Center for Health Sta- ical Journal, 316, 1636–1642. tistics, Hyattsville, Maryland. 3. Sorlie, P. D., Backlund, E. and Keller, J. (1995) US 15. Singh, G., Miller, B., Hankey, B. et al. (2002) Changing mortality by economic, demographic, and social char- area socioeconomic patterns in U.S. cancer mortality, acteristics: The National Longitudinal Mortality Study. 1950–1998: Part I – All cancers among men. Journal of American Journal of Public Health, 85, 949–956. the National Cancer Institute, 94, 904–915. 4. National Center for Health Statistics (2006) Health 16. Singh, G., Miller, B. and Hankey, B. (2002) Changing United States 2006: With Chartbook on Trends in area socioeconomic patterns in U.S. cancer mortality, the Health of Americans, National Center for Health 1950–1998: Part II – Lung and colorectal cancers. Statistics, Hyattsville, Maryland. Journal of the National Cancer Institute, 94, 5. Lantz, P. M., House, J. S., Lepkowski, J. M. et al. (1998) 916–925. Results from a nationally representative prospective 17. Hollingshead, A. and Redlich, F. (1958) Social Class study of US adults. Journal of the American Medical and Mental Illness, John Wiley and Sons, Inc., New Association, 279, 1703–1708. York. 6. Link, B. G. and Phelan, J. C. (1995) Social conditions as 18. Srole, L., Langner, T., Michael, S. et al. (1962) Mental fundamental causes of disease. Journal of Health and Health in the Metropolis: The Midtown Manhattan Social Behavior, 35 (Extra Issue: Forty Years of Med- Study, McGraw-Hill, New York. ical Sociology: The State of the Art and Directions for 19. Leighton, A. (1959) My Name is Legion: The Stirling the Future), 80–94. County Study of Psychiatric Disorder and Social 7. Link, B. G. and Phelan, J. C. (1996) Understanding Environment, vol. 1, Basic Books, New York. sociodemographic differences in health – the role of 20. Kessler, R. C., McGonagle, K., Zhao, S. et al. (1994) fundamental social causes. American Journal of Public Lifetime and 12-month prevalence of DSM-III-R psy- Health, 86, 471–473. chiatric disorders in the United States: results from the 8. Link, B. G. and Phelan, J. C. (2000) Evaluating the National Comorbidity Study. Archives of General Psy- fundamental cause explanation for social disparities in chiatry, 51, 8–19. health, in The Handbook of Medical Sociology, 5th edn 21. Kessler, R. C., Demler, O., Frank, R. G. et al. (2005) (eds C. E. Bird, P. Conrad andA. M. Freemont), Prentice Prevalence and treatment of mental disorders, 1990 to Hall, Upper Saddle River, New Jersey, pp. 33–46. 2003. New England Journal of Medicine, 352, 9. Phelan, J. C., Link, B. G., Diez-Roux, A. et al. (2004) 2515–2523. ‘Fundamental causes’ of social inequalities in mortal- 22. Klerman, G. L. (1989) The current age of youthful ity: a test of the theory. Journal of Health and Social melancholia: evidence for increase in depression Behavior, 45, 265–285. among adolescents and young adults, in Annual 10. Chapin, C. V. (1924) Deaths among taxpayers and non- Progress in Child Psychiatry and Child Development taxpayers income tax, Providence, 1865. American (eds S. Chess and M. Hertzig), Brunner/Mazel, Journal of Public Health, 4, 647–651. Philadelphia, Pennsylvania, pp. 333–335. 11. Villerme, L. (1840) Tableau d’Etat Physique et Moral 23. Murphy, J., Laird, N., Monson, R. et al. (2000) A 40- des Ouvriers, vol. 2, Jules Renouard et Cie, Paris. year perspective on the prevalence of depression: the 12. Lutfey, K. and Freese, J. (2005) Toward some funda- Stirling County Study. Archives of General Psychiatry, mentals of fundamental causality: socioeconomic sta- 57, 209–215. tus and health in the routine clinic visit for diabetes. 24. Dohrenwend, B. P., Levav, I., Shrout, P. et al. (1992) American Journal of Sociology, 110, 1326–1337. Socioeconomic status and psychiatric disorder: the 13. Rogot, E., Sorlie, P. D., Johnson, N. J. and Schmitt, C. causation-selection issue. Science, 255, 946–952. (1992) A Mortality Study of 1.3 Million Persons by 25. Aneshensel, C.S. (1992) Social stress: theory and Demographic, Social, and Economic Factors: research. Annual Review of Sociology, 18, 15–38. 1979–1985 Follow-Up, NIH publication 92-3297, 26. Pearlin, L. I. (1989) The sociological study of stress. National Institutes of Health, Bethesda, Maryland. Journal of Health and Social Behavior, 30, 241–256. 14. National Center for Health Statistics (2003) Health 27. Schwartz, S. and Meyer, I. (In press) Mental health United States 2003: With Chartbook on Trends in the disparities research: The impact of within and between

192 SOCIAL DETERMINANTS group analyses on tests of social stress hypotheses./ 32. Sirey, J. A., Bruce, M. L., Alexopoulos, G. S. et al. Social Science and Medicine. (2001) Perceived stigma as a predictor of treatment 28. Turner, R. J., Wheaton, B. and Lloyd, D. A. (1995) The discontinuation in young and older outpatients with epidemiology of social stress. American Sociological depression, American Journal of Psychiatry, 158 (3), Review, 60, 104–125. 479–481. 29. Wang, P. S., Lane, M., Olfson, M. et al. (2005) Twelve- 33. Sirey, J. A., Bruce, M. L., Alexopoulos, G. S. et al. month use of mental health services in the (2001) Stigmaas a barrier to recovery: perceived stigma United States. Archives of General Psychiatry, 62, and patient-rated severity of illness as predictors of 629–640. antidepressant drug adherence. Psychiatric Services, 52(12), 1615–1620. 30. Phelan, J. C., Link, B. G., Stueve, A. and Pescosolido, B. A. (2000) Public conceptions of mental illness in 34. Alegria, M., Canino, G., Rios, R. et al. (2002) Mental 1950 and 1996: What is mental illness and is it to be health care for Latinos: inequalities in use of specialty feared? Journal of Health and Social Behavior, 41, mental health services among Latinos, African Amer- 188–207. icans, and non-Latino whites. Psychiatric Services, 53, 1547–1555. 31. Kuppin, S. (2006) The American Public and Prozac: An In-depth Analysis of the American Public’s Views 35. Lorant, V., Kampfl, D., Seghers, A. et al. (2003) Socio- on the Causes of Depression and Its Treatment with economic differences in psychiatric impatient care. Antidepressant Medication, Unpublished doctoral Acta Psychiatrica Scandinavia, 107, 170–177. dissertation.

15 The sociodevelopmental origins of psychosis 1 Craig Morgan and Gerard Hutchinson 2 1 NIH Biomedical Research Centre and Institute of Psychiatry, King's College London, London, UK 2 Psychiatry Unit, Department of Clinical Medical Sciences, University of the West Indies, Mount Hope, Champs Fleurs, Trinidad In the past twenty years the dominant view, at least trauma and adversity [3,4] have been mired in see- until recently, has been that schizophrenia (and other mingly irresolvable debates about whether these are psychoses) is a genetic brain disease, the onset of causal (or index exposure to causal processes) or which is the product of a neurodevelopmental pro- simply reflect an underlying genetic or biological cess [1]. Social factors, if they have been assigned a vulnerability that compromises social development role at all, have been relegated to the status of triggers, (drift versus cause). Second, the mechanisms by serving merely to hasten the onset of a largely biolo- which social contexts and experiences impact on gically determined disease. Once disorder is estab- individuals to increase risk of schizophrenia and other lished, it is generally conceded that social environ- psychoses have been poorly specified. Only a small ments and experiences, particularly interpersonal proportion of those who are exposed to adverse social family relationships, have a role in course and out- conditions, traumatic life events, and so on, ever come. However, their role in aetiology, beyond that of develop a serious mental disorder. The types of ad- precipitants, remains contested. Here, in reviewing the verse social conditions associated with psychosis are relevant literature, we address the question: Is there not specific; they are also associated with a range of now substantive evidence that socioenvironmental other disorders including depression and anxiety [5]. factors are important in the aetiology of psychosis, If such experiences are relevant to the onset of psy- beyond merely triggering onset? chosis, why are so few people affected? Ironically, it is The proposition that socioenvironmental factors are advances in genetics, neuroscience and psychology important has, in the past, been undermined by two that are beginning to suggest plausible mechan- particular problems. First, repeated findings (dis- isms [6–10] and, in doing so, provide further impetus cussed fully below) that those with a serious mental to research on the relationship between social factors disorder are more likely to live in cities [2], to be of and psychosis. More than this, such research provides low socioeconomic status and to experience more the basis for more genuinely integrated models of Principles of Social Psychiatry, second edition Edited by Craig Morgan and Dinesh Bhugra Ó 2010 John Wiley & Sons, Ltd.

194 SOCIAL DETERMINANTS aetiology, which posit interactions between genes, Some of the impetus for this line of thought stems biology, psychology and society over the life from recent epidemiological research that suggest course [11,12]. We take this one step further and ask notable variations in the incidence of schizophrenia whether there is now evidence for a (predominantly) and other psychoses by place and social group. sociodevelopmental pathway to psychosis. 15.1 A CHANGING EPIDEMIOLOGICAL LANDSCAPE In the 1960s, 1970s and 1980s, the World Health ratio median of 1.4), and confirmed the higher rates in Organization conducted a series of seminal studies of urban (versus mixed rural–urban) areas and in mi- the epidemiology of schizophrenia [13,14]. Perhaps grant (versus host) populations (rate ratio median of the most influential of these, the Determinants of 4.6). These latter two findings have been replicated OutcomesofSevereMentalDisorder(DOSMeD)[14], and extended in other recent and more specific re- found that the incidence of narrowly defined schizo- views [2,17], and each is discussed in turn below. In phrenia, in the seven countries for which reliable fact, from the beginning, the interpretation of a uni- estimates were available, ranged from 7 per 100 000 form incidence did not go unchallenged. A number of persons per year in Aarhus (Denmark) to 14 per commentators pointed out that, although statistically 100 000 persons per year in Nottingham (UK), a sta- nonsignificant, there was a twofold difference be- tistically nonsignificant variation (at p < 0.05). This tween the highest and lowest reported incidence rates was widely interpreted as indicating that there is a for narrow schizophrenia in the DOSMeD study, and, uniform worldwide incidence of schizophrenia, an for broadly defined schizophrenia, there were even interpretation that has been taken as further evidence more marked differences among the various cen- that the disorder is primarily genetic – the usual tres [19]. The most parsimonious explanation for the variability that would be expected if the occurrence initial DOSMeD finding is that the study was under- of schizophrenia was influenced by external environ- powered to detect geographical variations in inci- ments was simply not evident [15,16]. For example, as dence (n ¼ 302 cases across 8 sites in 7 countries). As Crow states ([16], p. 15): ‘Incidences are relatively McGrath [20] has commented, ‘... schizophrenia is uniformacrosspopulations,particularlywhenthecore not the egalitarian disorder that we once thought it (nuclear) symptoms are assessed .... The fact that was’ (p. 14). To rephrase Crow, it seems that it is incidence is apparently unrelated to geographic or geographical and environmental variance (not invar- other environmental variation ... is the key observa- iance) that is the key observation that has to be tion that has to be explained by theories of aetiology.’ explained by theories of aetiology. However, in recent years, new research and meta- At the same time, a series of epidemiological analyses have challenged the interpretation that schi- surveys have reported surprisingly high rates of psy- zophrenia, even narrowly defined, has a uniform chosis or psychosis-like experiences (e.g. auditory incidence [17,18]. A comprehensive meta-analysis hallucinations, odd ideas, paranoia) in general popu- of 100 incidence studies by McGrath and collea- lation samples (estimates range from around 5% to gues [18], for example, found evidence of marked around 20%) [21,22]. This has reopened the debate variations across a number of domains. Overall, even about whether psychotic disorders are discrete enti- when the bottom 10% and top 10% were excluded, ties, marked by a clear disjunction from normal there was still more than a fivefold variation in experience, or whether they lie on a continuum with reported incidence rates among the included studies normality (see also Chapter 2) [23]. Of particular (median incidence rate: 15 per 100 000 persons per relevance here is that these experiences appear to year) – a variation that could not be explained by show similar (but perhaps less marked) variations by methodological heterogeneity. The review further area (urban versus rural) [24] and social group (mi- found evidence that rates are higher in men (rate grant versus host) [25] to those reported in studies of

THE SOCIODEVELOPMENTAL ORIGINS OF PSYCHOSIS 195 diagnostic groups. This has opened a particularly ment in the onset of psychosis (see below). Here, we fruitful avenue of research that is further contributing consider evidence concerning both psychotic disorder to renewed interest in the role of the (social) environ- and psychosis-like experiences. 15.2 URBANICITY: (1) EVIDENCE There is consistent evidence that urbanicity (variously of schizophrenia increased in line with length of defined) is associated with around a twofold increased residence in increasingly urbanized areas – a finding risk of schizophrenia (or psychosis more broadly). In a that suggests risk may accumulate over develop- meta-analysis of ten studies, for example, Krabben- ment as a consequence of continued exposure; and dam and van Os [2] report a pooled odds ratio of 1.72 (3) individuals living in a higher degree of urbaniza- (95% confidence interval (CI) 1.53–1.92), indicating tion than five years earlier had an increased risk on average a 70% higher incidence in urban compared (relative risk (RR) of 1.4), while individuals living with rural areas. Those studies that were able to adjust in a lower degree of urbanization than fiveyears earlier for potential confounders show that this association had a decreased risk (RR 0.8). In other words, moving cannot be entirely explained by the effects of age, to a less densely populated area during upbringing gender, ethnicity, social class or markers of genetic reduced the risk of schizophrenia (and vice versa). If risk [2]. Further, there is now evidence that subclinical the risk factors indexed by urbanicity, and any other psychosis-like experiences are more common in urban risks they interact with, are causal, then (based on areas, suggesting that on a population level urban calculations from the study by Pedersen and Morten- environments are associated with a general increase sen [28]) these may account for around 30% of all in vulnerability for psychosis [24]. In contrast, the cases of schizophrenia – ‘making it potentially the evidence on affective psychosis is more most important of all environmental risk factors pro- equivocal [26,27]. posed for schizophrenia’ ([2], p. 797). Nevertheless, with regard to schizophrenia at However, what it is about living in cities or more least, there is now compelling evidence that these densely populated areas that increases risk is un- findings are not simply a function of individuals clear [30]. It is consequently perhaps most useful to predisposed to develop, or in the process of devel- consider ‘urbanicity’ as a risk indicator, indexing oping, schizophrenia drifting into the anonymity of exposure to risk factors that coalesce in densely cities. In a study of 1.89 million people, drawn from populated environments. These risks are not necessa- Danish population registers, Pedersen and Morten- rily related to the social environment. However, where sen [28] investigated the relationship between place studies have included data on nuerodevelopmental of birth, place of residence before the age of 15 years markers (e.g. obstetric complications, neurocogni- and risk of schizophrenia. Place was characterized tion) or aspects of the physical environment (e.g. by degree of urbanization (population density) and traffic pollution, household crowding), they have not categorized into five levels from the most rural found evidence that these contribute to the urban (reference) to the most densely populated. They effect [2]. What limited evidence there is does tend found (in line with Marcelis et al. [29], in an earlier to implicate aspects of the social environment, parti- Dutch study): (1) at each age, risk of schizophrenia cularly organizational and structural aspects of city increased in line with degree of urbanization; (2) risk living [2]. 15.3 URBANICITY: (2) SOCIAL CONTEXTS AND STRUCTURES To begin with, some clues are offered by studies that schizophrenia (psychosis) within urban environments. have investigated variations in the distribution of This links back to the seminal study, conducted by

196 SOCIAL DETERMINANTS Faris and Dunham [31], of the social distribution of deprived areas of south-east London, area-level in- mental disorders in Chicago in the 1930s. In this, the equality was associated with an increased incidence addresses of all patients admitted with mental dis- of psychosis. The authors speculate that the associa- orders to four state and eight private hospitals in tion with inequality in the most deprived areas may Chicago over a twelve-year period were plotted on reflect the effects of reduced social cohesion and a census map of the city. Considerable variation in the social support and increased mistrust and isolation. rates of mental disorder according to place of resi- In a study of first admission rates for schizophrenia dence was found, with the highest rates being in those and area-level measures of social fragmentation, districts characterized by social disorganization and deprivation and urbanicity in Scotland, Allardyce fragmentation (i.e. areas with squalid housing, excess et al. [38] found that the association between urba- crime rates, a high level of social mobility, a high nicity and first admission rates was fully accounted proportion of foreign-born residents, a large propor- for by indices of social fragmentation (based on tion of the population living in single rooms or hostels, mobility in the previous year, single-person house- a high proportion who were unmarried and a high holds, number of unmarried persons) and deprivation proportion of people living below the official poverty (based on Carstairs scores). In the ÆSOP study of line). This finding was particularly marked for schizo- first-episode psychosis conducted in three UK cen- phrenia, where the range was from 700 per 100 000 in tres, the incidence of all psychoses was highest in city centre districts to 100 per 100 000 in the periph- south-east London (54.5 per 100 000), the most eral residential districts. In the 1950s, in a study in urbanized centre, compared with Nottingham (25.1 Bristol, Hare [32] found that rates of schizophrenia per 100 000) and Bristol (25.1 per 100 000) [39]. were associated with the proportion of single-person Within south-east London, there was evidence of households in an area, a finding he interpreted as additional variation at the small (ward) area-level in suggesting a role for social fragmentation and isola- the incidence of schizophrenia that could not be tion. These studies link to classic sociological theories accounted for by individual-level variables, such as that have located the origins of intrapsychic distress in age, sex and ethnicity [26,40]. Further, around 20% social organization and structure, notably the work of of the neighbourhood variation in incidence was Durkheim and his theories of suicide. Here, the con- accounted for by voter turnout (a proxy measure for cept of anomie (i.e. a shorthand for contexts in which social capital) and ethnic fragmentation (segrega- social norms have broken down) may be useful in tion), independent of area-level deprivation [26,40]. characterizing socially fragmented areas and under- Taking this one step further, Kirkbride et al. [41] standing the feelings of disaffection they engen- conducted another study in the same area of south- der [33–35]. This shifts attention from isolation per east London to investigate more directly the relation- se (or ‘social defeat’, see below) to an altogether more ship between low levels of ‘social capital’ (defined fraught state of antagonism between the individual primarily as social cohesion and trust and measured and his/her environment – one in which suspicious- using postal questionnaires) and rates of schizophre- ness and paranoia become understandable and (per- nia. Intriguingly, what they found was a U-shaped haps) even adaptive responses. relationship, such that incidence rates were highest in In an effort to gain some traction on the urban areas with low (incidence rate ratio (IRR) of 2.0) and effect, recent research has revisited these early stu- with high (IRR 2.5) levels of cohesion and trust, dies and sought to investigate the impact of aspects of compared with areas with medium levels of cohesion the wider social environment – notably social frag- and trust, independent of age, sex, ethnicity and other mentation, deprivation and cohesion (albeit with neighbourhood characteristics (e.g. ethnic fragmen- limited theorizing on how such contextual factors tation and deprivation). The authors concluded that impact on, and become manifest in, individuals). In areas with low social capital may compound expo- their review, Allardyce and Boydell [36] identified 13 sure to social stress, whereas areas with high social relevant studies. In one of the earliest, Boydell capital may be problematic for those who are ex- et al. [37] found that, in the most socioeconomically cluded, for whatever reasons, from accessing the

THE SOCIODEVELOPMENTAL ORIGINS OF PSYCHOSIS 197 available social capital. Tentatively, what these find- cerning social fragmentation and social capital need to ings suggest is that it is not so much deprivation per be investigated more directly and replicated, ideally in se but resulting area-level fragmentation, loss of prospective designs that can separate cause and effect cohesion and trust, and isolation that is associated (the feasibility of such studies notwithstanding). with an increased incidence of schizophrenia. This There is, moreover, some inconsistency in the litera- association, however, is not necessarily straightfor- ture. For example, in a study conducted in Australia, ward, and (as the findings from the latter study by McGrath et al. [43] failed to find an association Kirkbride et al. [41] suggest) may be modified by between urbanicity and schizophrenia, a finding that individual access to area-level social capital and perhaps hints at a possible protective role for climate. support. A study conducted in the Netherlands using Other studies have found that area-level effects (e.g. data from the Maastricht Mental Heath Case Register neighbourhood deprivation) are largely accounted for found that those who were single had a higher risk of by individual-level risk factors [2]. This may reflect schizophrenia if they lived in areas with fewer single the methodological challenges inherent in disentan- people (relative to areas with more single peo- gling effects that operate at individual and area levels. ple) [42]. The authors speculate that feelings of What it further reminds us is that at least some of the isolation may be exacerbated if single people live observed variation in incidence between urban and in areas where the majority live with a partner, an rural areas may be a consequence of the aggregation in interpretation that again allows a role for isolation or, cities of individual-level exposures (e.g. trauma, possibly, anomie. stressful life events, substance use). The thorny issue There remain caveats. The data are still relatively of quite how social contexts impact on individuals thin and largely cross-sectional, with limited account in such a way as to predispose to, or increase risk having been taken of potential confounders, notably a for, schizophrenia and other psychoses is discussed family history of psychosis. The speculations con- further below. 15.4 MIGRATION AND ETHNICITY: (1) EVIDENCE There is now strong evidence from studies in the UK, These findings have proved controversial, a key the Netherlands, Sweden, Denmark, Australia and the question being whether the reported rates are valid or a US that rates of schizophrenia and other psychoses are function of methodological artefact and misdiagno- elevated in migrant and minority ethnic popula- sis [48]. In response, recent studies have used more tions [44]. In their recent meta-analysis of popula- robust methodologies, with comprehensive case- tion-based incidence studies of schizophrenia, finding procedures, accurate population denominator Cantor-Graae and Selten [17] found (from a total of data and standardized diagnostic procedures with 50 effect sizes) a mean weighted relative risk (RR) for raters blind to the ethnicity of individuals being developing schizophrenia among migrant groups of assessed. Such studies continue to report very high 2.9 compared with indigenous or host populations. rates [48]. For example, the ÆSOP study (introduced The relative risk was particularly high in migrants above) found that the incidence of all psychoses was from developing countries (RR 3.3), in second- over 6 times higher in black Caribbean and over 4 generation migrants (RR 4.5) and in migrants from times higher in black African populations in the UK, countries where the majority population is black (RR compared with white British [47]. These findings 4.8). This latter finding is intriguing, as we know that it held across all three centres (south-east London, is more visible migrant and minority groups who Nottingham, Bristol), for men and women, and across experience more racism and discrimination (see all age groups [47]. Studies that have attempted to below). Since the publication of this review, there assess directly the potential role of misdiagnosis have have been further studies in Europe and the US failed to find clear evidence that this is sufficient to confirming this general pattern [44–47]. account for the reported rates [49].

198 SOCIAL DETERMINANTS The degree of increased risk, however, is not con- populations, this appeared to be evident for women sistent across migrant and minority ethnic groups – as only. In the Netherlands, the incidence appears to be hinted at in the meta-analysis conducted by Cantor- highest in Moroccan migrants [50]. The reasons for Graae and Selten [17]. For example, in the ÆSOP these variations are unclear; however, speculation has study, the incidence of psychosis was raised to a focused on differential exposure to discrimination much lesser extent in Asian and other white (i.e. non- and the buffering effects of family supports and social British) populations [47]. In a more recent study in networks. More generally, this leads to the question east London, the incidence was again found to be of why rates of schizophrenia and other psychoses are higher in most migrant and minority ethnic elevated (albeit to varying degrees) in migrant groups [45]. However, in Pakistani and Bangladeshi populations. 15.5 MIGRATION AND ETHNICITY: (2) DISCRIMINATION, DISADVANTAGE AND ANOMIE There are a number of recent reviews that have ad- above linking aspects of the wider social environment dressedthisquestion[44,48,51–53].Tocutalongstory with the risk of psychosis, and again hints at a possible short,allhaveconvergedontheconclusionthatthehigh role for isolation and discrimination in areas of low rates are likely to be a consequence of exposure, ethnic density or, alternatively, at a protective effect of broadly defined, to adverse social contexts and experi- living in areas of high ethnic density, which may ences.Inpart,thisconclusionderives (a) from a lack of mitigate the impact of discrimination, isolation and any evidence that genetic or known biological risk disadvantage. That said, the processes underpinning indicators (e.g. obstetric complications, neurological this repeated finding have not been directly investi- soft signs, minor physical anomalies) contribute to gated. What is perhaps most notable, however, is that increased risk in these populations [48] and (b) from such a social patterning of risk defies ready explana- the observation that, as rates are even more elevated in tion, or even plausible speculation, in terms of social second-generation migrants, selective migration is drift or known biological risk factors [48]. unlikely to play a significant role [17]. Where studies There are a small number of studies that have have been conducted, all show that rates of psychosis attempted to investigate directly the relationship be- are not elevated in the countries from which these tween discrimination and psychosis. For example, populations have migrated [54–56], which suggests Veling, Hoek and Mackenbach [59] used general population differences in genetic risk are unlikely to population data on perceptions of discrimination in explain the repeated findings. More than this, there is the Netherlands to order migrant and minority ethnic now direct evidence implicating aspects of the social groups according to levels of exposure to discrimina- environment and experience over the life course, tion. Using this index of discrimination, they found much of it from studies conducted in the UK and the clear evidence of a linear relationship with schizo- Netherlands. phrenia, such that the incidence increased in line with Perhaps most intriguing is the now replicated find- level of exposure to discrimination, the highest rates ing that the relative risk of schizophrenia increases as being in those with highest levels of perceived dis- migrant and minority ethnic groups form a decreasing crimination (i.e. Moroccan, with IRR 4.82). More proportion of local populations. In other words, the generally, an analysis of data from the Netherlands less ethnically dense an area, the higher the rates of Mental Health Survey and Incidence Study found psychosis (see Figure 15.1). This pattern was origin- tentative evidence that experiences of discrimination ally reported by Faris and Dunham and has now been at baseline predicted the occurrence of paranoid idea- reported for the black Caribbean population in the tion at three-year follow-up in otherwise healthy UK [40,57] and for migrant populations in the Nether- individuals. This was not, however, focused specifi- lands [58]. This relates back to research reviewed cally on racial discrimination and the number of

THE SOCIODEVELOPMENTAL ORIGINS OF PSYCHOSIS 199 High % minority population Medium % minority Low % minority population population RR 2.4† RR 3.6† RR 4.4† RR 3.8‡ RR 2.1‡ RR 6.5‡ RR 1.3* - RR 2.4* RR, Rate ratio (Migrant or ethnic group vs. White) †Boydell et al, 2001; ‡Kirkbride et al, 2007; *Veling et al, 2008 Figure 15.1 Rate ratios for schizophrenia by ethnic density in recent studies (Reproduced from C. Morgan and G. Hutchinson (2009) The social determinants of psychosis in migrant and ethnic minority populations: a public health tragedy, Psychological Medicine,p. 2) individuals who reported discrimination and paranoid population-based controls, after taking account of ideation was very small (n ¼ 7). This remains an area possible confounders. Most importantly, while the for which there is limited direct evidence and, as a effect of separation on the odds of psychosis was consequence, inferences drawn from indirect evi- similar for white British and black Caribbean indivi- dence (e.g. highest rate in black groups, ethnic density duals, separations were much more common for black effect), while intuitively appealing, need to be con- Caribbeans (e.g. 31% black Caribbean controls versus sidered with appropriate caution. 18% white British controls, p ¼ 0.03) [60]. This pat- There is some evidence that exposure to social tern (i.e. a similar effect size in both groups, a higher adversity, broadly defined, over the life course may prevalence in the black Caribbean population) was contributeto the high rates. This comes primarily from also evident for a number of indicators of adult social studies of the UK black Caribbean population. For disadvantage and isolation (both recent and long- example, a series of findings from the ÆSOP study standing), including unemployment, housing instabil- link indicators of childhood and adult disadvantage ity and limited social networks [61]. These findings with psychosis in general and with high rates in the are important as they are consistent with the high rates Caribbean population in particular [25,60,61]. To take being a function of a greater prevalence of social risk one example, this study found that long-term separa- factors in the black Caribbean population [61]. Re- tion from a parent before the age of 16 (a marker of turning to the previous paragraph, high levels of social childhood disadvantage, indexing exposure to intra- disadvantage in migrant and minority ethnic groups familial conflict, financial hardship and housing in- may reflect a subtle racism institutionalized in social stability) was around three times more common in structures that constrain access to opportunities and those with a first-episode of psychosis compared with resources [48].

200 SOCIAL DETERMINANTS There is, of course, a need for caution here. The ables index exposure to contexts and experiences that kinds of individual-level markers of adversity that contribute to higher rates in the black Caribbean have been considered are at best crude proxies for group. In other words, there is a convergence of what complex social experiences that occur over time and in might be termed circumstantial evidence on the inter- specific contexts. In relation to adult indicators at pretation that the high rates are a function of cumu- least, the problem of reverse causality is not overcome lative adversity over the life course, in interaction with byconsideration of linear relationships. These data are aspects of the wider environment (e.g. ethnic density) nonetheless intriguing, because they pattern in pre- – perhaps leading, for some, to a chronic state of cisely the way that would be expected if these vari- dissonance with the social environment (i.e. anomie). 15.6 SOCIAL ADVERSITY ACROSS THE LIFE COURSE: (1) CHILDHOOD The model of schizophrenia (and other psychoses) as a care for their relatives, without which mental health developmental disorder suggests that environmental services would be stretched to breaking point and the factors that operate early in life, from pregnancy lives of many sufferers would be considerably poorer. through infancy, childhood and adolescence, may be The mistake in the early theories was to suppose that particularly important. As already noted, the impact of most if not all cases of schizophrenia could be traced the urban environment on risk appears to accumulate back to poor parenting. What is lost in the under- throughout childhood, perhaps as a consequence of standable backlash is that – unfortunate as it indeed is repeated exposure to the unspecified ‘toxins’ (social – negative home environments can be sources of and physical) that are encountered in cities. The considerable stress for children; we know that various negative and enduring impact of adversities experi- forms of childhood adversity (e.g. parental separation, enced early in life are evident for a host of problematic intrafamilial conflict, and so on) can increase risk of outcomes (offending, substance use, all forms of negative outcomes in adulthood – including depres- mental disorder). It is no surprise, from these observa- sion, anxiety, self-harm, personality disorder and, tions, that childhood adversity in various forms (in- possibly, psychosis. trafamilial problems, traumas, bullying) has attracted A number of recent reviews have addressed the attention as a possible contributory factor in the question of whether childhood trauma (specifically, aetiology of schizophrenia and other psychoses. childhood sexual and physical abuse) is a causal factor Perhaps the most controversial area is family inter- in the aetiology of psychosis [3,11,69,70]. There are a actions and communication [62,63]. The idea that large number of studies that have reported high rates patterns of communication within families contribute of childhood abuse in samples of individuals with a to schizophrenia raises the spectre of the serious mental disorder (e.g. in a recent review, an ‘schizophrenogenic mother’ [64] and of families average of 50% for men and 50% for women) [3]. being blamed for causing serious mental disorder in However, as these are highly select samples of pre- their children [65,66]. In their early review, Hirsch and valent cases, they do not provide evidence concerning Leff [67] concluded that the evidence to support such the relationship with onset. A series of more recent theories was weak and methodologically limited. The population-based studies have reported marked asso- highly politicized nature of this topic has ensured that ciations between childhood abuse and trauma (var- aspects of family life have since been largely taboo iously defined and measured) and risk of psychosis or when considering the causes of schizophrenia and psychosis-like experiences. For example, in their other psychoses, and it is mainly in relation to course analysis of data on 4045 subjects aged 18–64 drawn and outcome that family interactions have been stu- from the Netherlands Mental Health Survey and In- died (through the concept of expressed emotion [68]). cidence Study, Janssen et al. [71] found that those who In these debates, it is all too frequently forgotten that had experienced emotional, physical or sexual abuse, families provide a tremendous amount of informal or neglect before the age of 16 were more likely to

THE SOCIODEVELOPMENTAL ORIGINS OF PSYCHOSIS 201 report experiencing psychotic symptoms during a British National Survey of Psychiatric Morbidity, three-year follow-up period. The effect was strongest found that those who met criteria for a definite or for the most severe psychosis groups, and held after probable psychotic disorder (n ¼ 60) were over 15 adjusting for a range of potential confounding vari- times more likely to have been sexually abused at ables (e.g. need for care-level psychosis: Adj. OR 7.3). some point in their lifetime. When the interrelation- However, the number of subjects with psychotic ship between other negative life events and level of symptoms was very small, particularly those with the depression were controlled, the odds ratio was mark- most severe symptoms (n ¼ 7). Other studies have edly reduced, though still significant (Adj. OR 2.9). focused on specific experiences or symptoms. Whit- However, the measure of sexual abuse was crude (a field et al. [72], in a large population-based cross- single question), no account was taken of timing, sectional study (n ¼ 17 337), found that those who duration or severity of abuse, and childhood and adult reported a history of hallucinations (assessed using a exposures were not distinguished. single question) were more likely to have been both There is, moreover,evidence that exposureto multi- physically (Adj. OR 1.7) and sexually (Adj. OR 1.7) ple traumas is associated with a linear increase in risk abused during childhood. This is consistent with other for psychosis. Using data from both the National studies that suggest a specific link between hallucina- Comorbidity Survey and the British Psychiatric Mor- tions and prior abuse [73,74]. The evidence for links bidity Survey, Shevlin et al. [79] found that risk of with other symptoms (e.g. delusions) is less clear. psychosis increased in dose-response fashion with the Janssen et al. [71], for example, reported higher rates number of traumas experienced. Notably, the kinds of of both hallucinations and delusional ideation in those traumatic events that were most strongly associated who had experienced childhood abuse, but others have with psychosis were those involving violence and found no association between early trauma and threat, as in the studies by Bebbington et al. [78] and delusions [75]. Spauwen et al. [77] (e.g. molestation, physical abuse, It is not only abuse that has been linked with serious injury or assault, and violence in the home) – psychosis or psychosis-like experiences. In a study or what Tirril Harris has termed intrusive life events. of adolescents (aged 12–15 years) in Ireland, Kelleher The finding of an effect for cumulative trauma is et al. [76] found that those who reported one or more particularly interesting when set alongside the only psychotic symptom were more likely to report past study in which the occurrence of sexual abuse has exposure to physical abuse, domestic violence and been determined using contemporaneous records. In bullying (as victim and/or perpetrator), suggesting a this, Spataro et al. [80] compared rates of subsequent broader range of adverse experiences may be relevant. hospital admissions in those who had been sexually Once more, however, this study was relatively small, abused before the age of 16 (n ¼ 1612), according to with only 14 (6.6%) of the sample reporting a psy- official records, with admission rates in a large popu- chotic symptom. Nevertheless, other studies have lation-based control sample (n ¼ 3 139 745). They found associations with a wider range of traumas (not found no association between child sexual abuse and necessarily restricted to childhood). Spauwen later admission to hospital with a diagnosis of schizo- et al. [77], using data on 2524 subjects aged 14–24 phrenia (RR 1.2). There are of course a number of from the Early Developmental Stages of Psycho- methodological problems (e.g. most cases of abuse do pathology Study, found that the experience of any not come to the attention of services, so that the lifetime trauma (from a list of nine events) was comparison group is likely to have included signifi- associated with the development of three or more (but cant numbers who were exposed) and strengths (the not fewer) psychotic symptoms during an average large sample, contemporaneous records of abuse), but follow-up period of 42 months (Adj. OR 1.9). The the most intriguing possibility is that intervention by trauma exerting the strongest independent effect was child protection services (which all received) miti- natural catastrophe (Adj. OR 15.1) followed by phy- gated risk. Just as prolonged and repeated exposure to sical threat (Adj. OR 2.1). Bebbington et al. [78], cumulative adversity over time may compound risk, using data on 8580 subjects aged 16–74 from the so intervention may reduce risk.

202 SOCIAL DETERMINANTS This, of course, remains speculative, and recent associated with an increased risk of later psychosis. reviews have highlighted the many methodological This links back to research noted above, and other limitations that characterize this research (e.g. selec- studies, that have found an association with parental tion and information bias, uncontrolled confounding, separation [60,83]. What has not been fully resolved is crude measurement of abuse), all of which urges the question of whether these associations reflect caution [3,70]. The most robust research has been on gene–environment correlations, in which exposure to psychosis-like experiences in general population sam- childhood adversity is a consequence of familial ples, which have an indeterminate relationship with liability to psychosis. These more recent studies have psychotic disorder. It may be that many positive adjusted for markers of genetic risk [60,83], but responses to psychosis screening questionnaires (as doubts continue to be expressed. In addition, evidence are used in these studies) reflect dissociative and/or on parental social class is inconsistent (e.g. see post-traumatic experiences. In one of the few studies References [84] and [85]), though this may be more to investigate individuals meeting criteria for a first- a reflection of the limitations of such a crude proxy episode psychotic disorder, a relationship with sexual variable, which has been inconsistently defined across and physical abuse was found in women only, and the studies. Where experiences have been more directly, effect size was generally weaker than in other studies though still somewhat crudely, considered, the evi- (i.e. odds ratios of around 2) [81]. dence is stronger and more consistent. Intriguing There is also evidence that the risk of psychosis is additional evidence comes from the study by Pedersen associated with broader indicators of social adversity and Mortensen [28], discussed above. In this, they also in childhood; it is in fact difficult to separate out the found that geographical mobility (i.e. movement out- effects of abuse and indicators of socioeconomic side the prior municipality of residence) during child- disadvantage, as they are often correlated. In a study hood was associated with an increased risk of schizo- using Swedish register data on 2.1 million individuals, phrenia. Such movements almost certainly involve a Wicks et al. [82] found that a range of indicators of change of school, raising the possibility that some of childhood socioeconomic adversity (i.e. rented the difficulties in forming relationships and a tendency accommodation, low socioeconomic status, single- to social isolation seen in those who later develop parent households, parental unemployment and schizophrenia reflects socio- rather than neurodeve- households receiving social welfare benefits) were lopmental processes. 15.7 SOCIAL ADVERSITY ACROSS THE LIFE COURSE: (2) ADULTHOOD There are a reasonable number of studies that suggest relationship between life events and first-episode significant stressful life events are associated with psychosis. There have been a small number of studies relapse in those with schizophrenia or other psy- that have considered the relationship between life choses [86,87]. There is, however, uncertainty about events and psychosis-like experiences, which show the period over which life events are rele- an excess of such events in the preceding six months vant [86,88,89] (i.e. proximal to relapse or through among those who report psychosis-like experi- a cumulative impact over time) and, while these ences [90,91]. More broadly, and as noted above, studies indicate a heightened sensitivity to stress for there is evidence that, on a series of fairly crude those with psychosis, they do not help clarify whether indicators, social disadvantage is more pronounced this is relevant to the onset of disorder. In the literature, in those with a first episode of psychosis compared in fact, there is a confusing use of terminology, with with population-based controls [61]. onset and relapse both being used, when what is meant The scepticism noted above is, however, more is onset of relapse rather than onset of the illness (i.e. pronounced for data on adult adversity and stress. the first episode) [86]. As far as we are aware, there has On the one hand, this is because of the apparent not been a single study that has investigated the decline in social function that occurs for many prior

THE SOCIODEVELOPMENTAL ORIGINS OF PSYCHOSIS 203 to the onset of overt positive psychotic symp- perspective [96]. They used Experience Sampling, toms [92], which may in turn increase the risk of which assesses the occurrence of daily stresses and exposure to life events (e.g. job loss, relationship hassles in the flow of everyday life, to investigate the break-up) and disadvantaged social circumstances. impact of daily stresses and emotional responses on On the other, there is evidence that likelihood of the likelihood of the occurrence or exacerbation of exposure to stressful events and adverse circum- psychosis (or psychosis-like) experiences. In one stances is genetically mediated [93]. This raises the study of general practice attenders, for example, possibility that associations with life events (includ- exposure to childhood trauma was found to mediate ing childhood life events) are a product, in part at the relationship between daily stresses and emotional least, of gene–environment correlations, where genes responses, with the greatest increases in negative that increase risk for exposure to stressful environ- affect being seen in those with a history of trau- ments also increase risk of psychosis [93]. Conse- ma [97]. This provides some evidence that early quently, the decline in social function (and indeed experiences sensitize individuals and increase nega- early markers of problematic development, such as tive responsivity to stress – albeit not in relation to solitary play and isolation) and increased exposure to psychosis. In a further study, Myin-Germeys stress and trauma is usually interpreted as the ex- et al. [98] found that daily stress had a direct impact pression of underlying genetic and neurological ab- on what they term ‘moment to moment fluctuations’ normalities that predispose to schizophrenia [94]. in psychotic experiences, with the intensity of symp- However, it is equally plausible that these develop- toms increasing in line with the experience of hassles mental precursors are a consequence of problematic and stressors. This is a particularly encouraging and social development, stemming from the kinds of innovative line of research, which (using quasi- early adverse experiences discussed above [28]. Of experimental designs) overcomes many limitations course, disentangling these alternatives is methodo- of traditional observational social epidemiology. logically daunting, and it may be that both processes There is a need for further studies to replicate and are involved to varying degrees in different indivi- extend these findings, but the initial results are con- duals. For example, even if the tendency to isolation sistent with the proposition that childhood adversity is initially driven by neurological and cognitive contributes to creating an enduring liability to nega- deficits, the exacerbation of this through exposure tive emotional responses and, in some, psychosis-like to familial disadvantage, frequent moves and/or abu- experiences following exposure to daily hassles and sive relationships may compound risk and push in- stress. By extension, psychotic disorder may emerge dividuals along a largely sociodevelopmental path- in those exposed to early adversity as a consequence way to psychosis – sociodevelopmental in the sense of ongoing exposure to problematic social contexts that intervention to modify this social trajectory, even and experiences in adulthood. in the biologically vulnerable, may interrupt the In sum, for all there are methodological limitations pathway to psychosis. In other words, the pessimism in the research to date (which to a degree reflect the engendered by a predominantly neurodevelopmental inherent difficulties in studying social factors and the perspective (‘doomed from the womb’) is turned onset of psychosis), there is a growing and converging on its head; framed in terms of sociodevelopment, body of evidence that adverse social contexts and the trajectory can be altered – psychosis, at least experiences over the life course are relevant to the for some, is not inevitable. Further, this perspective aetiology of psychosis – contexts and experiences, fits the current evidence on migrant and minority moreover, that may create a disjunction between the ethnic populations, where it seems to be social individual, others and the wider society (or what contexts and experiences that make the difference, sociologists have conceptualized as anomie [35]). To that underpin a significant increased risk of paraphrase Jones et al. [94], early disadvantage can set psychosis [48,95]. in train a cascade of problematic social development, A series of innovative studies conducted by Myin- one consequence of which may be an increased risk Germeys and van Os provide some support for this of psychosis.

204 SOCIAL DETERMINANTS 15.8 MECHANISMS: (1) SENSITIZATION In her poetic account of rural Irish life and the possible sical or social), perhaps through a final common links between socialization processes, isolation, pathway of dysregulation of the mesolimbic dopami- anomie and schizophenia, Scheper-Hughes [99] sug- nergic system [100,101]. With regard to socioenvir- gested that, ‘(w)rit large it [the development of schi- onmental insults, much of the evidence above suggests zophrenia] is the translation of social ills into private that it is cumulative exposure to adverse social con- troubles’ (p. 305). In our somewhat more mechanical texts and experiences over time, particularly during age, this begs the question of how, through what development, that increases risk of psychosis. This mechanisms, do social ills (e.g. fragmented environ- suggests that such exposures may operate through a ments, trauma, etc.) impact on individuals (or become process of sensitization, in which repeated exposure embodied) to increase risk of such a rare set of private progressively increases behavioural, psychological troubles (voices, delusions, thought disturbance, and biological responses to further stresses (even blunted affect, and so on)? Ironically, it is advances where these are relatively mild, e.g. daily hassles), in genetics, neuroscience and psychology that suggest with a consequent increased risk of psychosis-like a number of plausible mechanisms, which (crucially) experiences and, in some, psychotic disorder [6,8]. are consistent with what is known more generally This is also consistent with evidence that much more about the genetic and biological basis of schizophre- common psychosis-like experiences are associated nia and other psychoses. with similar, but fewer and milder, social adversi- One increasingly popular model of onset suggests ties [22]. In other words, exposure to cumulative that multiple genes of small effect create an enduring adversity over time may push some individuals along liability to psychotic disorder that becomes manifest, a continuum, through affective responses and psycho- in the presence of further environmental insults (phy- sis-like experiences, to psychotic disorder. 15.9 MECHANISMS: (2) DOPAMINE AND THE HYPOTHALAMIC–PITUITARY–ADRENAL AXIS Importantly, there is emerging (indirect) evidence that isolation of subordinated rodents is associated with such exposures may plausibly link to psychosis persistence of changes in the dopaminergic sys- through sensitization of the dopaminergic sys- tem [107]. Similar findings have been reported in tem [6,101] and dysregulation of the hypothalamic– studies of macaque monkeys [108]. There is some pituitary–adrenal (HPA) axis [102] (both of which are tentativeevidence ofsimilarprocessesinhumansfrom implicated in psychosis [100,103]). To begin with, asinglestudythatfound,inasampleof10students,that there is some evidence that stress is associated with the effect of psychosocial stress on mesolimbic dopa- dopamine release in humans and animals [8]. More mine release was dependent on early life maternal specifically, there is accumulating evidence from ani- care [109]. Further, elevated dopamine metabolism mal studies that negative and threatening events (e.g. has been found in girls who have been sexually abused maternal deprivation in neonatal rats [104] and ‘social compared with nonabused controls [110]. defeat’ in mice [105]) can produce dopaminergic Interest in the possible role of the HPA axis is more hyperactivity in the mesocorticolimbic system, and recent, and stems from evidence of dysregulation of thatprolongedexposuretosuchaversiveenvironments this system in those with psychosis [8]. In the most can lead to sensitization of this system. For example, recent formulation of their neural-stress–diathesis rodents exposed to ‘social defeat’ or subordination model of schizophrenia, Walker, Mittal and Tess- show increased levels of dopamine in the nucleus ner [102] speculate that dysregulation of the HPA acumbens and preforontal cortex [106]; subsequent axis may ‘trigger a cascade of events resulting in

THE SOCIODEVELOPMENTAL ORIGINS OF PSYCHOSIS 205 neural circuit dysfunction, including alterations in As appealing as this hypothesis is, it is doubtful that dopamine signalling’ (van Winkel and Stefanis [8], the range of adversities that appear relevant over the p. 1097). Of particular relevance here is further evi- life course can be so readily collapsed into a single dence that exposure to stress, particularly childhood exposure. In humans, moreover, the kinds of experi- trauma, is associated with alterations in HPA axis ences that might comprise social defeat (entrapment, function [111,112]. These findings are the basis for loss, humiliation) have been more consistently linked the traumagenic model of psychosis proposed by Read with depression. In contrast, exposure to intrusive and and colleagues [69,113]. anxiety-provoking events and contexts may be parti- These mechanisms of course remain largely hy- cularly relevant to psychosis – not a state of defeat and pothetical, and there is a need for caution. For exam- learned helplessness, but a state of disillusion with and ple, on the basis of the evidence from animal studies heightened sensitivity to the external environment and above, Jean-Paul Selten and Elizabeth Cantor- its perceived hazards. This reminds us that, while Graae [114] have hypothesized that it is exposure to animal research can provide clues, we need to be chronic discrimination and ‘social defeat’ that is the cautious in borrowing terminology and applying find- unifying experience underpinning the high rates of ings to humans. There is much that can be lost in psychosis in migrant and minority ethnic populations. translation. 15.10 MECHANISMS: (3) GENE–ENVIRONMENT INTERACTION It remains that not all individuals exposed even to exposure to stress in humans early in life may impact prolonged and cumulative traumas and adversity de- on risk for later disorder by altering gene expression. velop a psychotic disorder; vulnerability varies. This Where gene–environment interactions have been susceptibility may be linked to genotype [7,8]. investigated, these have been based on statistical There has been considerable interest in potential models of interaction between a putative environmen- gene–environment interactions in psychosis, with tal factor and a proxy genetic marker or specific some interesting initial findings [7]. Two types of genotype. For example, Caspi et al. [117], in an gene–environment interaction can be specified: (1) in analysis of data from the Dunedin longitudinal study which genotype moderates sensitivity to the environ- of a representative birth cohort, showed that a func- ment, such that individuals differ in their response to tional polymorphism in the promoter region of the environmental exposure dependent on genotype; and serotonin transporter (5-HT T) gene moderated the (2) in which the environment impacts on gene expres- influence of stressful life events on depression. In- sion, through effects on DNA sequence (i.e. causing dividuals with one or two copies of the short allele of de novo mutations) and methylation (i.e. causing the 5-HT T promoter polymorphism exhibited more altered gene expression through epimutations) [7]. depressive symptoms, diagnosable depression and In relation to this latter possibility, there is no direct suicidality when exposed to stressful life events than evidence for psychosis, but there is research to individuals homozygous for the long allele. This may suggest this is a plausible potential mechanism that help explain why only some of those exposed to could, in part at least, account for the impact of stressful life events develop depressive symptoms social experience on risk of psychosis. A series of (i.e. the first type of interaction noted above, where animal experiments, for example, have found that responses to stress are moderated by genotype). There maternal behaviour in rats (pup licking and grooming; is similar evidence concerning cannabis use and the arched-back nursing) can affect stress responses in catechol-O-methyl-transferese (COMT) gene, such offspring through DNA methylation of relevant that risk of psychosis for those who use cannabis genes [115,116], raising the intriguing possibility that appears to vary as a function of COMT– risk is highest

206 SOCIAL DETERMINANTS for those with one or two copies of the Valine (Val) dependent on pre-existing indicators of genetic risk allele compared with those homozygous for the for psychosis [121]. Methionine (Met) allele [118]. The evidence for so- This line of research is not, however, without cioenvironmental exposures is limited. In one of the limitations. A recent meta-analysis of studies of the few studies to investigate this, Stefanis et al. [119] serotonin transporter gene (5-HT T), life events and found that significant life stress (in this case, induction depression failed to find evidence of gene–environ- into the Greek army) was associated with the occur- ment interaction [122], and others have questioned rence of psychotic symptoms; this association was whether such studies can provide additional insights strongest in those with either the Val/Met or Val/Val (Stanley Zammit’s ‘Genes  environment’ presenta- allele, as opposed to the Met/Met allele. What is tion at the UK Mental Health Research Network particularly noteworthy here is that COMT encodes Conference, Nottingham, 2009). A key considera- a key enzyme that metabolizes dopamine in the tion here is that statistical interaction does not frontal cortex. Additional, but less specific, evidence necessarily equate with biological interaction, and comes from family and adoption studies that have the processes underlying the interaction effects so found adopted-away offspring of mothers with psy- far observed are unclear. Evidently, the most sig- chosis are at highest risk when brought up in home nificant advances are likely to be achieved where the contexts that score highly on ‘critical/conflictual’, underlying processes can be identified, which is ‘constricted’ and ‘boundary problem’ scales [62,120]. what makes interactions with COMT (because of Further, there is evidence from population-based re- its role in dopamine metabolism) a potentially im- search in the Netherlands that the urban effect is portant line of investigation. 15.11 MECHANISMS: (4) COGNITION AND EMOTION In recent years, there has been a rapid development of affective biases may develop particularly in those cognitive models of psychosis that tend to focus on who are susceptible due to underlying genetic or specific symptoms (e.g. hallucinations, delusions) neurodevelopmental factors. Interestingly, there is rather than diagnostic categories and which posit a tentative evidence for such models in relation to black key role for affective and cognitive biases in symptom Caribbeans in the UK [125] and, related to this, there is formation [10,12,123,124]. In these models, initial some evidence that delusions of persecution and vulnerability (which may stem from genetic, neuro- reference are more common in the black Caribbean developmental or social processes) is expressed and other minority ethnic groups [126]. In introducing through a number of affective and cognitive biases cognitive schema, these models provide a meaningful (e.g. a tendency to jump to conclusions, an external account of how hostile environments and resulting locus of control, deficits in source monitoring); these anomie may link to, say, delusion formation. In short, tendencies in turn increase the risk of symptom for- these models are genuinely biopsychosocial, provid- mation in response to subsequent stresses and trou- ing explanatory mechanisms at the level of symptom bling experiences (e.g. exposure to threat may result in formation that incorporate findings from genetics, tendencies to attribute misfortune to external causes neuroscience and social epidemiology [11,12]. One and hypersensitivity to the environment, which if such model is shown in Figure 15.2 [11]. continued over time may increase the risk of suspi- To reiterate, on the basis of the evidence reviewed ciousness, paranoia and, ultimately, delusion forma- above, there is increasing evidence that exposure to tion). There is now a significant body of evidence to adverse social contexts and experiences over the life support this broad model in relation to a number of course is crucial both in conferring vulnerability and symptoms [123]. Such models are compatible with in the transformation of vulnerability into disorder. In neurobiological research [9], in that cognitive and other words, aberrant social development arising from

THE SOCIODEVELOPMENTAL ORIGINS OF PSYCHOSIS 207 Intrusive Event Self-protective social withdrawal Lack normalising social experiences Behavioural Cannabis & other “mind altering” exposure Faulty Source Event Salience Hypervigilance Monitoring Psychological ‘Hostile World’ External locus External attribution of control bias Biological HPA axis / Dopamine Dysregulation Symptom Formation Figure 15.2 Childhood trauma and psychosis – an integrated model (Reproduced from Fisher H, Craig T. Childhood adversity and psychosis. In: Morgan C, ed. Society and Psychosis. Cambridge: Cambridge University Press; 2008:95–111.) cumulative adversities (that impact on brain develop- one (but by no means the only) critical pathway to the ment and cognitive and affective processes) may be formation and expression of psychosis. 15.12 SOCIAL DISORDER Our focus here has been on aetiology, on the emer- to the development and persistence of psychosis; in gence of psychotic disorder. However, just as social many respects it is a social (and mental) disorder, an ills, to borrow Scheper-Hughes’ [99] words, find intense form of social suffering [128], affecting the expression in the private troubles of psychosis, so susceptible, that is felt in the challenge, both before this private turmoil reverberates back outwards to and after onset, of navigating and engaging with a shape the local social and moral worlds of the frequently hostile social world – a hostility more sufferer – both by affecting the individual’s capacity frequently encountered as a consequence of living in to interact with the social environment and through urban centres, of being an outsider, of living in the negative and stigmatizing responses of poverty. Such a formulation has significant implica- others [127]. There is substantial evidence that con- tions for prevention and intervention; it focuses tinued exposure to social difficulties (limited social attention, broadly, on modifying aspects of the social networks, unemployment, problematic family rela- environment (especially for those groups most at tionships, stigma and discrimination) worsens the risk) and on intervening to promote social reintegra- course and outcome of disorder. This evidence is tion for those with a disorder. This is not to negate the reviewed extensively elsewhere in this book (see utility of psychotropic medication and symptom re- Chapters 26 and 31). The crucial point here is that duction; it is to highlight the centrality of individuals’ the social environment, broadly construed, is central social lives in recovery.

208 SOCIAL DETERMINANTS 15.13 CONCLUSION There remain those who are sceptical that socioenvir- conditional on genetic risk. Schizophrenia Bulletin, onmental factors are of any significance in the aetiol- 31 (4), 795–799. ogy of psychosis (e.g. see References [15] and [129]). 3. Morgan, C. and Fisher, H. (2007) Environment and Some of the associations observed between social schizophrenia: environmental factors in schizophre- factors and psychosis may be a function of the fact nia: childhood trauma – a critical review. Schizophre- that they increase exposure to other factors that impact nia Bulletin, 33 (1), 3–10. directly on risk (e.g. substance use [101]). At this 4. Morgan, C., McKenzie, K. and Fearon, P. (2008) Society stage, it is reasonable to acknowledge that there is still and Psychosis, Cambridge University Press, Cambridge. sufficient uncertainty and ambiguity in the available 5. Kendler,K. S.,Neale, M. C.,Kessler,R. C. etal. (1992) evidence that any all-embracing or dogmatic claims Childhood parental loss and adult psychopathology in women – a twin study perspective. Archives of General concerning the origins of psychosis will be premature Psychiatry, 49 (2), 109–116. and ultimately unhelpful. The model we propose here, 6. Collip, D., Myin-Germeys, I. and Van Os, J. (2008) based on a specific reading of the literature – of, Does the concept of ‘sensitization’ provide a plausible among others, a sociodevelopmental trajectory – is mechanism for the putative link between the environ- consequently tentative, involves speculation beyond ment and schizophrenia? Schizophrenia Bulletin, 34 the evidence and no doubt requires much further (2), 220–225. empirical testing. Herein lies the challenge. We began 7. van Os, J., Rutten, B. P. and Poulton, R. (2008) this chapter by noting two specific problems that have Gene–environment interactions in schizophrenia: undermined efforts to investigate the relationship review of epidemiological findings and future direc- between aspects of the social environment and psy- tions. Schizophrenia Bulletin, 34 (6), 1066–1082. chosis: direction of causation and mechanisms. The 8. van Winkel, R. and Stefanis, N. C. (2008) Myin- recent upsurge of interest in social factors has, in part, Germeys I. Psychosocial stress and psychosis. A re- been fuelled by research that has disentangled cause view of the neurobiological mechanisms and the evi- and effect for some exposures (e.g. urbanicity) and by dence for gene–stress interaction. Schizophrenia Bul- research providing plausible and, to some extent, letin, 34 (6), 1095–1105. evidenced mechanisms that link experience and psy- 9. Garety, P. A., Bebbington, P., Fowler, D. et al. (2007) chosis. These findings need replication and exten- Implications for neurobiological research of cognitive sion, and exciting work is already underway (e.g. models of psychosis: a theoretical paper. Psychologi- gene–environment interactions; quasi-experimental cal Medicine, 37 (10), 1377–1391. designs; effects of social experiences on stress res- 10. Garety, P. A., Kuipers, E., Fowler, D. et al. (2001) A ponsivity, brain structure and function). The promise cognitive model of the positive symptoms of psycho- sis. Psychological Medicine, 31 (2), 189–195. is that this will generate knowledge about areas in which we can intervene to interrupt problematic social 11. Fisher, H. and Craig, T. (2008) Childhood adversity and psychosis, in Society and Psychosis (eds. C. Morgan, development and prevent onset and continued suffer- K. McKenzie and P. Fearon), Cambridge University ing, at least for some. After all, it is surely this that Press, Cambridge, pp. 95–111. drives all research into these devastating disorders. 12. Bebbington, P., Fowler, D., Garety, P. et al. (2008) Theories of cognition, emotion and the social world: missing links in psychosis, in Society and Psychosis REFERENCES (eds C. Morgan, K. McKenzie and P. Fearon), Cam- bridge University Press, Cambridge, pp. 219–237. 1. Andreasen, N. (2000) Schizophrenia: the fundamental 13. WHO (1974) Schizophrenia: A Multinational Study, questions. Brain Research Reviews, 31, 106–112. World Health Organization, Geneva. 2. Krabbendam, L. and van Os, J. (2005) Schizophrenia 14. Jablensky, A., Sartorius, N., Ernberg, G. et al. (1992) and urbanicity: a major environmental influence – Schizophrenia: manifestations, incidence and course

THE SOCIODEVELOPMENTAL ORIGINS OF PSYCHOSIS 209 in different cultures: a World Health Organization ten- 27. Pedersen, C. B.and Mortensen, P. B.(2006) Urbanicity country study. Psychological Medicine, 20 (suppl. 20), during upbringing and bipolar affective disorders in 1–97. Denmark. Bipolar Disorders, 8, 242–247. 15. Crow, T. (2000) Schizophrenia as the price that Homo 28. Pedersen, C. B. and Mortensen, P. B. (2001) Evidence sapiens pays for language: a resolution of the central of a dose–response relationship between urbanicity paradox in the origin of the species. Brain Research during upbringing and schizophrenia risk. Archives of Reviews, 31, 118–129. General Psychiatry, 58 (11), 1039–1046. 16. Crow, T. J. (2007) How and why genetic linkage has not 29. Marcelis, M., Navarro-Mateu, F., Murray, R. et al. solved the problem of psychosis: review and hypoth- (1998) Urbanization and psychosis: a study of esis. American Journal of Psychiatry, 164 (1), 13–21. 1942–1978 birth cohorts in The Netherlands. Psycho- 17. Cantor-Graae, E. and Selten, J. P. (2005) Schizophre- logical Medicine, 28 (4), 871–879. nia and migration: a meta-analysis and review. Amer- 30. McGrath, J. and Scott, J. (2006) Urban birth and risk of ican Journal of Psychiatry, 162, 12–24. schizophrenia: a worrying example of epidemiology 18. McGrath, J., Saha, S., Welham, J. et al. (2004) A where the data are stronger than the hypotheses. systematic review of the incidence of schizophrenia: Epidemiologia e Psichiatria Sociale, 15, 243–246. the distribution of rates and the influence of sex, 31. Faris, R. and Dunham, H. (1939) Mental Disorders in urbanicity, migrant status and methodology. BMC Urban Areas, University of Chicago Press, Chicago, Medical, 2 (1), 13. Illinois. 19. Kleinman, A. (1991) Rethinking Psychiatry: From 32. Hare, E. (1956) Mental illness and social conditions in Cultural Category to Personal Experience, The Free Bristol. Journal of Mental Science, 102, 349–357. Press, New York. 33. Durkheim, E.(1952) Suicide, TheFree Press, Glencoe, 20. McGrath, J. J. (2007) The surprisingly rich contours of Illinois. schizophrenia epidemiology. Archives of General Psy- 34. Merton, R. K. (1938) Social structure and anomie. chiatry, 64 (1), 14–16. American Sociological Review, 3, 672–682. 21. Verdoux, H. and van Os, J. (2002) Psychotic symptoms 35. Deflam, M. (2007) Anomie, in The Blackwell Ency- in non-clinical populations and the continuum of clopedia of Sociology (ed. G. Ritzer), Blackwell Pub- psychosis. Schizophrenia Research, 54 (1–2), 59–65. lishing, Oxford. 22. van Os, J., Linscott, R. J., Myin-Germeys, I. et al. 36. Allardyce, J. and Boydell, J. (2006) Review: the wider (2009) A systematic review and meta-analysis of the social environment and schizophrenia. Schizophrenia psychosis continuum: evidence for a psychosis Bulletin, 32 (4), 592–598. proneness–persistence–impairment model of psycho- 37. Boydell, J., van Os, J., McKenzie, K. and Murray, R. tic disorder. Psychological Medicine, 39 (2), 179–195. M. (2004) The association of inequality with the 23. Dutta, R., Greene, T., Addington, J. et al. (2007) incidence of schizophrenia – an ecological study. Biological, life course, and cross-cultural studies all Social Psychiatry and Psychiatric Epidemiology, point toward the value of dimensional and develop- 39 (8), 597–599. mental ratings in the classification of psychosis. Schi- 38. Allardyce, J., Gilmour, H., Atkinson, J. et al. (2005) zophrenia Bulletin, 33 (4), 868–876. Social fragmentation, deprivation and urbanicity: re- 24. Spauwen, J., Krabbendam, L., Lieb, R. et al. (2004) Does lation to first-admission rates for psychoses. British urbanicity shift the population expression of psychosis? Journal of Psychiatry, 187, 401–406. Journal of Psychiatric Research, 38 (6), 613–618. 39. Kirkbride, J. B., Fearon, P., Morgan, C. et al. (2006) 25. Morgan, C., Fisher, H., Hutchinson, G. et al. (2009) Heterogeneity in incidence rates of schizophrenia and Ethnicity, social disadvantage and psychotic-like ex- other psychotic syndromes: findings from the 3-center periences in a healthy population based sample. Acta ÆSOP study. Archives of General Psychiatry, 63 (3), Psychiatrica Scandinavica, 119 (3), 226–235. 250–258. 26. Morgan, C., Kirkbride, J., Leff, J. et al. (2007) Parental 40. Kirkbride, J. B., Morgan, C., Fearon, P. et al. separation, loss and psychosis in different ethnic (2007) Neighbourhood-level effects on psychoses: re- groups: a case-control study. Psychological Medicine, examining the role of context. Psychological Medicine, 37 (4), 495–503. 37 (10), 1413–1425.

210 SOCIAL DETERMINANTS 41. Kirkbride, J. B., Boydell, J., Ploubidis, G. B. et al. 53. Cooper, B. (2005) Immigration and schizophrenia: the (2008) Testing the association between the incidence social causation hypothesis revisited. British Journal of schizophrenia and social capital in an urban area. of Psychiatry, 186, 361–363. Psychological Medicine, 38 (8), 1083–1094. 54. Bhugra, D., Hilwig, M., Hossein, B. et al. (1996) First- 42. van Os, J., Driessen, G., Gunther, N. and Delespaul, P. contact incidence rates of schizophrenia in Trinidad (2000) Neighbourhood variation in incidence of schi- and one-year follow-up. British Journal of Psychiatry, zophrenia. Evidence for person–environment interac- 169, 587–592. tion. British Journal of Psychiatry, 176, 243–248. 55. Hickling, F. and Rodgers-Johnson, P. (1995) The in- 43. McGrath, J., El-Saadi, O., Cardy, S. et al. (2001) Urban cidence of first-contact schizophrenia in Jamaica. birth and migrant status asrisk factors forpsychosis: an British Journal of Psychiatry, 167, 193–196. Australian case-control study. Social Psychiatry and 56. Mahy, G., Mallett, R., Leff, J. and Bhugra, D. (1999) Psychiatric Epidemiology, 36 (11), 533–536. First-contact incidence of schizophrenia in Barbados. 44. Fearon, P. and Morgan, C. (2006) Environmental British Journal of Psychiatry, 175, 28–33. factors in schizophrenia: the role of migrant studies. 57. Boydell, J., van Os, J., McKenzie, K. et al. (2001) Schizophrenia Bulletin, 32 (3), 405–408. Incidence of schizophrenia in ethnic minorities in 45. Kirkbride, J. B., Barker, D., Cowden, F. et al. (2008) London: ecological study into interactions with envir- Psychoses, ethnicity and socio-economic status. Brit- onment. British Medical Journal, 323, 1336–1338. ish Journal of Psychiatry, 193 (1), 18–24. 58. Veling, W., Susser, E., van Os, J. et al. (2008) Ethnic 46. Bresnahan, M., Begg, M. D., Brown, A. et al. (2007) density of neighborhoods and incidence of psychotic Race and risk of schizophrenia in a US birth cohort: disorders among immigrants. The American Journal of another example of health disparity? International Psychiatry, 165 (1), 66–73. Journal of Epidemiology, 36 (4), 751–758. 59. Veling, W., Hoek, H. W. and Mackenbach, J. P. (2008) 47. Fearon, P., Kirkbride, J. B., Morgan, C. et al. (2006) Perceived discrimination and the risk of schizophrenia Incidence of schizophrenia and other psychoses in in ethnic minorities: a case-control study. Social Psy- ethnic minority groups: results from the MRC chiatry and Psychiatric Epidemiology, 43 (12), ÆSOP Study. Psychological Medicine, 36 (11), 953–959. 1541–1550. 60. Dean, K., Dazzan, P., Lloyd, T. et al. (2007) Minor 48. Morgan, C. and Hutchinson, G. (in press) The social physical anomalies across ethnic groups in a first determinants of psychosis in migrant and ethnic min- episode psychosis sample. Schizophrenia Research, ority populations: a public health tragedy. Psycholo- 89 (1–3), 86–90. gical Medicine, 1–5. 61. Morgan, C., Kirkbride, J., Hutchinson, G. et al. (2008) 49. Hickling, F. W., McKenzie, K., Mullen, R. and Murray, Cumulative social disadvantage, ethnicity and first- R. (1999) A Jamaican psychiatrist evaluates diagnoses episode psychosis: a case-control study. Psychological at a London psychiatric hospital. British Journal of Medicine, 38 (12), 1701–1715. Psychiatry, 175, 283–285. 62. Tienari, P. and Wahlberg, K. (2008) Family environ- 50. Veling, W., Selten, J. P., Veen, N. et al. (2006) In- ment and psychosis, in Society and Psychosis (eds C. cidence of schizophrenia among ethnic minorities in Morgan, K. McKenzie and P. Fearon), Cambridge the Netherlands: a four-year first-contact study. Schi- University Press, Cambridge, pp. 112–126. zophrenia Research, 86 (1–3), 189–193. 63. Read, J., Seymour, F. and Mosher, L. (2004) Unhappy 51. Sharpley, M., Hutchinson, G., McKenzie, K. and families, in Models of Madness (eds J. Read, L. Mosher Murray, R. (2001) Understanding the excess of psy- and R. Bentall), Brunner-Routledge, East Sussex, pp. chosis among the African-Caribbean population in 253–268. England. British Journal of Psychiatry, 178 (suppl. 64. Fromm-Reichmann, F. (1948) Notes on the develop- 40), s60–s68. ment of treatment of schizophrenics by psycho-ana- 52. McKenzie, K., Fearon, P. and Hutchinson, G. (2008) lytic psychotherapy. Psychiatry, 11, 263–273. Migration, ethnicity and psychosis, in Society and 65. Bateson, G., Jackson, D., Haley, J. and Weakland, J. Psychosis (eds C. M. Morgan, K. McKenzie and P. (1956) Toward a theory of schizophrenia. Behavioural Fearon), Cambridge University Press, Cambridge. Science, 1, 251–264.

THE SOCIODEVELOPMENTAL ORIGINS OF PSYCHOSIS 211 66. Lainge, R. D. and Esterson, D. (1964) Sanity, Madness 79. Shevlin, M., Houston, J. E., Dorahy, M. J. and Adam- and the Family, Basic Books, New York. son, G. (2008) Cumulative traumas and psychosis: an 67. Hirsch, S. and Leff, J. (1975) Abnormalities in the analysis of the national comorbidity survey and the Parents of Schizophrenics, Oxford University Press, British Psychiatric Morbidity Survey. Schizophrenia Oxford. Bulletin, 34 (1), 193–199. 68. Kavanagh, N. (1992) Recent developments in ex- 80. Spataro, J., Mullen, P. E., Burgess, P. M. et al. (2004) pressed emotion and schizophrenia. British Journal Impact of child sexual abuse on mental health: pro- of Psychiatry, 160, 601–620. spective study in males and females. British Journal of Psychiatry, 184, 416–421. 69. Read, J., van Os, J., Morrison, A. P. and Ross, C. A. (2005) Childhood trauma, psychosis and schizophre- 81. Fisher, H., Morgan, C., Dazzan, P. et al. (2009) Gender nia: a literature review with theoretical and clinical differences in the association between childhood abuse implications. Acta Psychiatrica Scandinavica, 112 and psychosis. British Journal of Psychiatry, 194 (4), (5), 330–350. 319–325. 70. Bendall, S., Jackson, H. J., Hulbert, C. A. and McGor- 82. Wicks, S., Hjern, A., Gunnell, D. et al. (2005) Social ry, P. D. (2008) Childhood trauma and psychotic adversity in childhood and the risk of developing disorders: a systematic, critical review of the evidence. psychosis: a national cohort study. American Journal Schizophrenia Bulletin, 34 (3), 568–579. of Psychiatry, 162, 1652–1657. 71. Jannsen, I., Krabbendam, L., Bak, M. et al. (2004) 83. Agid, O., Shapira, B., Zislin, J. et al. (1999) Environ- Childhood abuse as a risk factor for psychosis. Acta ment and vulnerability to major psychiatric illness: a Psychiatrica Scandinavica, 109, 38–45. case control study of early parental loss in major depression, bipolar disorder and schizophrenia. Mo- 72. Whitfield, C. L., Dube, S. R., Felitti, V. J. and Anda, R. F. (2005) Adverse childhood experiences and halluci- lecular Psychiatry, 4 (2), 163–172. nations. Child Abuse and Neglect, 29, 797–810. 84. Mulvany, F., O’Callaghan, E., Takei, N. et al. (2001) Effect of social class at birth on risk and presentation of 73. Read, J., Agar, K., Argyle, N. and Aderhold, V. (2003) schizophrenia: case-control study. British Medical Sexual and physical assault during childhood and Journal, 323 (7326), 1398–1401. adulthood as predictors of hallucinations, delusions and thought disorder. Psychology and Psychotherapy: 85. Harrison, G., Gunnell, D., Glazebrook, C. et al. (2001) Theory, Research and Practice, 76, 1–22. Association between schizophrenia and social in- equality at birth: case-control study. British Journal 74. Hammersley, P., Dias, A., Todd, G. et al. (2003) Child- of Psychiatry, 179, 346–350. hood traumas and hallucinations in bipolar affective disorder. British Journal of Psychiatry, 182, 543–547. 86. Fallon, P. (2008) Life events; their role in onset and relapse inpsychosis, research utilizing semi-structured 75. Morrison, A. and Petersen, T. (2003) Trauma and interview methods: a literature review. Journal of metacognition as predictors of predisposition to hal- Psychiatric and Mental Health Nursing, 15 (5), lucinations. Behavioural and Cognitive Psychother- 386–392. apy, 31, 235–246. 87. Norman, R. M. and Malla, A. K. (1993) Stressful life 76. Kelleher, I., Harley, M., Lynch, F. et al. (2008) Associa- events and schizophrenia. I: A review of the research. tionsbetween childhood trauma, bullying and psychotic British Journal of Psychiatry, 162, 161–166. symptoms among a school-based adolescent sample. British Journal of Psychiatry, 193 (5), 378–382. 88. Bebbington, P., Wilkins, S., Jones, P. et al. (1993) Life events and psychosis. Initial results from the Camber- 77. Spauwen, J., Krabbendam, L., Lieb, R. et al. (2006) well Collaborative Psychosis Study. British Journal of Impact of psychological trauma on the development Psychiatry, 162, 72–79. of psychotic symptoms: relationship with psychosis proneness. BritishJournalofPsychiatry, 188, 527–533. 89. Malla, A. K., Cortese, L., Shaw, T. S. and Ginsberg, B. (1990) Life events and relapse in schizophrenia. A one 78. Bebbington, P., Bhugra, D., Singleton, N. et al. (2004) Psychosis, victimisation and childhood disadvantage: year prospective study. Social Psychiatry and Psy- evidence from the second British National Survey of chiatric Epidemiology, 25 (4), 221–224. Psychiatric Morbidity. British Journal of Psychiatry, 90. Johns, L. C., Cannon, M., Singleton, N. et al. (2004) 185, 220–226. Prevalence and correlates of self-reported psychotic

212 SOCIAL DETERMINANTS symptoms in the British population. British Journal of opmental course of schizophrenia. Annual Review of Psychiatry, 185, 298–305. Clinical Psychology, 4, 189–216. 91. Wiles, N. J., Zammit, S., Bebbington, P. et al. (2006) 103. Pariante, C. M., Vassilopoulou, K., Velakoulis, D. Self-reported psychotic symptoms in the general po- et al. (2004) Pituitary volume in psychosis. British pulation: results from the longitudinal study of the Journal of Psychiatry, 185, 5–10. British National Psychiatric Morbidity Survey. British 104. Hall, F. S., Wilkinson, L. S.,Humby, T. and Robbins, T. Journal of Psychiatry, 188, 519–526. W. (1999) Maternal deprivation of neonatal rats pro- 92. Jones, P. B., Bebbington, P., Foerster, A. et al. (1993) duces enduring changes in dopamine function. Sy- Premorbid social underachievement in schizophrenia. napse, 32 (1), 37–43. Results from the Camberwell Collaborative Psychosis 105. Berton, O., McClung, C. A., Dileone, R. J. et al. (2006) Study. British Journal of Psychiatry, 162, 65–71. Essential role of BDNF in the mesolimbic dopamine 93. Kendler, K. S. and Baker, J. H. (2007) Genetic influ- pathway in social defeat stress. Science, 311 (5762), ences on measures of the environment: a systematic 864–868. review. Psychological Medicine, 37 (5), 615–626. 106. Tidey, J. W. and Miczek, K. A. (1996) Social defeat 94. Jones, P., Rodgers, B., Murray, R. and Marmot, M. stress selectively alters mesocorticolimbic dopamine (1994) Child development risk factors for adult schi- release: an in vivo microdialysis study. Brain Re- zophrenia in the British 1946 birth cohort. Lancet, 344 search, 721 (1–2), 140–149. (8934), 1398–1402. 107. Isovich, E., Engelmann, M., Landgraf, R. and Fuchs, 95. Selten, J. P. and Cantor-Graae, E. (in press) The denial E. (2001) Social isolation after a single defeat reduces of a psychosis epidemic. Psychological Medicine, 1–3. striatal dopamine transporter binding in rats. Eur- 96. Myin-Germeys, I. and van Os, J. (2008) Adult adver- opean Journal of Neuroscience, 13 (6), 1254–1256. sity: do early environment and genotype create lasting 108. Morgan, D., Grant, K. A., Gage, H. D. et al. (2002) vulnerability for adult social adversity in psychosis?, Social dominance in monkeys: dopamine D2 receptors in SocietyandPsychosis (eds C.Morgan,K. McKenzie and cocaine self-administration. Nature Neu- and P. Fearon), Cambridge University Press, Cam- roscience, 5 (2), 169–174. bridge, pp. 127–142. 109. Pruessner, J. C., Champagne, F., Meaney, M. J. and 97. Glaser, J. P., van Os, J., Portegijs, P. J. and Myin- Dagher, A. (2004) Dopamine release in response to a Germeys, I. (2006) Childhood trauma and emotional psychological stress in humans and its relationship to reactivity to daily life stress in adult frequent attenders early life maternal care: a positron emission tomogra- of general practitioners. Journal of Psychosomatic phy study using [11C]raclopride. Journal of Neu- Research, 61 (2), 229–236. roscience, 24 (11), 2825–2831. 98. Myin-Germeys, I., van Os, J., Schwartz, J. E. et al. 110. De Bellis, M. D., Lefter, L., Trickett, P. K. and (2001) Emotional reactivity to daily life stress in PutnamJr, F. W. (1994) Urinary catecholamine excre- psychosis. Archives of General Psychiatry, 58 (12), tion in sexually abused girls. Journal of the American 1137–1144. Academy of Child and Adolescent Psychiatry, 33 (3), 99. Scheper-Hughes, N. (2001) Saints, Scholars and Schi- 320–327. zophrenics: Mental Illness in Rural Ireland, Twentieth 111. Heim, C., Newport, D. J., Heit, S. et al. (2000) Anniversary Edition, University of California Press, Pituitary–adrenal and autonomic responses to stress Los Angeles, California. in women after sexual and physical abuse in child- 100. Howes, O. D. and Kapur, S. (2009) The dopamine hood. Journal of the American Medical Association, hypothesis of schizophrenia: version III – the final com- 284 (5), 592–597. mon pathway. Schizophrenia Bulletin, 35 (3), 549–562. 112. De Bellis, M. D., Chrousos, G. P., Dorn, L. D. et al. 101. Di Forti, M., Lappin, J. M. and Murray, R. M. (2007) (1994) Hypothalamic–pituitary–adrenal axis dysregu- Risk factors for schizophrenia – all roads lead to lation in sexually abused girls. Journal of Clinical dopamine. European Neuropsychopharmacology, 17 Endocrinology and Metabolism, 78 (2), 249–255. (suppl. 2), S101–S107. 113. Morrison, A., Read, J. and Turkington, D. (2005) Trauma 102. Walker, E., Mittal, V. and Tessner, K. (2008) Stress and and psychosis: theoretical and clinical implications. Acta the hypothalamic pituitary adrenal axis in the devel- Psychiatrica Scandinavica, 112 (5), 327–329.

THE SOCIODEVELOPMENTAL ORIGINS OF PSYCHOSIS 213 114. Selten, J. P. and Cantor-Graae, E. (2005) Social defeat: ing psychosis? American Journal of Psychiatry, 160 risk factor for schizophrenia? British Journal of Psy- (3), 477–482. chiatry, 187, 101–102. 122. Risch, N., Herrell, R., Lehner, T. et al. (2009) Inter- 115. Weaver, I. C., Cervoni, N., Champagne, F. A. et al. action between the serotonin transporter gene (5- (2004) Epigenetic programming by maternal behavior. HTTLPR), stressful life events, and risk of depression: Nature Neuroscience, 7 (8), 847–854. a meta-analysis. Journal of the American Medical 116. Meaney, M. J. and Szyf, M. (2005) Environmental Association, 301 (23), 2462–2471. programming of stress responses through DNA methy- 123. Bentall, R. P. and Fernyhough, C. (2008) Social pre- lation: life at the interface between a dynamic envir- dictors of psychotic experiences: specificity and psy- onment and a fixed genome. Dialogues in Clinical chological mechanisms. Schizophrenia Bulletin, NeuroSciences, 7 (2), 103–123. 34 (6), 1012–1020. 117. Caspi, A., Sugden, K., Moffitt, T. E. et al. (2003) 124. Freeman, D., Garety, P. A. and Kuipers, E. (2001) Influence of life stress on depression: moderation by a Persecutory delusions: developing the understanding polymorphism in the 5-HTT gene. Science, 301 of belief maintenance and emotional distress. Psycho- (5631), 386–389. logical Medicine, 31 (7), 1293–1306. 118. Caspi, A., Moffitt, T. E., Cannon, M. et al. (2005) 125. Sharpley, M. S. and Peters, E. R. (1999) Ethnicity, Moderation of the effect of adolescent-onset cannabis class and schizotypy. Social Psychiatry and Psychia- use on adult psychosis by a functional polymorphism tric Epidemiology, 34 (10), 507–512. in the catechol-O-methyltransferase gene: longitudi- 126. Demjaha, A., Morgan, K., Morgan, C. et al. (2006) nal evidence of a gene X environment interaction. Symptom dimensions and ethnicity in the ÆSOP first Biological Psychiatry, 57 (10), 1117–1127. onset psychosis study. Schizophrenia Research, 81, 119. Stefanis, N. C., Henquet, C., Avramopoulos, D. et al. 233–1233. (2007) COMT Val158Met moderation of stress-in- 127. Thornicroft, G. (2006) Shunned: Discrimination duced psychosis. Psychological Medicine, 37 (11), Against People with Mental Illness, Oxford University 1651–1656. Press, Oxford. 120. Tienari, P., Wynne, L. C., Sorri, A. et al. (2004) 128. Kleinman, A., Das, V. and Lock, M. (eds) (1997) Genotype–environment interaction in schizophrenia- Social Suffering, University of California Press, Los spectrum disorder. British Journal of Psychiatry, 184, Angeles, California. 216–222. 129. Oh, G. and Petronis, A. (2008) Environmental studies 121. van Os, J., Hanssen, M., Bak, M. et al. (2003) Do of schizophrenia through the prism of epigenetics. urbanicity and familial liability coparticipate in caus- Schizophrenia Bulletin, 34 (6), 1122–1129.



16 Depression Tom K. J. Craig Health Services and Population Research Department, Institute of Psychiatry, King's College London, London, UK 16.1 INTRODUCTION There are probably few psychiatrists now who doubt increased stress and a higher incidence of depressive thecentralcausalroleplayedbythesocialenvironment symptoms [7]. Women are twice as likely to develop in the onset and course of major depression. Meta- depression than are their male partners following analyses of population surveys have consistently crises involving children, procreation or housing, shown higher rates among the economically disadvan- while there is no difference in risk for other crises [8]. taged[1],thoselivinginlowstatusneighbourhoods[2] Considerable progress has been made in terms of and with a low standard of living [3,4]. However, this moving beyond these broad associations towards de- effect of poverty is largely negated once account has tailed aetiological models that specify the kinds of been taken of social network size and quality [5]. experience and their interactions with wider social and Women are twice as likely as men to develop biological vulnerabilities that are important for de- depression in adulthood and this increased risk is pression. This chapter draws together disparate largely down to social factors. Marriage is protective strands of research based on the strongest current for men but not for women, where it may even confer a aetiological model for depression in women that owes higher risk [6]. Women take the main responsibility much to three decades of work by Professor George for child-rearing and this role is associated both with Brown and colleagues. 16.2 MEASURING STRESSFUL EXPERIENCE Establishing a causal relationship between stressful because individuals reorder experiences in their strug- experience and depression requires instruments that gle to make sense of why they became ill when they can establish time order and capture the essence of the did. Similarly, events with little emotional impact ‘depressogenic’ experience. Without very careful at- when they occurred may, through the lens of a sub- tention, events that occurred shortly after an onset may sequent depression, be recalled as having been of far be erroneously recalled as occurring before it, partly greater significance. These problems are mitigated by Principles of Social Psychiatry, second edition Edited by Craig Morgan and Dinesh Bhugra Ó 2010 John Wiley & Sons, Ltd.

216 SOCIAL DETERMINANTS using semi-structured interviews that pay close atten- guesses about how others would feel in response to tion to time order and which distinguish experiences particular situations. The accuracy of these judge- that may be the products of incipient illness, e.g. being ments is increased if the person doing the judging dismissed from work because of poor time-keeping, knows something of the other’s biography and the from ‘independent’ events that are logically outside wider social context. So, for example, most people the control of the individual [9,10]. would agree that the emotional reaction to a positive Pencil and paper questionnaires also do not deal pregnancy test is likely to be very different for a very well with defining and assessing the different comfortably well-off woman in a stable partnership aspects of ‘stressfulness’ contained within any one from that of a single mother on welfare benefits who type of experience. Simply asking a depressed parti- has just been abandoned by her lover. That this con- cipant to report their feelings in retrospect is clearly textual judgement will apply to most women in similar open to biases of recall and reappraisal. The impact of circumstances is sufficient for aetiological enquiry, some experiences may be played down and others despite the fact that a few may report a different exaggerated and there is always the possibility that the reaction. This approach is the basis for the contextual crucial meaning of the experience is not even acces- measure of meaning used by the Life Events and sible to consciousness – for example, a young woman Difficulties Schedule (LEDS) developed by Brown expecting her first child but abandoned by her lover and Harris [9]. The measure distinguishes ‘events’ as might well dwell on anxiety about her future and not discrete and datable experiences from ‘difficulties’ report the extent to which she felt betrayed. that can last for months or years without any discrete An alternative to relying solely on self-report is to crisis. Ratings of a variety of qualitative dimensions of use the investigator to pass judgement on the likely relevant experiences are made in consensus meetings meaning of the crisis, taking account of the context in by investigators who are ‘blind’ to the subject’s which it occurred. This is possible because humans reported reaction (which is separately recorded but have the capacity to empathize and make reasonable not discussed). 16.3 BUILDING AN AETIOLOGICAL MODEL 16.3.1 Events and difficulties 80% of the depressed women but only 30% of the healthy women in the general population. Since this In the earlier studies [9] emphasis was placed on the study there have been numerous replications of the importance of a broad notion of contextual ‘threat’ basic result [11–16]. reflecting how unpleasant the average person in a A very important aspect of the rating of contextual similar context would find a particular event. Threat threat is the assumption that the experiences most was rated in terms of the immediate impact (short- relevant for depression are those that disrupt relation- term threat) and its more protracted implications some ships and goals to which the individual is committed. 10–14 days later (long-term threat). The first study to This was confirmed in a longitudinal study of working use this system of contextual measurement took place class women with at least one child at home living in some 30 years ago in Camberwell, London [9]. For Islington (a London borough). At baseline, measures both patients and depressive cases in the community, were taken of the woman’s commitment to six key the vast majority of onsets were preceded by the domains of her life including motherhood, employ- occurrence of experiences with long-term threat. ment and very close relationships. At follow-up a year Events with short-term threat only were of no aetio- later all severe events were rated according to whether logical importance. Only the most severe difficulties they ‘matched’ one of the domains in which women lasting for two or more years seemed to be important. had demonstrated high commitment at baseline. Wo- Taken together, one or more of these severe events or men who experienced such a ‘matching’ severe event difficulties had occurred prior to onset in more than were three times more likely to develop depression

DEPRESSION 217 than women with nonmatching but equally severe one of these events were three times more likely to events [15]. develop depression than were women who experi- Cooper and Paykel [17], in the earlier edition of this enced events characterized by loss alone. Further- book, point out that there is no evidence to suggest that more, only losses involving death of a loved one really life event stress as a causal agent is specific to depres- contributed, while events that did not involve humi- sion but note that there may be some qualities of events liation, entrapment or loss were not significantly that are more likely to lead to depression than others. associated with onset. Robert Finlay-Jones and Brown [18] refined the con- These observations were replicated by Kendler and cept of threat to distinguish severe events characterized colleagues[23]inastudyof7322maleandfemaletwins by ‘loss’ and ‘danger’. Loss was defined to include recruited from a community case register in Virginia, deaths, financial and material losses, and the loss of a USA.Semi-structuredinterviews wereusedtoestablish ‘cherishedidea’(e.g.thelossofbeliefthatone’sspouse the presence of major depression and generalized anxi- is faithful following the discovery of his extramarital etyandtheoccurrenceof‘personal’ stressfulevents(i.e. affair).Danger wasdefinedasthe threatofafutureloss, events directly affecting the participant, such as divorce which might occur as the result of the present event but or separation, job loss, legal problems) and ‘network’ was not inevitable (e.g. a professional dancer who events(i.e.thoseaffectingothersclosetotherespondent, following an injury is told by her doctor that it is such as the death or serious illness of a parent). For each possible that she may never dance again). Loss events oftheeventsthatmetcriteriafor‘highthreat’,additional were causally linked to the occurrence of depression ratings were made of loss, humiliation, entrapment and anddangertoanxietywitheventscharacterizedbyboth danger using an adaptation of the LEDS definitions. In qualities more frequently reported by subjects who had bothmenandwomen,onsetsofdepressionwerestrongly symptoms of both conditions [18]. These findings have associatedwithlossandhumiliationeventswhileonsets been replicated in subsequent studies [15,19]. of anxiety were predicted by higher ratings of danger. Further results from the Islington study suggest that Mixed states of both major depression and generalized loss itself may not be the most important characteristic anxiety were associated with events that shared char- of stress involved in depression. Close attention to the acteristics of loss, humiliation and danger. Events that qualitative descriptions of severe events suggested contained elements of both loss and humiliation were that many of those that preceded depression were particularly strongly associated with depression. characterized by humiliation or defeat and the sheer Most of the research reviewed to this point concerns lack of any prospect that things might get better in the major depression as encountered in the general po- future. This observation chimes with speculations of pulation and in outpatient samples within which stu- some evolutionary theorists that depression has its dies suggest that 66–90% of episodes can be traced to origins in the experience of defeat, loss of rank and a severe event within the prior 6 months. However, this value, and submission, such as is seen in many group- still leaves some onsets of depression without a pre- living species after being outranked by a rival or losing cipitating stressor, particularly among severe cases in a battle to defend territorial boundaries [20–22]. This treatment settings. At one time this was taken as led to a refinement of the contextual rating system to indicating that there were two types of depression, classify severely threatening events further in terms of some of which were less reactive to the environment the experience of loss, humiliation and entrap- and so had a more biological basis. While a simplistic ment [16]. The category of humiliation involves the dichotomy is now thought to be incorrect, it has been likelihood that the event would provoke a sense of found that there may be a difference between first and being put down or lead to a marked loss of face or self- subsequent episodes of depression, with first episodes esteem and that of entrapment the extent to which the being more strongly associated with life stress event emphasized the fact of being trapped in a than subsequent recurrences; the fact that more hos- punishing situation. Using these classifications,events pitalized cases are of recurrent disorders probably involving humiliation or entrapment preceded the explains the weaker association with events in these bulk of onsets of depression. Women experiencing populations [24–28].

218 SOCIAL DETERMINANTS 16.3.2 Amplifying the impact of events: childhood maltreatment to the adult vulnerability fac- psychosocial vulnerability tors is illustrated by the observation that poor-quality parental care in childhood is associated with early While the majority of onsets of depression are pre- cohabitation and premarital pregnancy that can trap ceded by severe events or difficulties, only about a fifth young women in unsatisfactory relationships and can of those experiencing these stressors go on to develop become a source of other chronic financial and inter- depression. In the original Camberwell study, having personal difficulties [33]. three or more children, lack of paid employment and The majority of episodes of depression in the the lack of a close confiding relationship all increased community resolve without treatment [34] but a sig- the likelihood of depression in the face of a severely nificant minority of people with major depression are threatening event [9]. Subsequent studies have un- still depressed a year or two later [35]. Clinical factors packed more precisely what these broad vulnerability such as the severity of symptoms and the duration of indicators were picking up. Thus, for example, in the previous episodes are important [36], but so are broad Islington studyreferred to earlier,two indices assessed social conditions such as low socioeconomic sta- at the first interview together predicted the majority of tus [1,4,37]. In the general population studies carried onsets of depression following a severe event in the out by George Brown and colleagues, the likelihood of follow-up year. These were negative psychological a depression taking a chronic course (i.e. lasting at factors such as negative evaluation of self (i.e. low least a year) was strongly associated with the presence self-esteem) and negative environmental factors com- of an interpersonal difficulty at onset, again particu- prising negative interactions in the home (particularly larly where these involved a core sexual relation- involving a core sexual relationship) or, for single ship [38,39]. As discussed earlier, these difficulties parents, the absence of a close confiding relationship. were more likely in women who had experienced On their own these factors were not associated with childhood maltreatment, but the latter also had a the onset of depression but, in the presence of an substantial direct effect on whether an onset took a appropriate severe event, both indices related to onset chronic course [32]. and were additive in effect [29]. These findings were replicated in a further longitudinal study by the same research team [30]. 16.3.3 Biological vulnerability and Thevulnerability factors not only reduce a woman’s gene–environment interaction capacitytosurviveacrisisbutalsoactuallyincreasethe risk of a crisis occurring. Given some thought, this is Geneticfactorsprobablyaccountforaround37%ofthe hardly surprising. For example, the likelihood of an variance of depression [40], though this is significantly infidelity or separation event is greater in disharmo- higher in women than men [41]. There is a significant nious relationships even where the degree of marital gene by environment (GxE) interaction, such that ge- tensionisnotsufficienttocrossthethresholdforratinga neticfactorsinfluencetheonsetofdepressionbyaffect- marital difficulty. Thesevulnerability factors are stable ingthesensitivityoftheindividualtosevereevents[42]. over quite long periods of time and interrelated. For Genes may contribute indirectly to the occurrence of example, the experience of childhood maltreatment, stressfuleventsthroughtheirinfluenceontemperament defined by the presence of parental rejection, neglect, and through behaviours that make the occurrence of severe physical abuse or the experience of sexual abuse stressful events more likely [43,44]. Early studies ex- increases the likelihood of both negative psychological plored the contribution of temperament to the occur- and environmental vulnerabilities in adulthood and is rence and response to adversity through the concept of associated through these to a significantly increased neuroticism, but the results of these studies are some- risk of exposure to severe events, all together account- what inconsistent and it has been suggested that the ing fora doublingoftherisk of depressioninthosewith neuroticism measure reflects current mood state rather childhood maltreatment [31,32]. The pathway from than a stable temperamental characteristic [45]. On the

DEPRESSION 219 other hand, in a study of twins Kendler, Gardner and stressful events may operate is the response of the Prescott [46] found that the highly inherited trait of hypothalamic–pituitary–adrenal (HPA) axis and the neuroticism in one twin predicted not only the occur- secretion of cortisol. Cortisol has profound effects on rence of life events in the other twin but also the quality the brain, including on memory and the emotional of interpersonal relationships. In another study, child- appraisal of events. Hyperactivity of the HPA axis is hoodsexualabusewaspredictiveofadultmajordepres- the most consistent biological finding in major depres- sion, the relationship beingclearest in thosewith low to sion. The basic sensitivity of this axis is determined average levels of neuroticism [47]. genetically and is present from birth, but subsequent A way forward in disentangling the genetic and experiences have profound effects [49]. Studies of rats environmental contributions to aetiology has been and nonhuman primates show that separating neo- provided by studies based on samples from the nates from their mothers produces long-lasting Virginia Twin Registry in North America. In one of changes in the HPA axis that resemble those seen in these studies, Kendler and Gardner [48] identified depressed patients. Emotional deprivation and mal- female monozygous (MZ) twins who were discordant treatment in childhood have a profound impact on the for a lifetime history of major depression. As MZ axis [50] and these effects persist into adulthood. twins are genetically identical, the study of discordant Women with histories of sexual or severe physical pairs who grew up in the same household allows the abuse in childhood show increased HPA activation, investigator to tease out environmental variables that reflected in an enhanced heart rate response and characterize the affected sibling. Among the environ- ACTH release when stressed. If they are currently mental variables identified in this way were a history depressed, they also show a very large increase in of much lower levels of parental warmth, higher levels cortisol secretion [51]. Early life experiences sensitize of overprotectiveness and greater maternal authoritar- the parts of the central nervous system related to the ianism. The affected twin was more likely to be single emotional response to stress and produce hyper-re- or divorced and far more likely than their nonaffected activity of a variety of neurotransmitter systems [52]. sister to report stressful life events, particularly those These and related observations are taken to indicate involving romantic problems and job loss, higher rates that a disturbed HPA axis is not just a response to of problems with relatives and poorer social support. depression but an underlying neurobiological ab- The analysis, in a nutshell, replicates many of the normality that predisposes to the condition. linkages illustrated in the childhood to adulthood In healthy individuals, there is a peak of circulating pathway discussed earlier (Figure 16.1). cortisol about half an hour after waking that A possible biological mechanism that is partially gradually returns to awakening levels over the next 60 genetically determined but which has a powerful minutes. In depressed people the response is markedly environmental pathway through which the impact of increased [53]. Both cortisol and another corticosteroid, Severe Event onset Chronic Course Risky (sexual) behaviours (e.g. Interpersonal PMP) and aversive difficulty partnerships Negative childhood experiences e.g. Abuse/neglect BIOLOGICAL Low self – esteem VULNERABILITY Figure 16.1 Theoretical causal model

220 SOCIAL DETERMINANTS dehydroepiandrosterone(DHEA),havebeenfoundtobe 5-year period. This effect was also seen for early elevated on awakening among adolescents who subse- childhood maltreatment, which was most strongly quently develop major depression following a stressful associated with later depression in people with the experience[54].Extendingthisworktoan adult popula- short allele [57]. The publication of these findings led tion, Tirril Harris and colleagues [55] carried out a to a flurry of replications with mixed results (see prospective study of 116 women of whom 83 were Reference [58] for a detailed review). Two studies thought to be vulnerable to depression on the basis of have used a similar five-year index of life events: one, negativeaspectsofcorerelationshipsornegativeevalua- based on a sample of school teachers in Sydney, tions of self (see above). Salivary cortisol and DHEA replicated the results for first-ever episodes of de- werecollectedat0800and2000 hours.Participantswere pression [59]; the second, based on a large population followedat3monthlyintervalsfor12monthstomonitor survey with a questionnaire-based measure of stress, the onset of depression and, if onset occurred, were failed to replicate [60]. By far the majority of studies interviewed soon afterwards in order to obtain informa- have examined events occurring in the 12 months tion about life events as close in time as possible to the before onset, using a host of different ways of onset date. If therewas no onset, follow-up took place at measuring and dating stress. Not surprisingly per- the 12 month point. The onset of depression was asso- haps, the results are very mixed from studies showing ciated with severe events and with morning cortisol but, no evidence for GxE at all (e.g. see References [60] contrary to expectations, these appeared to be indepen- and [61]) through others showing a weak effect (e.g. dent effects. see Reference [62]). In one of the more careful Although much research has concentrated on the studies, Kendler et al. [63] studied participants taken HPA axis, there are several closely related systems from their earlier Virginia Twin Register studies with that are under genetic and environmental control. For careful dating of both the onset of depression and of example, the 5-hydroxytryptamine (5-HT, serotonin) events, excluding those who were already depressed neurotransmitter system is also modified by early at the start of the 12-month study period and events life experience [56]. Perhaps the most exciting that occurred after onset. Although the GxE interac- development in understanding the mechanism tion was replicated, the effect was considerably through which genetic susceptibility and the envir- smaller than that reported by earlier studies. onment interact is in the observation that the short In addition to the association with recent life events, allele of the serotonin transporter protein (5-HTT) on the observation of GxE involving early childhood has chromosome 17 is associated with susceptibility to also been replicated in four studies, two of which stressful events (or, conversely, that people with the specifically deal with maltreatment – one a study of long allele may be resilient to stress for genetic young adults [64] and the other with children [65]. All reasons). This was first shown by data from the four studies show strong effects and, of the studies to Dunedin Longitudinal Birth Cohort Study, in which date, only one has failed to replicate [60]. the rate of depression was markedly increased in Figure 16.1 pulls together these strands to provide a subjects with the short allele who had also been general model of the link between childhood and adult exposed to stressful life events compared to those vulnerability and their relationship to the onset of who had the long allele form. The life event measure depression. was an index of stressful events occurring over a 16.4 SOCIAL CAPITAL, SOCIAL SUPPORT AND SUPPORTIVE INTERVENTIONS: PREVENTION AND RECOVERY FROM DEPRESSION Social capital can be conceptualized as the ‘features of participants to act together more effectively to pursue social life – networks, norms and trust – that enable shared objectives’ [66]. It encompasses social

DEPRESSION 221 structures and networks and the resources embedded support can protect a person’s mental health or help within them [67,68]. Social network relationships are them recover from depression. sometimes referred to in terms of ‘bonding’ and The strands of evidence linking social support to ‘bridging’ ties between people. Bonding ties are those depression are complex, not least because the term that strengthen the cohesion of a social group, typified itself conceals many overlapping components. Sup- by high levels of mutual trust, loyalty and shared port may be classified in terms of actual resources responsibility. The benefits include caring for vulner- provided (e.g. see Reference [76]), which may be in able members of the group, but the other side of the terms of practical resources (e.g. a loan or the provi- coin are more negative characteristics of intolerance sion of direct assistance) or expressions of care and of people who do not ‘fit in’ to the structures and rules concern. Such objective support may be distinguished of the group. Bridging capital, on the other hand, from perceived support (e.g. see Reference [77]). It describes the wider links between diverse groups, may refer to a ‘background’ state such as the presence including broad networks of friends, acquaintances, of a close confidant or loving partnership or to what business partners, clubs and so forth. It is thought to was actually provided at the time of a crisis. Back- promote social inclusion and is generally viewed as ground objective support is only weakly correlated positive. Communities with high levels of social with perceived support [78], while it is this subjective capital, as measured by trust and civic participation, perception that has the stronger association with have lower death rates [69] and better self-rated mental health (e.g. see References [79] and [80]). health [70]. McCulloch [71] analysed data from the Women appear to provide more support than men [81] British Household Survey and found that people with and benefit more from perceived support than men, low social capital (measured as perceived character- having largerand morevaried social networks [82,83]. istics of neighbourhoods) had higher rates of common There is a broad consensus that positive social support mental disorders but, being a cross-sectional study, it is both protective of mental health [84,85] and capable is not possible to infer causality from this data. of promoting recovery [86,87]. Perceived lack of Correlations have also been shown between low levels support is associated with an increased incidence of of trust in people in general and perceptions of com- common mental disorders [72,79]. munity problems and common mental disor- Support measured at one point in time is no guar- ders [72,73]. Although social capital is usually con- antee of its availability in a crisis. In fact, not receiving ceptualized at an ecological level (i.e. the total amount support that one might have expected seems particu- of resources in a society potentially available to an larly depressogenic [84,88]. This is the converse of the individual), the evidence for an association between stress-buffering hypothesis where social support acts social capital measured at this ecological level and as a protection against the impact of a crisis [89,90]. mental health is inconsistent and weak at best [74]. Not all studies provide support for the stress-buffer Webber [75] points out that, at any absolute level of hypothesis (e.g. see References [91] and [92]) and the social capital within a community, individuals may evidence is probably stronger for the negative effect of have unequal access to the constituent resources be- the absence of support at times of crisis than for its cause of their location in the social hierarchy or the buffering capacity. strength of their interpersonal ties, so that an indivi- Recovery from depression may involve the reversal dual-level measure of social capital might be a better of the factors that are implicated in onset, such as a way of appraising social capital in relation to health. reduction in the severity of an ongoing difficulty or the He gives as an example the inability of a single mother occurrence of a ‘fresh-start’ event that gives renewed to obtain child care from friends and family that may hope for the future. Such ‘fresh starts’ might involve, contribute to her risk of depression or of gaining for example, a new job or the separation from an employment through informal social contacts that abusive spouse. In the follow-up of depressed women could provide a positive life change and promote the in the Islington studies described above, such events recovery from depression. This leads to a considera- and difficulty reductions occurred in more than half tion of social ties and the ways in which personal of the remissions in both community and patient

222 SOCIAL DETERMINANTS samples, and contributed to recovery in patients taking the chances of remission while new severe stressors antidepressant medication [93,94]. were found to inhibit remission [96]. Interventions for depression based onsocial support It is perhaps surprising given the extent of our have been described – they are, however, methodo- understanding of how the social environment affects logically very challenging to undertake. One such both the onset and course of depression that there intervention involved training female volunteers re- have not been more attempts to explore how the cruited through local press advertisements to providea social factors described in this chapter might influ- befriending service to 86 women suffering from ence response to common pharmacological and chronic depression. The volunteers were asked to psychological treatments. Nemeroff and collea- meet with the depressed woman for a minimum of gues [97] report an analysis of a large multicentre an hour each week, acting as a friend, providing study of chronic depression in which 681 patients practical support with ongoing difficulties and work- were randomized to 12 weeks of treatment with an ing to create fresh-start experiences. A randomized antidepressant or psychotherapy or a combination of controlled trial with a waiting list control design was the two. The combination therapy was more effec- employed with half of the depressed women being tive than either alone, but among patients with a allocated a volunteer befriender immediately and half history of childhood trauma, psychotherapy alone being placed on a waiting list and offered the service at was superior to pharmacotherapy and the combined follow-up a year later. Remission was attained by 65% treatment was only marginally superior to psy- of the women allocated to befriending and 39% of the chotherapy alone. This suggests that psychotherapy waiting list controls – a statistically significant dif- is an essential element in treating patients with ference of modest effect size [95]. Fresh-start events chronic depression who have histories of childhood and standard attachment style were found to increase trauma [97]. 16.5 CONCLUSIONS The studies described in this chapter converge on a 2. Galea, S. et al. (2007) Urban neighborhood poverty and plausible biopsychosocial model for the aetiology of the incidence of depression in a population-based co- depression. The vast majority of the research behind hort study. Annals of Epidemiology, 17 (3), 171–179. this model comes from observational studies, yet many 3. Lewis, G. et al. (1998) Socioeconomic status, standard of of the elements of the various pathways are open to living, and neurotic disorder. Lancet, 352, 605–609. testing through experimental studies, of which there 4. Ostler, K. et al. (2001) Influence of socio-economic are only a handful of examples to draw upon. Studies deprivation on the prevalence and outcome of depres- are underway that will test the impact of ‘parenting’ sion in primary care: the Hampshire Depression Pro- ject. British Journal of Psychiatry, 178 (1), 12–17. programmes, of supportive interventions for young people leaving care and of social factors as mediators 5. Bruce, M. L. and Hoff, R. A. (1994) Social and physical health risk factors for first-onset major depressive dis- inthepharmacologicaltreatmentofdepression.Design- order in a community sample. Social Psychiatry and ing such methodologically rigorous experiments invol- Psychiatric Epidemiology, 29 (4), 165–171. ving manipulation of the social environment is a 6. Weissman, M. M. (1987) Advances in psychiatric particularlychallengingtask,buttheessentialnextstep. epidemiology: rates and risks for major depression. American Journal of Public Health, 77 (4), 445–451. REFERENCES 7. Radloff, L. S. (1975) Sex differences in depression: the effect of occupation and marital status. Sex Roles, 1, 1. Lorant, V. et al. (2003) Socioeconomic inequalities in 249–265. depression: a meta-analysis. American Journal of Epi- 8. Nazroo, J. Y., Edwards, A. C. and Brown, G. W. (1997) demiology, 157 (2), 98–112. Gender differences in the onset of depression following

DEPRESSION 223 a shared life event: a study of couples. Psychological tion of onsets of major depression and generalised Medicine, 27, 9–19. anxiety. Archives of General Psychiatry, 60, 789–796. 9. Brown, G. W. and Harris, T. O. (1978) Social Origins of 24. Farmer, A. et al. (2000) Cardiff depression study. A sib- Depression: A Study of Psychiatric Disorder in Women, pair study of life events and familiality in major depres- Tavistock Press, London. sion. British Journal of Psychiatry, 176, 150–155. 10. Paykel, E. S. (1983) Methodological aspects of life 25. Ormel, J., Oldehinkel, A. J. and Brilman, E. I. (2001) events research. Journal of Psychosomatic Research, The interplay and etiological continuity of neuroticism, 27 (5), 341–352. difficulties, and life events in the etiology of major and 11. Bebbington, P. et al. (1984)Misfortune and resilience: a subsyndromal, first and recurrent depressive episodes community study of women. Psychological Medicine, in later life. American Journal of Psychiatry, 158 (6), 14, 347–363. 885–891. 12. Campbell, E., Cope, S. and Teasdale, J. (1983) Social 26. Frank, E. et al. (1994) Life events and the research factors and affective disorder: an investigation of diagnostic criteria endogenous subtype. A confirmation Brown and Harris’ model. British Journal of Psy- of the distinction using the Bedford College methods. chiatry, 143, 548–553. Archives of General Psychiatry, 51 (7), 519–524. 13. Costello, C. G. (1982) Social factors associated with 27. Brown, G. W., Harris, T. O. and Hepworth, C. (1994) depression: a retrospective community study. Psycho- Life events and endogenous depression: a puzzle reex- logical Medicine, 12, 329–339. amined. Archives of General Psychiatry, 51, 525–534. 14. Surtees, P. G. et al. (1986) Life events and the onset of 28. Mitchell, P. B. et al. (2003) Severity of stressful life affective disorder: a longitudinal general population events in first and subsequent episodes of depression: study. Journal of Affective Disorders, 10, 37–50. the relevance of depressive subtype. Journal of Affec- 15. Brown, G. W., Bifulco, A. and Harris, T. O. (1987) Life tive Disorders, 73 (3), 245–252. events, vulnerability and onset of depression: some 29. Brown, G. W., Bifulco, A. and Andrews, B. (1990) refinements. British Journal of Psychiatry, 150, 30–42. Self-esteem and depression: 3. Aetiological issues. 16. Brown, G. W., Harris, T. O. and Hepworth, C. (1995) Social Psychiatry and Psychiatric Epidemiology, 25, Loss, humiliation and entrapment among women de- 235–243. veloping depression: a patient and non-patient compar- 30. Bifulco, A. et al. (1998) Predicting depression in ison. Psychological Medicine, 25, 7–21. women: the role of past and present vulnerability. 17. Cooper, Z. and Paykel, E. S. (2003) Social factors in the Psychological Medicine, 28 (1), 39–50. onset and maintenance of depression, in Principles of 31. Bifulco, A., Brown, G. W. and Harris, T. O. (1994) Social Psychiatry (eds D. Bhugra and J. Leff), Childhood experience of care and abuse (CECA): a Blackwell, Oxford, pp. 99–121. retrospective interview measure. Child Psychology and 18. Finlay-Jones, R. and Brown, G. W. (1981) Types of Psychiatry, 35, 1419–1435. stressful life events and the onset of anxiety and depres- 32. Brown, G. W. et al. (2008) Parental maltreatment and sive disorders. Psychological Medicine, 11, 803–815. adulthood cohabiting partnerships: a life-course study 19. Miller, P. M. and Ingham, J. G. (1983) Dimensions of of adult chronic depression–4. Journal of Affective experience. Psychological Medicine, 13, 417–429. Disorders, 110 (1–2), 115–125. 20. Abramson, L. Y., Seligman, M. E. and Teasdale, J. D. 33. Harris, T. O. (1988) Psycho-social vulnerability to (1978) Learned helplessness in humans: critique and depression: the biographical perspective of the Bedford reformulation. Journal of Abnormal Psychology, 87 College studies, in Handbook of Social Psychiatry (1), 49–74. (eds A. S. Henderson and G. D. Burrows), Elsevier, 21. Gilbert, P. (1992) Depression: The Evolution of Power- Amsterdam, pp. 55–71. lessness, Lawrence Erlbaum, Hillside, New Jersey. 34. Goldberg, D. and Goodyer, I. (2005) The Origin and 22. Price, J. et al. (1994) The social competition hypothesis Course of Common Mental Disorder, Routledge, of depression. British Journal of Psychiatry, 164, London. 309–315. 35. Spijker, J. et al. (2002) Duration of major depressive 23. Kendler, K. S. et al. (2003) Life event dimensions of episodes in the general population: results from loss, humiliation, entrapment and danger in the predic- The Netherlands Mental Health Survey and Incidence

224 SOCIAL DETERMINANTS Study (NEMESIS). British Journal of Psychiatry, 181, exploration of the role of environmental experiences in 208–213. the aetiology and course of illness. Psychological Med- 36. Spijker, J. et al. (2001) Determinants of poor 1-year icine, 31 (3), 411–423. outcome of DSM-III-R major depression in the general 49. Bradley, R. G. et al. (2008) Influence of child abuse on population: results of The Netherlands Mental Health adult depression: moderation by the corticotropin-re- Survey and Incidence Study (NEMESIS). Acta Psy- leasing hormone receptor gene. Archives of General chiatrica Scandinavica, 103 (2), 122–130. Psychiatry, 65 (2), 190–200. 37. Keller, M. B. (1994) Depression: a long-term illness. 50. Gunnar, M. R. et al. (2001) Salivary cortisol levels in British Journal of Psychiatry, 165 (suppl. 26), 9–15. children adopted from Romanian orphanages. Devel- 38. Brown, G. W. and Moran, P. (1994) Clinical and opment and Psychopathology, 13 (3), 611–628. psychological origins of chronic depressive episodes: 51. Heim, C. and Nemeroff, C. B. (2002) Neurobiology of I. A community survey. British Journal of Psychiatry, early life stress: clinical studies. Seminars in Clinical 165, 447–456. Neuropsychiatry, 7 (2), 147–159. 39. Brown, G. W., Craig, T. K. and Harris, T. O. (2008) 52. Heim, C. and Nemeroff, C. B. (2001) The role of Parental maltreatment and proximal risk factors using childhood trauma in the neurobiology of mood and the Childhood Experience of Care and Abuse (CECA) anxiety disorders: preclinical and clinical studies. Bio- instrument: a life-course study of adult chronic logical Psychiatry, 49 (12), 1023–1039. depression – 5. Journal of Affective Disorders, 110 (3), 53. Bhagwagar, Z., Hafizi, S. and Cowen, P. J. (2003) 222–233. Increase in concentration of waking salivary cortisol 40. Sullivan, P. F., Neale, M. C. and Kendler, K. S. (2000) in recovered patients with depression. American Jour- Genetic epidemiology of major depression: review and nal of Psychiatry, 160 (10), 1890–1891. meta-analysis. American Journal of Psychiatry, 157 54. Goodyer, I. M. et al. (2000) Recent life events, cortisol, (10), 1552–1562. dehydroepiandrosterone and the onset of major depres- 41. Kendler, K. S. et al. (2006) A Swedish national twin sion in high-risk adolescents. British Journal of Psy- study of lifetime major depression. American Journal chiatry, 177, 499–504. of Psychiatry, 163 (1), 109–114. 55. Harris, T. O. et al. (2000) Morning cortisol as a risk 42. Kendler, K. S. et al. (1995) Stressful life events, genetic factor for subsequent major depressive disorder in adult liability, and onset of an episode of major depression in women. British Journal of Psychiatry, 177, 505–510. women. American Journal of Psychiatry, 152 (6), 56. Gardner, K. L. et al. (2005) Early life experience alters 833–842. behavior during social defeat: focus on serotonergic 43. Plomin, R. et al. (1990) Genetic influence on life events systems. Neuroscience, 136 (1), 181–191. during the last half of the life span. Psychology and 57. Caspi, A. et al. (2003) Influence of life stress on Aging, 5 (1), 25–30. depression: moderation by a polymorphism in the 44. Kendler, K. S. et al. (1993) A twin study of recent life 5-HTT gene. Science, 301 (5631), 386–389. events and difficulties. Archives of General Psychiatry, 58. Uher, R. and McGuffin, P. (2008) The moderation by 50 (10), 789–796. the serotonin transporter gene of environmental adver- 45. Farmer, A. et al. (2002) Neuroticism, extraversion, life sity in the aetiology of mental illness: review and events and depression. The Cardiff Depression Study. methodological analysis. Molecular Psychiatry, 13 British Journal of Psychiatry, 181, 118–122. (2), 131–146. 46. Kendler, K. S., Gardner, C. and Prescott, C.A. (2003) 59. Wilhelm, K. et al. (2006) Life events, first depression Personality and experience of psychological adversity. onset and the serotonin transporter gene. British Jour- Psychological Medicine, 33, 1193–1202. nal of Psychiatry, 188, 210–215. 47. Kendler, K. S., Kuhn, J. W. and Prescott, C. A. (2004) 60. Surtees, P. G. et al. (2006) Social adversity, the ser- Childhood sexual abuse, stressful life events and risk otonin transporter (5-HTTLPR) polymorphism and for major depression in women. Psychological Medi- major depressive disorder. Biological Psychiatry, 59 cine, 34 (8), 1475–1482. (3), 224–229. 48. Kendler, K. S. and Gardner, C. O. (2001) Monozygotic 61. Gillespie, N. A. et al. (2005) The relationship twins discordant for major depression: a preliminary between stressful life events, the serotonin transporter

DEPRESSION 225 (5-HTTLPR) genotype and major depression. Psycho- 75. Webber, M. (2005) Social capital and mental health, in logical Medicine, 35 (1), 101–111. Social Perspectives in Mental Health. Developing So- 62. Jacobs, N. et al. (2006) Stress-related negative affectiv- cial Models to Understand and Work with Mental ity and genetically altered serotonin transporter func- Distress (ed. J. Tew), Jessica Kingsley, London, pp. tion: evidence of synergism in shaping risk of depres- 90–111. sion. Archives of General Psychiatry, 63 (9), 989–996. 76. Weiss, R. S. (1974) The provision of social relation- 63. Kendler, K. S. et al. (2005) The interaction of stressful ships, in Doing Unto Others: Joining, Molding, Con- life events and a serotonin transporter polymorphism in forming, Helping, Loving (ed. Z. Rubin), Prentice-Hall, the prediction of episodes of major depression: a re- Englewood Cliffs, New Jersey, pp. 17–26. plication. Archives of General Psychiatry, 62 (5), 77. Cobb, S. (1976) Presidential Address – 1976. Social 529–535. support as a moderator of life stress. Psychosomatic 64. Taylor, S. E. et al. (2006) Early family environment, Medicine, 38 (5), 300–314. current adversity, the serotonin transporter promoter 78. Haber, M. G. et al. (2007) The relationship between polymorphism, and depressive symptomatology. Bio- self-reported received and perceived social support: a logical Psychiatry, 60 (7), 671–676. meta-analytic review. American Journal of Community 65. Kaufman, J. et al. (2004) Social supports and serotonin Psychology, 39 (1–2), 133–144. transporter gene moderate depression in maltreated 79. Henderson, S. (1981) Social relationships, adversity children. Proceedings of the National Academy of and neurosis: an analysis of prospective observations. Sciences of the United States of America, 101 (49), British Journal of Psychiatry, 138, 391–398. 17316–17321. 80. Kessler, R. C. et al. (1994) Perceived support and 66. Putnam, R. (1996) The strange disappearance of civic adjustment to stress in a general population sample of America. The American Prospect, 7, 34–48. female twins. PsychologicalMedicine,24 (2), 317–334. 67. Lin, N. (2001) Social Capital: A Theory of Social 81. Kessler, R. C. and McLeod, J. D. (1984) Sex differences Structure and Action, Cambridge University Press, in vulnerability to undesirable life events. American Cambridge. Sociological Review, 49, 620–631. 68. Bourdieu, P. (1986) The forms of capital, in Handbook 82. Fuhrer, R. and Stansfeld, S. A. (2002) How gender of Theory and Research for the Sociology of Education affects patterns of social relations and their impact on (ed. J. Richardson), Greenwood Press, New York, pp. health: a comparison of one or multiple sources of 241–258. support from ‘close persons’. Social Science and Med- 69. Lochner, K. A. et al. (2003) Social capital and neigh- icine, 54 (5), 811–825. borhood mortality rates in Chicago. Social Science and 83. Antonucci, T. C. (1994) A life span view of womens Medicine, 56 (8), 1797–1805. social relations, in Women Growing Older (eds B. F. 70. Subramanian, S. V., Kawachi, I. and Kennedy, B. P. TurnerandL.E.Troll),Sage,ThousandOaks,California, (2001) Does the state you live in make a difference? pp. 239–269. Multilevel analysis of self-rated health in the US. Social 84. Brown, G. W. et al. (1986) Social support, self-esteem Science and Medicine, 53, 9–19. and depression. Psychological Medicine, 16, 813–831. 71. McCulloch, A. (2001) Social environments and health: 85. Turner, R. J. and Marino, F. (1994) Social support and cross sectional national survey. British Medical Jour- social structure: a descriptive epidemiology. Journal of nal, 323 (7306), 208–209. Health and Social Behavior, 35 (3), 193–212. 72. Boreham, R., Stafford, M. and Taylor, R. (2002) Health 86. Leenstra, A. S., Ormel, J. and Giel, R. (1995) Positive Survey of England 2000: Social Capital and Health, life change and recovery from depression and anxiety. The Stationary Office, London. A three-stage longitudinal study of primary care atten- 73. Hendryx, M. S. and Ahern, M. M. (1997) Mental health ders. British Journal of Psychiatry, 166 (3), 333–343. functioning and community problems. Journal of Com- 87. Pevalin,D.J.andGoldberg,D.P.(2003)Socialprecursors munity Psychology, 25, 147–157. to onset and recovery from episodes of common mental 74. De Silva, M. J. et al. (2005) Social capital and mental illness. Psychological Medicine, 33 (2), 299–306. illness: a systematic review. Journal of Epidemiology 88. Pagel, M. D., Erdly, W. W. and Becker, J. (1987) Social and Community Health, 59 (8), 619–627. networks: we get by with (and in spite of) a little help

226 SOCIAL DETERMINANTS from our friends. Journal of Personality and Social disorders. A test of specificity. British Journal of Psy- Psychology, 53 (4), 793–804. chiatry, 161, 44–54. 89. Broadhead, J. et al. (2001) Social support and life 94. Brown, G. W. (1993) Life events and affective disorder: events as risk factors for depression amongst women replications and limitations. Psychosomatic Medicine, in an urban setting in Zimbabwe. Social Psychiatry and 55 (3), 248–259. Psychiatric Epidemiology, 36 (3), 115–122. 95. Harris, T., Brown, G. W. and Robinson, R. (1999) 90. Lin, N., Woelfel, M. W. and Light, S. C. (1985) The Befriending as an intervention for chronic depression buffering effect of social support subsequent to an among women in an inner city. 1: randomised con- important life event. Journal of Health and Social trolled trial. British Journal of Psychiatry, 174, Behavior, 26 (3), 247–263. 219–224. 91. Brugha, T. S. et al. (1997) Predicting the short-term 96. Harris, T., Brown, G. W. and Robinson, R. (1999) outcome of first episodes and recurrences of clinical Befriending as an intervention for chronic depression depression: a prospective study of life events, difficul- among women in an inner city. 2: role of fresh-start ties, and social support networks. Journal of Clinical experiences and baseline psychosocial factors in remis- Psychiatry, 58 (7), 298–306. sion from depression. British Journal of Psychiatry, 92. Wade, T. D. and Kendler, K. S. (2000) Absence of 174, 225–232. interactions between social support and stressful life 97. Nemeroff, C. B. et al. (2003) Differential responses to events in the prediction of major depression and de- psychotherapy versus pharmacotherapy in patients pressive symptomatology in women. Psychological with chronic forms of major depression and childhood Medicine, 30 (4), 965–974. trauma. Proceedings of the National Academy of 93. Brown, G. W., Lemyre, L. and Bifulco, A. (1992) Social Sciences of the United States of America, 100 (24), factors and recovery from anxiety and depressive 14293–14296.

17 Common mental disorders Christoph Lauber Department of Population and Behavioural Sciences, University of Liverpool, Liverpool, UK 17.1 DEFINITION: WHAT ARE COMMON MENTAL DISORDERS? Over time, the language used to describe depression, such as anxiety and affective well-being [3]. When anxiety and related common psychiatric disorders has compared to disorders usually referred to as ‘severe changed, especially in English-speaking countries. mental illness’, it may be interpreted that CMDs Increasingly, the term ‘common mental disorders’ cannot reach a serious illness level. Although severe (CMDs) has come into general use and includes mental illnesses are often highly disabling and there- depressive disorders, anxiety disorders, obsessive- fore cause a huge personal and societal burden, CMDs compulsive disorders and phobias [1]. CMDs are can be as disabling and handicapping as severe mental commonly encountered in community settings, and illness. The significance of CMDs for public health their occurrence signals a breakdown in normal func- stems primarily from the recognition that they are tioning. Also referred to as nonpsychotic or neurotic much more prevalent and, thus, are responsible for the disorders they often manifest with a mixture of so- majority of illness burden worldwide. Global esti- matic, anxiety and depressive symptoms. This implies mates for developed and developing countries suggest the exclusion of organic disorders like dementia, that the burden caused by CMDs is similar to, and in confusional states and delirium, schizophrenia and many cases greater than, that of many physical health other psychotic disorders, and substance abuse [2]. disorders [4]. CMDs are, hence, one important cause Following from this, it is a widely held belief that of ill-health, social disability and health service use CMDs are minor problems of general mental health and, thus, pose a significant public health problem [5]. 17.2 EPIDEMIOLOGICAL FINDINGS The WHO multinational study of the prevalence, rities across centres included general consensus on the nature and determinants of CMDs in general medical ubiquity of CMDs, the comorbidity of anxiety and care settings was conducted in 14 countries [6]. De- depression, and the association of CMDs with disabil- spite enormous differences in most variables, simila- ity. These results are supported by different surveys Principles of Social Psychiatry, second edition Edited by Craig Morgan and Dinesh Bhugra Ó 2010 John Wiley & Sons, Ltd.

228 SOCIAL DETERMINANTS in Western countries. For instance, the National proportion of adults with at least one CMD was 14%, Comorbidity Survey (NCS) Replication Study from while in 2000 the proportion was 14.5%. However, the US reported that 28.8% of people have had some there was a slight but significant increase in the kind of CMD during their lifetime (‘any anxiety dis- prevalence of any CMD among men, from 10.7% in order’ according to DSM-III-R) and, of these, 18.1% 1993 to 11.8% in 2000 [8], with an increase mostly in had experienced a CMD in the last 12 months [7]. The mixed anxiety and depressive disorders in those aged Psychiatric Morbidity Survey (PMS) of 2000 [8] found 35 to 54. In contrast, the NCS Replication Study found that about one in six adults (16.4%) in the UK were an increase of the lifetime prevalence of CMDs by assessed as having a CMD in the week before the nearly 4% over one decade [7,9]. interview. The most prevalent CMD among the popu- Although the diagnostic (American surveys usually lation as a whole was mixed anxiety and depressive use the Structured Clinical Interview for DSM-IV disorder(8.8%).Generalizedanxietydisorderwasnext (SCID), the PMS used the Clinical Interview most commonly found (4.4%). The remaining disor- Schedule – revised) and the time criteria (the PMS ders were less prevalent, ranging from 0.7% (panic usedatimeframeof1weekpriortotheinterviewwhile disorder) to 1.8% (phobias). theNCSaskedfora12-monthandlifetimeprevalence) The overall prevalence for anyCMD for all adults in used in these surveys may differ and, thus, the data are the UK remained stable between 1993 and 2000 (the not easily comparable, the numbers of people affected years when the PMSs were conducted): in 1993 the gives a clear indication of the size of the problem. 17.3 RISK FACTORS FOR COMMON MENTAL DISORDERS Sociodemographic variables are consistently reported domestic problems [12]. Their multiple roles include to be associated with CMD. In brief, people with child-bearing, child-rearing and running the family CMDs are more likely to be female, aged between home, and in an increasing proportion of families 35 and 54 years old, separated or divorced, and of earning income is likely to lead to considerable stress. ‘lower social position’ ([10]; [11], p. 14), including The reproductive roles of women, the consequences of living as a one-person family or as a lone parent, being infertility, postnatal depression and in some cultures the unemployed or economically inactive, living in an failure to produce a male child are examples of mechan- economically deprived area, and having poor material isms that make women vulnerable to CMDs. Moreover, circumstances and lower education [8,12,13]. violence against women is emerging as a pervasive global issue and contributes significantly to preventable morbidity and mortality for women across cultures. 17.3.1 Gender The PMS (2000) showed that prevalence rates were 17.3.2 Age higher among women than men for all CMDs except panic disorders, where therewas no gender difference. In the absence of cohort effects, differential mortality The disparity between the rates for women and men and age-related differences in the willingness to report was significant for phobias and mixed anxiety and symptoms,onewouldexpecttofindincreasinglifetime depressive disorders. Similar patterns have also been prevalence of all disorders with age [9]. However, the found in non-Western cultures [14,15]. NCS and the Epidemiologic Catchment Area (ECA) Explanations for this frequently invoke social posi- Study showed the group aged 25 to 34 years as having tion and processes. For example, an explanation for the the highest prevalence with declining figures at later high prevalence of CMDs in women may be the ages, while the highest prevalence in the PMS (2000) difficulties they encounter in a number of different was in the group aged 40–54 years (16%). The lowest areas, such as their social position, aspirations and prevalence rates of any CMD were found among older

COMMON MENTAL DISORDERS 229 people. The prevalence among those aged 65–69 was 17.3.5 Socioeconomic factors 9.6% and among those aged 70–74 was 8.3%. In line with this, Melzer et al. [10,11] found that the There are some results that describe a relationship prevalence rates for having any CMD in men aged between CMD and socioenvironmental factors. The 65–69 years (5%) were dramatically lower than in the NCA reported that respondents living in rural areas age group 60–64 years (14.5%). Prevalence rates in had a 40% lower risk of 12-month comorbidity than women peaked at age 50 to 54 years (21.3%) and their urban counterparts [9]. The PMS (2000) con- declined thereafter, but no large changes in prevalence firmed this finding for the UK and additionally are evident around age 60 or 65 years. In men leaving showed some geographical variability, with the work early (before the age of 65 years) the prevalence Northwest of England (200 per 1000) and London of CMDs remains high until the conventional retire- (182 per 1000) having the highest prevalence of ment age. The authors concluded that in the general CMDs. Regions with prevalence rates lower than the population aged 50–74 years, there is a dramatic UK average were the West Midlands, Scotland and improvement in mental health in men after the con- parts of the South. ventional retirement age, but not in women. Rates of all CMDs are declining with the increase of employment, income and education. This holds true for the UK, the US, the Netherlands and Australia 17.3.3 Living circumstances ([8–10]; [11], p. S7). The association between CMD and economic deprivation is supported by another People with CMDs are more likely to be separated or result from the PMS (2000) in that people with a divorced and to live as a one-person family or as a lone CMD, when compared with those who have no such parent. In the PMS (2000), lone mothers were twice as disorder, are more likely to be tenants of local autho- likely to have a CMD as other women. This was not rities and housing associations and have moved three significant after controlling for financial strain or or more times in the last two years. Finally, a study social support. Lone fathers were nearly four times from Chile found a negative association between the more likely to have a CMD than other men, and this quality of the built environment of small geographical risk remained undiminished by controlling for age, sectors and the presence of CMDs among its income, debt and levels of social support [13]. Thus, residents [17]. debt management would be a rational strategy to reduce psychiatric morbidity in lone parents. 17.3.6 Comorbidity 17.3.4 Ethnicity The Epidemiologic Catchment Area (ECA) Study, conducted in the US in the 1990s, showed that The PMS (2000) reports clear ethnic differences in the comorbidity among mental disorders is high. More prevalence of CMDs in the UK. While the overall than 60% of the ECA respondents with at least one prevalence was 16.4%, South Asians (Indian, Pakis- lifetime disorder had two or more disorders [9]. tani and Bangladeshi) were reported to have a pre- Similar results are reported from the NCS Replica- valence of 19.2% and whites one of 16.3%. Black tion Study [7]. The same holds true for the comor- people (black Caribbean, black African and other bidity of CMDs with physical illnesses. In the PMS black groups) have the lowest prevalence with (2000), 42% reported a physical complaint (all the 14.1%. All other ethnicities have a prevalence of following data referring to CMDs include people 20.4%. This partly contradicts results from the US with a depressive episode!), with women being where neither the NCS nor the ECA found any black– slightly more likely than men to report a longstanding white or Hispanic–white differences for panic disor- illness (43% versus 40%). There is a clear relation- ders [16], but an increased prevalence for simple ship between the number of CMDs and the reporting phobia and agoraphobia in black people [9]. of a physical complaint. Just under two-fifths of

230 SOCIAL DETERMINANTS adults with no CMD (38%) report having a physical Haynes et al. [18] examined whether excessive complaint. This rises to over half of those with one alcohol consumption is a risk factor for anxiety and CMD (57%), while among those with two or more depression in the general population, and whether CMDs, two-thirds (67%) report at least one physical anxiety and depression are risk factors for excessive complaint. The prevalence of musculoskeletal or skin alcohol consumption. They found that excessive alco- complaints, complaints of the digestive system, the hol consumption was not associated with the onset of nervous system and the genitourinary system is al- anxiety and depression, but abstinence was associated most twice as high in people with CMDs when withalowerrisk.Subthresholdsymptomswereweakly compared to those without. associated with new-onset alcohol dependence. 17.4 MENTAL HEALTH SERVICE USE The majority of people with CMDs do not receive any anxiety and depressive disorder were least likely to be professional treatment at all. This has been found in receiving counselling or therapy (5%). The most both the NCS and PMS. The PMS (2000) showed that common types of therapy were counselling (4%) and just 24% of people with a CMD (data including psychotherapy (3%). Behavioural or cognitive ther- depressive episodes) were receiving treatment of apy was being given to 1% of people with CMDs. some kind for mental or emotional problems at the Those with obsessive-compulsive disorder (12%) and time of interview, with 20% taking psychoactive phobias (11%) were the most likely to be treated by medication, while 9% were receiving counselling or psychotherapy. These groups were also the most likely therapy. A small proportion (4%) was receiving both to have behavioural or cognitive therapy, 5% of those forms of treatment. The proportion receiving treat- with obsessive compulsive disorder and 3% of those ment rose with the number of CMDs present. with phobias. Counselling was most often prescribed The PMS (2000) showed that the group most likely for people with phobias (15%) and obsessive-com- to be receiving treatment were those with phobias. pulsive disorder (10%) (PMS, 2000). Over half of this group (54%) were receiving treat- In the previous year almost two-fifths of those ment in some form, with 27% receiving medication included in the PMS (2000) sample with CMDs only, 9% receiving therapy or counselling alone and (39%) had spoken to their GP about a mental or 18% receiving both forms of treatment. Least likely to emotional problem, compared with 6% of those with- be receiving any treatment were those with mixed out a CMD. There was a clear association between the anxiety and depressive disorder (16%). number of disorders and the likelihood of having Antidepressants were the most common psychoactive spoken to a GP. Most likely to have seen a GP in the medication being used (PMS, 2000), being prescribed to last two weeks were those with phobias (19%), panic 16% of peoplewith a current CMD, mostly to thosewith disorder (14%) and obsessive-compulsive disorder phobias (40%) and obsessive compulsive disorder (13%). Those with mixed anxiety and depressive (30%). Again, there was an increase in the proportion disorder were the least likely to have talked to a GP ofpeoplebeingprescribedantidepressantsrelatingtothe about an emotional or mental problem (3%), while 3% number of disorders present. Overall, 6% of respondents of those with a CMD visited an outpatient department with CMD were taking hypnotics or anxiolytics. These for treatment or a checkup because of a mental or were most commonly prescribed to those with phobias emotional problem in the last three months, compared (17%) and generalized anxiety disorder (14%). with less than 1% of those without a disorder. Just under a tenth (9%) of people with CMDs were Those with phobias (13%) were almost twice as receiving counselling or therapy, compared with 1% likely as those with any other disorder to have visited a of thosewith no CMD. Most likely to be treated by this hospital as an outpatient for a psychiatric problem. method were those with phobias (27%) and obsessive Among thosewith obsessive-compulsive disorder, 7% compulsive disorder (20%), while those with mixed had made an outpatient visit for their problems, while

COMMON MENTAL DISORDERS 231 6% of those with generalized anxiety disorder had services were nursing services; 4% of people with a done so. Among respondents assessed as having a CMD had seen a social worker, compared with 1% of CMD, 16% had used one or more of the community those with no CMD, while 3% of people with a CMD care services in the last year, compared with 4% of mentioned using a psychiatrist, psychologist, home those with no CMD. The most frequently used help or care worker in the previous twelve months. 17.5 THE ECONOMIC IMPACT OF COMMON MENTAL DISORDERS There are relatively few data available on the eco- costs and £1.2 billion from service costs. As about nomic impact of CMDs although a considerable half of the people with anxiety disorders are not in number of the days taken off work are related to contact with services and of those who are, 46% do these disorders. A recently published study by Das- not receive medication or psychological therapy, Munshi et al. [19] showed that a fifth of all days off these costs might grow considerably higher in the work in the UK occurred in people with mixed near future, if those in contact with services but who anxiety and depressive disorder. In a report on the are not currently receiving treatment do receive any cost of mental health care in England [20], the total kind of therapy. These costs, however, will be offset costs of CMDs (data without both post-traumatic by reduced lost employment costs. If combined stress disorder and mixed anxiety and depressive medication and psychological therapy are used, disorder) were approximately £8.9 billion in 2007. there will be less of an offset and costs may actually This includes £7.7 billion from lost employment exceed benefits. 17.6 AN INTERNATIONAL PERSPECTIVE ON THE RELATIONSHIP BETWEEN SOCIETY, CULTURE AND COMMON MENTAL DISORDERS Most research on CMDs has been carried out in Wes- indicators of poverty such as poor housing or low tern countries. However, as CMDs are a major con- income. Finally, Pothen et al. [23] reported from India tributortotheburdenofdiseaseindevelopingcountries that indicators of low socioeconomic status, i.e. pov- as well, some studies have emerged from outside the erty (e.g. being in debt or inability to buy food) and Western world. These results do not necessarily differ illiteracy were significantly associated with CMDs from those reported in Western countries. Even more, seen in primary health care settings. These findings theymightcontributeto ashared worldwideand cross- suggest that the association between poverty and cultural understanding of factors associated with risk CMDs is a universal one, occurring in all societies for and course of CMDs. irrespective of their levels of development. 17.6.1 Poverty 17.6.2 Violence Different studies have found that numerous indicators Ludermir et al. [24] reported from Brazil that a large of poverty are strongly associated with CMDs. For proportion of women are exposed toviolence (50.7%). instance, Patel et al. [12] compared data from India, The most frequent forms were psychological violence Zimbabwe, Chile and Brazil and found a relationship alone (18.8%) or accompanied by physical violence between CMDs, low education and poverty. These (16.0%). Most forms of violence studied (physical, findings were replicated by Ludermir and Lewis [21] psychological or sexual violence alone or mixed in a Brazilian sample. A literature review by Patel and forms) were significantly associated with mental dis- Kleinman [22] described a relationship of CMDs with orders. The prevalence of CMD was 49.0% among

232 SOCIAL DETERMINANTS women who reported any type of violence and 19.6% common in Caribbean and African groups, anxiety among those who did not report violence. These more common in Irish-born populations and phobias findings confirmed previous results by Lopes more common in Asian and Oriental groups [27]. et al. [25] from Brazil. Different conceptualizations of health and disease may contribute to misrecognition of CMD in ethnic 17.6.3 Education and employment minorities.Forinstance,inastudyfromSouthLondon, Bhui et al. [28] found that the prevalence of CMD and somatic symptoms does not differ across cultures. A study from Chile found a strong, inverse and Among English subjects, general practitioners were independent association between education and more likely to identify correctly pure psychiatric ill- CMDs [17]. The prevalence of CMDs was also higher ness and mixed pathology, but Punjabi subjects with amongst people with manual unskilled occupations, common mental disorders were more often assessed as overcrowded housing and lower per capita income. having ‘subclinical disorders’ and ‘physical and Similar results have been found in other Latin Amer- somatic’disorders.Punjabicaseswithdepressiveideas ican studies [26]. were less likely to be detected compared with English ones. English women were less well detected than 17.6.4 Migrants English men by Asian general practitioners. Help- seeking English subjects were more likely to be cor- CMDs in ethnic minorities have been the subject of rectlyidentifiedascases.Thismis-orunderrecognition very little research. In general, CMDs appear to be of CMDs in ethnic minorities may have significant more prevalent in many ethnic minority groups in healthserviceimplications.Forexample,itmayleadto the UK compared with the general population, with higher service use due to unexplained symptoms, but one report suggesting mood disorders are more lower rates of effective treatment [27]. 17.7 COMMON MENTAL DISORDERS AND THE SOCIAL ENVIRONMENT People suffering from a CMD impact on their social indicators of CMDs is consistent with studies from environment (as well as vice versa). For example, other developing countries. For instance, Harpham Stewart et al. [29] found an association between ma- et al. [31] detected in their community-based cross- ternal CMDs and infant growth impairment in rural sectional survey in Ethiopia, India, Vietnam and Peru Malawi–themeanlength-for-ageforinfantsofmothers high levels of maternal CMD and child malnutrition in with a CMD was significantly lower than for infants of each study setting. While there was a relation between mothers without CMD. The same holds true for mean high maternal CMD and poor child nutritional status in weight-for-age, but this difference was not significant. IndiaandVietnam,thefindingsfromPeruandEthiopia, Hadleyetal.[30]foundthatfoodinsecurity,stressful however, do not provide clear evidence for a similar life events and symptoms of CMD were highly pre- association being present in non-Asian countries. valent in sub-Saharan Africa. Food insecurity and These findings suggest that the negative effects of stressfullifeeventswereindependentlyassociatedwith food insecurity extend beyond nutritional outcomes high symptoms of CMDs. The finding that food in- and that interventions to promote food security may security and stressful life events are associated with also positively influence adult mental health. 17.8 PUBLIC HEALTH IMPLICATIONS CMDs are highly prevalent. However, this has so far health professionals, research studying CMDs in re- not sufficiently influenced the attitudes of mental lation to influencing and explaining factors, and