Migration, ethnicity and psychoses: evidence, models and future directions

Abstract

There is a large body of research reporting high rates of psychotic disorders among many migrant and minority ethnic groups, particularly in Northern Europe. In the context of increasing migration and consequent cultural diversity in many places worldwide, these findings are a major social and public health concern. In this paper, we take stock of the current state of the art, reviewing evidence on variations in rates of psychoses and putative explanations, including relevant theories and models. We discuss in particular: a) the wide variation in reported rates of psychotic disorders by ethnic group, and b) the evidence implicating social risks to explain this variation, at ecological and individual levels. We go on to set out our proposed socio‐developmental model, that posits greater exposure to systemic social risks over the life course, particularly those involving threat, hostility and violence, to explain high rates of psychoses in some migrant and minority ethnic groups. Based on this analysis, the challenge of addressing this social and public health issue needs to be met at multiple levels, including social policy, community initiatives, and mental health service reform.

A defining feature of the modern world is large scale migration, both within and between countries, one consequence of which is increasing ethnic and cultural diversity in many places. In this context, repeated reports that some migrant and minority ethnic populations experience high rates of psychotic disorders are particularly concerning.

In this paper, we take stock of the current evidence and related debates, focusing on variations in incidence of psychoses and related explanations. We draw some, we hope, thought‐provoking conclusions about the socio‐developmental, and ultimately structural, roots of ethnic disparities and how these relate to other reports of poor clinical and social outcomes and more negative interactions with mental health services in these populations.

The pressing challenge is how to harness what we know a‐bout ethnic disparities to develop social policies, community initiatives, and mental health services to address this major public health issue.

VARIATIONS BY ETHNIC GROUP: INCIDENCE OF PSYCHOTIC DISORDERS

There is an extensive literature from the past sixty years reporting high rates of psychotic disorders among several minority ethnic populations in high‐income countries. Several overlapping systematic reviews and meta‐analyses have summarized these findings1, 2, 3, 4, 5, 6, 7, 8, 9 (see Table 1). Overall, these reviews suggest that the incidence of psychotic disorders in all minority ethnic populations combined is around 1.5 to 3.0 times the incidence in majority populations.

Table 1.

Summary of findings from meta‐analyses, showing overall rate or risk ratio (RR) for all psychotic disorders, unless otherwise specified, in minority vs. majority ethnic groups

	Overall		First generation		Second generation		Men		Women
	RR	95% CI	RR	95% CI	RR	95% CI	RR	95% CI	RR	95% CI
Scope: International
McGrath et al1 (schizophrenia)
Migrants, minorities (vs. majority)	4.6	1.0‐12.9
Cantor‐Graae & Selton2 (schizophrenia)
Migrants, minorities (vs. majority)	2.9	2.5‐3.4	2.7	2.3‐3.2	4.5	1.5‐13.1	2.5	2.0‐3.2	2.4	1.8‐3.1
From developed countries	2.3	1.7‐3.1
From developing countries	3.3	2.8‐3.9
From areas where majority population is White	2.3	1.8‐3.0
From areas where majority population is Black	4.8	3.7‐6.2
From areas where majority population is neither	2.2	1.6‐3.0
Bourque et al3
Migrants, minorities (vs. majority)			2.3	2.0‐2.7	2.1	1.8‐2.5
Men			2.1	1.7‐2.6	2.5	1.8‐3.4
Women			2.4	1.9‐2.9	3.0	2.1‐4.2
White			1.8	1.6‐2.1	2.3	2.1‐2.7
Black Caribbean			3.9	3.4‐4.6	5.8
Black African			4.3	2.8‐6.8	3.7	2.2‐6.3
Asian			1.7	1.3‐2.3	1.3	0.8‐2.1
Middle Eastern			2.3	1.4‐4.0	2.3	1.4‐4.0
Castillejos et al4
Migrants, minorities (vs. majority)	3.1	2.7‐3.5
Schizophrenia	2.7	2.0‐3.7
Non‐affective psychoses	3.1	2.6‐3.6
Affective psychoses	1.3	1.2‐1.3
Jongsma et al5
Migrants, minorities (vs. majority)	1.8	1.5‐2.0
Schizophrenia	1.4	1.2‐1.7
Non‐affective psychoses	1.7	1.4‐2.1
Selten et al6
Migrants, minorities (vs. majority)	2.1	2.0‐2.3
From developed countries	1.7	1.5‐1.8
From developing countries	2.5	2.3‐2.9
White	1.7	1.5‐1.9
Black	4.2	3.4‐5.1
To UK	2.7	2.2‐3.3
To Scandinavia	1.9	1.8‐2.0
To Netherlands	3.0	2.4‐3.7
To Southern Europe	2.8	1.9‐4.0
To Canada	1.2	0.9‐1.7
To Israel	1.2	1.0‐1.5
To Australia	2.1	1.2‐3.8
Refugee	1.9	1.6‐2.2
Non‐refugee	1.8	1.6‐2.9
Scope: Netherlands, Belgium, France, Italy
van der Ven et al7 (non‐affective psychoses)
Migrants, minorities (vs. majority)							2.9	2.7‐3.2	2.6	2.2‐3.1
From the Maghreb							2.9	2.0‐4.1	1.4	0.7‐2.6
From Asia							1.1	0.7‐1.6	0.9	0.3‐2.8
From Central and South America							3.0	1.7‐5.3	3.2	2.1‐4.8
From Western countries							1.3	0.9‐1.9	1.4	0.9‐2.3
From Sub‐Saharan Africa							4.9	2.4‐10.1	4.8	1.6‐14.3
From Morocco							3.2	2.0‐5.0	1.6	0.8‐3.1
Scope: UK
Tortelli et al8 (schizophrenia)
Black Caribbean vs. majority (White)	4.7	3.9‐5.7
Kirkbride et al9 (schizophrenia)
Black Caribbean vs. majority (White)	5.6	3.4‐9.2
Black African vs. majority (White)	4.7	3.3‐6.8
Asian vs. majority (White)	2.4	1.3‐4.5

However, these overall estimates can be misleading. The extent to which rates are elevated, relative to the majority population, varies considerably. For example, the highest reported rates are among Black minority groups (i.e., 4 to 6 times higher than in majority groups)6. This finding is driven in part by studies from the UK that report especially high rates for Black Caribbean and Black African populations8. The evidence for other minority ethnic groups in the UK suggests that rates are either not increased or only modestly so (e.g., around 1.5 times, at most, for White non‐British and Asian populations)9, 10, 11.

Variations are also evident in other countries. For example, in the Netherlands, rates are particularly high for Moroccan and Surinamese populations, but less so for Turkish12, 13.

These variations should not be surprising. These are populations with different migratory histories and cultural heritages, living in diverse social contexts, and occupying varying social positions. The place of migration may also matter. In the most recent review, for example, there was no strong evidence that rates of psychotic disorders were elevated in migrants to Canada or Israel6.

Further, there may be variations in relative risks by gender. In a study in East London11, evidence was found that rates of psychotic disorders may be specifically elevated among women from Pakistan (incidence rate ratio, IRR of 3.1) and Bangladesh (IRR of 2.3). In the Netherlands, there is strong evidence from several studies that the incidence of psychoses is substantially higher among men from minority ethnic groups compared with women, especially among those from the Maghreb (Morocco, Algeria, Lybia, Tunisia), with a ratio as high as 5:17.

More recent studies have reinforced the complexity of patterns of risk. For example, analyses of data from our study of psychoses in urban and rural sites in five European countries14 found marked variations in the extent of elevated risk in minority ethnic populations depending upon setting. Further, a study using Swedish register data suggests that the incidence of psychotic disorders may be particularly high among refugees relative to other migrants (i.e., IRR of 2.9 for refugees vs. 1.7 for other migrants), a difference that was more pronounced among men15.

Moreover, incidence rates may change over time. In our most recent study16, in which we compared incidence rates in South London between 1997‐1999 and 2010‐2012, we found notable changes in incidence by ethnic group. For example, rates in the White British population increased from 20 per 100,000 in 1997‐1999 to 39 per 100,000 in 2010‐2012. By contrast, rates in the Black Caribbean population declined from 141 per 100,000 in 1997‐1999 to 94 per 100,000 in 2010‐2012. As a consequence, the relative risk for the Black Caribbean population decreased considerably (from 6.7 to 2.8).

These differences over time may, of course, be due to methodological inconsistencies in, for example, case detection. The main point, however, is that we should not assume that rates and rate ratios are constant over time. There is no single, universal, time‐invariant migrant effect, as is often implied. Variation is the norm. This further implies that there is unlikely to be one simple explanatory factor that can account for such patterns. A multifaceted explanation is required.

A final note on the current epidemiological evidence. Most studies have been conducted in Northern Europe – in fact, in the UK and the Netherlands – with some studies in Southern Europe, the US, Canada and Israel. We do not know to what extent these findings generalize to other countries and contexts, in particular in the global South, where most migration occurs. Recent work on urbanicity and psychosis is a timely reminder that what we find in high‐income settings does not necessarily extend to other contexts17.

Moreover, we know very little about the effects of internal migration, which often involves movement over considerable distances and to very different social and cultural contexts. In a study in Northern Italy18, incidence rates for psychotic disorders were similarly elevated for internal migrants (IRR=1.93, 95% CI: 1.19‐3.13), predominantly from Southern Italy, and external migrants (IRR=1.79, 95% CI: 1.06‐3.02).

Methodological artefacts

Since the earliest studies, the validity of reports of high rates of psychotic disorders, particularly in Black minority groups, has been the subject of much debate. There are several potential biases that could create spurious differences. These relate, for example, to selective migration, case identification, and under‐enumeration of minority denominator data.

However, without entirely discounting these potential biases, it is unlikely that those noted can fully account for the observed patterns. For instance, various lines of reasoning suggest that selective migration is unlikely. In an intriguing thought experiment, Selten et al19 re‐calculated incidence rates for the Surinamese population in the Netherlands, assuming that the entire population of Suriname had migrated. After doing this, rates were still elevated relative to the Dutch majority.

Further, in more recent studies, methods for identifying cases have become more varied and comprehensive (e.g., use of case registers, multiple sources), and denominator data for minority groups are more accurate, certainly in the UK. However, reports of disparities in incidence persist.

Misdiagnosis

More challenging is the suggestion that high rates result from systematic misdiagnosis of psychosis in minority ethnic groups20, 21. This possibility merits careful consideration. In addressing this issue, it is worth stepping back to consider why these findings initially proved so controversial.

The earliest reports in the UK, where most studies initially originated, focused on schizophrenia. At that time, in the late 1970s and early 1980s, the prevailing consensus within psychiatry was that schizophrenia was primarily a genetic brain disorder. To link ethnicity or race with an increased risk of a heavily stigmatized disorder that was considered to be primarily genetic understandably provoked a reaction. It was uncomfortably close to racist ideologies of the genetic inferiority of Black people20. Considered from this perspective, the over‐diagnosis of schizophrenia among Black people stemmed from wider stereotypes which, when refracted through the lens of psychiatry, led to the pathologization of culturally grounded beliefs, behaviours and expressions of distress as signs of psychosis.

For example, Littlewood and Lipsedge21 argued that acute distress arising from difficult conditions and life experiences was systematically misdiagnosed as schizophrenia in the Black Caribbean population in the UK. Echoes of this history can be heard today, particularly outside of academia, in the framing of this issue as one of mis‐ or over‐diagnosis.

To be sure, as recent evidence and trends re‐highlight (see below), the nature of psychoses is clearly such that diagnosis is often challenging, especially across diverse cultural groups. Further, given that low‐level psychotic experiences are common and frequently co‐occur with symptoms of depression, anxiety and post‐traumatic stress disorder22, 23, it is plausible that predominantly affective disorders may sometimes be misdiagnosed as psychotic disorders. What is more, there is some direct evidence that misdiagnosis does occur in relation to some minority populations (e.g., in the US)24. Certainly, this is an important clinical issue, with implications for management and treatment.

However, there are lines of reasoning and evidence that, on balance, suggest that variations in incidence by ethnic group are not simply an artefact of mis‐ or over‐diagnosis. For example, there have been several attempts to assess the extent to which stereotyping and misdiagnosis occur and might explain reported variations in incidence rates. Two studies used vignettes to investigate racial stereotyping in diagnosis. Neither found strong evidence that psychiatrists are more likely to diagnose schizophrenia when the ethnicity of individuals described in case vignettes is Black25, 26.

Similarly, Hickling et al26, in a study that compared diagnoses made by British and Jamaican psychiatrists in the same patients, found no differences in the percentage of Black patients diagnosed with schizophrenia. Further, recent studies focus on all psychotic disorders, not just schizophrenia, and tend to report high rates for all disorders. This is not, then, an issue of specifically mis‐ or over‐diagnosing schizophrenia.

Finally, in the past 20 years, our understanding of the nature and aetiology of psychoses has changed considerably. It is now clear that schizophrenia and other psychotic disorders are shaped by a complex array of factors, including social conditions and experiences, that combine and interact over time to increase risk. High rates of psychoses in some migrant and minority ethnic groups does not, then, imply an excess of a genetic or purely biologically induced brain disorder.

VARIATIONS BY ETHNIC GROUP: PREVALENCE OF PSYCHOTIC EXPERIENCES

In parallel with research on psychotic disorders in migrant and minority ethnic groups, substantial evidence has emerged in recent years that low‐level psychotic (or anomalous and unusual) experiences – such as fleeting and non‐distressing hallucinations, suspiciousness, and magical thinking – are somewhat common in the general population27, 28.

This raises the possibility that psychotic experiences are continuously distributed, varying in frequency, severity and intensity, with disorder at the extreme end of this distribution. If this is so, in populations with high rates of disorder, we would also expect low‐level psychotic experiences to be more common (see Figure 1). There are now several studies suggesting that this is indeed the case30, 31, 32.

Figure 1.

Hypothesized rightward shift in the continuum of psychosis in some minority ethnic groups

In a recent systematic review and meta‐analysis, Tortelli et al33 identified 19 studies of adults (age 16 or over) that have reported data on psychotic experiences in migrant or minority ethnic groups. The most consistent finding was that people from Black groups, compared with majority groups, more often reported psychotic experiences, both current (from 7 studies with 26 effect sizes: odds ratio, OR=1.8, 95% CI: 1.4‐2.3) and lifetime (from 4 studies with 9 effect sizes: pooled OR=1.3, 95% CI: 1.1‐1.6).

In line with this, in a survey we conducted in the same area of South London where many studies of ethnicity and psychotic disorder have been conducted34, we found that, compared with White British, people from Black Caribbean and Black African populations were more likely to report psychotic experiences (see Figure 2).

Figure 2.

Prevalence of psychotic experiences by ethnic group in a community survey in South London

AN EXPLANATORY FRAMEWORK

In considering why there are variations in the occurrence of psychotic disorders by ethnic group, it is useful to have in mind a framework for psychotic disorders in general.

Psychoses are highly heterogeneous – in their symptomatology, course and outcome – and our current diagnostic categories, at best, capture syndromes that may comprise multiple underlying disorders35. Moreover, these disorders, as noted above, may be the extreme, distressing end of a spectrum of beliefs and perceptual experiences that are somewhat common in populations.

As noted, the underpinning aetiological architecture of psychoses, across the spectrum, is complex. An array of factors, that are neither necessary nor sufficient alone to cause disorder, are associated with an increased risk, spanning genetic, neurobiological, substance use, psychological and social domains36. For example, several social factors have been implicated at area (urbanicity, social fragmentation, ethnic density) and individual (bullying, abuse, life events, discrimination) levels37. That none are sufficient or necessary means that multiple factors must co‐participate over time – no doubt in various combinations – to push individuals along a developmental pathway to psychosis.

This may also explain the heterogeneity in the manifestations and outcomes of psychotic disorders. Particular clusters of causes may underpin different clusters of symptoms and subsequent trajectories. There is some evidence to support this. For example, childhood adversities are associated with more positive symptoms38, while neurodevelopmental markers are associated with more negative symptoms39. Further, evidence is converging on interrelated psychological and biological mechanisms through which the array of factors increase risk, notably via effects on affective and cognitive processes and on physiological stress response and the dopamine system40, 41, 42, 43.

This evidence was synthesized by Howes and Murray44, who drew from our socio‐developmental model45 to propose an integrated socio‐developmental‐cognitive model that is applicable to all psychoses. At the core of this model is the idea that psychoses emerge and fluctuate over the life course as a consequence of clusters of causal and protective factors operating at multiple levels, from the molecular to the social. It follows that variations across populations will arise where there are differences in the distribution and/or effects of clusters of causal, and protective, factors.

CANDIDATE CAUSES

Several previous reviews, including our own, have considered the evidence for a range of putative causal factors that may contribute to ethnic variations45, 46, 47.

Population differences in genetic risk, obstetric complications or viral infections were proposed in early work on psychoses in minority populations48, fuelling initial concerns about the links being drawn between race, ethnicity, biological deficits, and schizophrenia. However, there seems to be no evidence to suggest that variations in genetic or neurodevelopmental risk markers can explain the high rates of psychotic disorders in some populations. Rather, there is now a broad consensus that variations in incidence by migrant and minority group are most likely a consequence of external, environmental factors, in particular related to social conditions, position, and experience across the life course.

Substance use

There is strong evidence that the use of certain substances, particularly cannabis, is associated with an increased risk of psychosis. This applies especially to high‐potency forms with high concentrations of tetrahydrocannabinol (e.g., skunk)49, 50. This finding is particularly relevant here because cannabis use was one of the earliest and most controversial explanations proposed for the high rates of psychosis observed among the Black Caribbean population in the UK.

Previous work, however, has not provided any strong evidence to support this notion45, 51. More recent work has been slow to emerge. However, it is likely that current interest in cannabis will produce samples that are sufficiently ethnically diverse to allow this question to be considered more fully.

Migration and acculturation

Migration is an unsettling, stressful experience that involves severe disruption to many aspects of individuals’ lives. It is inevitably followed by a prolonged period of adaptation and acculturation, processes that may be more or less difficult depending on the reasons for migration (e.g., economic vs. forced), available economic and social resources, cultural distance from the new society (especially language), and levels of discrimination and hostility faced.

The recent findings that incidence of psychotic disorders is higher among refugees15 hints at the potential importance of these processes and experiences. Asylum seekers and refugees are fleeing natural disaster, war, violence and persecution, and as such often arrive traumatized, with fewer resources and contacts, and face greater challenges in integrating with the host society. There is, however, surprisingly little research on the impact of experiences of migration and settlement on risk of psychosis.

In analyses of data from a small case‐control study conducted in the mid‐1990s, we found some evidence that cultural marginalization (i.e., distance from both culture of heritage and of majority society) distinguished Black Caribbean cases with a first‐episode psychosis from Black Caribbean controls52.

More recently, in analyses of data from the EU‐GEI study, concerning 1,088 cases with a first‐episode psychosis and 1,495 controls from sites in six countries, we found that adjusting for a measure of linguistic distance attenuated the association between migrant or minority group and case status53. Perhaps most intriguingly, the confounding effect of linguistic distance was stronger in first generation migrants, while social disadvantage was more important in second and subsequent generations. This hints at the possibility that the salient socio‐cultural experiences are different for migrants and settled minority populations.

Still, migration and initial experiences of acculturation can at most provide a partial explanation for the high incidence rates of psychotic disorders in some groups. The time between migration and onset of psychosis is typically several years. Hollander et al15, for example, found that the time from migration to first diagnosis was around three years for non‐refugee migrants, which is in line with earlier reports. For refugees, the time was shorter, but still relatively long, at around 2 years14.

Veling et al54, in a study of incident cases in the Netherlands, found that earlier age of migration was associated with a greater risk. It may be that migration at an earlier age has a particularly pernicious impact on risk of psychosis, but earlier migration also means living longer, during formative years of childhood and adolescence, in the host society. Social conditions and experiences in host societies are then likely to be as, if not more, important.

Social contexts and experiences: ecologies of risk

Those from minority ethnic groups are more likely to live in densely populated and relatively disadvantaged and socially fragmented urban areas. Living in these types of places is – in general – associated with an increased risk of psychosis, at least in Northern Europe. However, it seems that these contextual characteristics of areas alone do not account for the variability in rates of psychotic disorders by ethnic group.

In an early study, Harrison et al55 found no evidence that the area of residence could account for observed differences in rates of severe mental disorder between White British and Black Caribbean populations in Nottingham, UK, and our subsequent study in three UK centres (AESOP) found similarly elevated rates by ethnic groups in all centres, despite varying degrees of urbanicity (population density)56. Further, in a more recent study in the UK, Kirkbride et al57 found that similar ethnic disparities were evident in rural and urban areas in a region in the east of England (i.e., 2‐ to 4‐fold increased rates of psychotic disorders among Black Caribbean, Black African, and Pakistani populations, compared with White British).

This is not to say that context does not matter. One of the most striking and consistent findings in the literature is that rates of psychosis are higher among minority ethnic groups where they form a smaller proportion of the local population58, 59, 60. This ethnic density effect was reported as early as the 1930s in the US, in the seminal study by Faris and Dunham61, and has been replicated in many studies since, including several recent reports60, 62, 63, 64, 65.

Interpreting these findings, however, is difficult. When set alongside individual level data suggesting that repeated exposure to discrimination is important (see below), it is possible that living in areas of low ethnic density may increase risk because of exposure to more discrimination and hostility. Conversely, areas of high ethnic density may mitigate risk and promote resilience, possibly via access to more social supports.

Das Munshi et al66, in analyses of data from a UK national survey, did find some evidence to support this: people living in areas of low own ethnic group density did report more experiences of racism and discrimination, and fewer social supports. However, no studies have directly investigated these possibilities in relation to psychotic disorder.

Further, the effects for psychosis may not be uniform. Schofield et al67, in a study using Danish register data, found evidence of ethnic density effects for second, but not first, generation migrants, a finding again suggesting that different clusters of causes may underpin high rates of psychotic disorders in recent migrant vs. settled minority populations.

Social contexts and experiences: disadvantage, discrimination and hostility

At the level of individual experience, there is some evidence that more frequent exposure to social adversities over the life course, particularly discrimination, may be important. For example, family breakdown during childhood (indexed by separation from parents) is both associated with increased odds of psychosis and more common among some minority populations (e.g., Black Caribbean in the UK)68. The adversities indexed by separation (e.g., household discord, housing instability, and financial difficulties) that are experienced more often by those from minority ethnic groups may, then, contribute to pushing more in these populations along pathways to psychoses.

There is similar evidence for markers of social and economic disadvantage and isolation in adulthood69. However, the indicators used in these studies are crude and they do not tell us anything about the more specific exposures they may index or the mechanisms through which they work. Further, in ecological studies, socio‐economic status does not tend to explain much of the variance in rates of psychotic disorder by ethnic group.

Several studies have examined the potential role of discrimination and perceived disadvantage. Of particular note, Karlsen and Nazroo70, in an analysis of data from the Fourth National Survey of Ethnic Minorities in the UK, found an association between the estimated annual prevalence of psychosis and reports of experienced racism (OR=1.6), verbal abuse (OR=2.9), and racial attacks (OR=4.8). It is notable that the strongest effect was for experiences involving physical threat and violence (racial attacks).

The limited relevant data from first‐episode samples broadly supports these findings, albeit the approaches and measures used to capture discrimination and disadvantage vary. In the AESOP, we found that perceptions of disadvantage partly explained the high rates of psychosis in Black Caribbean and Black African groups71. Veling et al72, in the Netherlands, reported that the highest incidence rates were among those populations known to have the highest levels of perceived discrimination (i.e., Moroccan: IRR=4.8).

In other analyses of case‐sibling‐control data on non‐Western migrants, Veling et al73 found that cases were more likely to have a negative ethnic identity, compared with their matched controls. These findings are reinforced by a recent review of 24 studies of perceived discrimination that found overall support for an association with psychoses74. Together, these findings point to discrimination and perceived disadvantage as potentially important factors among minority ethnic groups.

VARIATIONS IN SYMPTOMATIC PRESENTATION

Several studies have examined whether there are ethnic variations in the nature and presentation of psychoses. It is possible that, if a broad set of social factors underpin the higher rates, there would also be more positive and affective symptoms among those from minority ethnic groups. In other words, if there are different aetiological pathways to psychosis, we might expect that to be reflected in differences in the manifestations of symptoms.

A few studies have found some evidence that Black patients present with more positive (e.g., paranoid delusions, hallucinations) and affective symptoms and fewer negative symptoms. For example, in recent analyses of data from our Europe‐wide incidence study, we found that migrant and minority patients presented with more positive (reality distortion) symptoms75. Veling et al76, in a study in the Netherlands, also found evidence of similar ethnic variations, with Moroccan patients reporting more overall symptoms and more persecutory delusions, and both Moroccan and Turkish patients more often meeting criteria for a depressive disorder.

These findings fit with the hypothesis that minority patients will present with more positive and affective symptoms. However, as far as we are aware, no studies have directly linked these variations with hypothesized social factors, so a link remains conjectural. A recent study in Canada did not find any ethnic variations in symptoms at first presentation77. This again points to the possibility that patterns and associations vary across contexts and by ethnic group.

MECHANISMS

There are a small number of studies that have sought to investigate putative mechanisms that may link experience and psychosis in minority groups. For example, Gevonden et al78 examined reactivity to daily stress in a sample of Moroccan and Dutch men, using experience sampling and an experimental exposure to social peer evaluations. They found no evidence that reactivity to stressors was more pronounced among Moroccan men. This is interesting, because it fits the epidemiological evidence. It is not that social factors have stronger effects in minority groups (i.e., that reactivity to stress is more pronounced); it is that they are more common.

Akdeniz et al79, in a sample of 124 young men comprising Germans and second generation migrants, investigated the impact of migrant status on brain structure using structural magnetic resonance imaging. They found that grey matter volume was reduced among migrant men only, a finding they tentatively speculate may indicate effects of environmental stress (associated with migrant status) on brain development, providing a possible mechanistic link between social stressors among those from minority ethnic groups and psychoses.

SOCIAL DEFEAT

Selten et al80, 81, 82 have hypothesized that the experience of social defeat is the common denominator explaining the high rates of psychoses in some minority ethnic groups. This model proposes that the long‐term experience of being excluded from the majority group (i.e., social defeat) increases risk via effects on the mesolimbic dopamine system. This idea arose from analogy with animal studies showing that rodents subject to threatening and intimidating behaviour by other rodents become passive and submissive (i.e., defeated) and that this is associated with sensitization of the mesolimbic dopamine system83, which has been implicated, in humans, in the underlying biology of psychoses43. This hypothesis has gained some traction and so merits careful consideration.

On the face of it, the hypothesis is plausible and has value in providing a catchy and memorable term that serves to highlight the centrality of social factors in generating variations in the incidence of psychoses by migrant and minority ethnic group. However, when probed further, there are some issues that – for the hypothesis to have potential explanatory power – require further clarification and development. First, the hypothesis is largely tautological: it posits minority status (i.e., being excluded from a majority group) to account for high rates of disorder among those who occupy minority statuses. Second, it cannot be minority status alone that explains the observed patterns: psychoses are just too rare for that to be the case. At the very least, other factors must be involved. Third, in the animal studies that were the basis for the original formulation, social defeat is the outcome, not the exposure. It is prolonged intimidating and threatening behaviour (not outsider status) that produces the outcome – passivity and submission (social defeat)82. In other words, it is excessive and repeated threat that is associated with – or leads to – sensitization of the mesolimbic dopamine system in this model. Further, the end state of defeat that characterizes the rodents is reminiscent of a state of helplessness. This is why the social defeat paradigm is usually considered a model of depression83.

In short, the social defeat hypothesis promises a single, elegant explanation – a characteristic that satisfies Ocam's razor. However, the high rates of psychoses in some minority groups are unlikely to be explained so simply81. The range of exposures involved and the mechanisms through which they impact on risk in minority groups are likely to be much more complex – and we need to embrace and seek to understand this complexity. As Einstein commented, “things should be made as simple as possible – but not simpler”85. In relation to mental health problems, Kendler has recently articulated this, noting the need to move beyond monocausal thinking86.

A SOCIO‐DEVELOPMENTAL PATHWAY TO PSYCHOSIS

In synthesizing the evidence around ten years ago, we proposed a socio‐developmental pathway to account for the high rates of psychosis in many migrant and minority populations45. That is, we hypothesized a developmental pathway in which exposure to adversity and trauma (particularly in childhood and/or prior to and during migration) – in the absence of buffers and protective factors – interacts with underlying genetic risk and impacts on neurobiological development (in particular the stress response and dopamine systems) to create an enduring liability to psychosis (reflected in, for example, expression of low‐level psychotic experiences). This liability becomes manifest (primarily as positive and affective symptoms) in the event of further cumulative stressors and/or prolonged substance use, especially high‐potency cannabis. As noted, this proposal has been incorporated into broader models of psychosis.

Our purpose in highlighting a socio‐developmental pathway is to draw attention to the possibility that there are some individuals for whom adverse social conditions and experiences are the primary factors in the development of psychoses: that is, in the absence of these exposures, psychotic disorders would not have developed. It further follows from this that, in populations where adverse social conditions and experiences are more common, rates of psychosis will be higher. Our hypothesis is that this explains the high rates in many migrant and minority populations.

The evidence that has accumulated in recent years – albeit fragmented and sporadic – both fits with this model and suggests refinements. Two lines of research are particularly noteworthy, and are described below.

Psychotic experiences during childhood and adolescence

Several studies have compared the prevalence of low‐level psychotic, or anomalous, experiences in young people from diverse ethnic groups. For example, Laurens et al32, in a study of 595 children aged 9 to 12 years in London, found that Black Caribbean children were around two times more likely to self‐report psychotic experiences compared with White British (OR=1.92). There were, however, no differences between Black African and White British children (OR=0.96).

In a study of 1,545 children with a mean age of 13 years in the Netherlands, Adriaanse et al30 found that Moroccan Dutch (OR=3.0) and Turkish Dutch (OR=2.2) children were more likely to report anomalous experiences with high impact, compared with Dutch children. In another Dutch sample, this time of young adults, Vanheusden et al87 found that self‐reported hallucinations were more common among most minority ethnic groups compared with Dutch participants (ORs ranging from 1.6 to 5.8). They further found interesting patterns by gender: for example, compared with Dutch participants, self‐reported hallucinations were especially common among Turkish women (OR=13.5) and among Moroccan men (OR=8.4).

This raises the question of whether these patterns of low‐level psychotic experiences in childhood, adolescence and young adulthood foreshadow the development of psychotic disorders later in life and, as such, represent opportunities both to understand the developmental origins of these later disparities and, more importantly, to intervene to prevent progression to more serious and intractable mental disorders.

Some studies have sought to investigate putative risk factors for psychotic experiences by ethnic group in young people, but this line of research is very much in its infancy. Adriaanse et al88 extended their work on psychotic experiences in children to consider other problems and risk factors. They found that children from minority ethnic groups reported fewer internalizing but more externalizing problems than Dutch children, which was in part explained by indicators of social disadvantage. In a further set of analyses, they identified several risk and protective factors for mental health problems in general that were evident among minority ethnic children, including trauma, conflicts with parents, and perceived discrimination. These findings resonate with an earlier study of 3,426 children and adolescents in the UK, which found that a migratory history and family dysfunction were associated with around a four‐fold increased risk of psychotic experiences89.

Together, these findings are especially intriguing because they point to variations in mental health during childhood that mirror what is seen in adulthood, i.e. similar or lower levels of internalizing, emotional or common mental disorders and higher levels of psychoses. It is possible, then, that similar experiences of adversity – centred around family conflict and breakdown, and perceived disadvantage and discrimination – are expressed and manifest differently by ethnic group. This would also explain why greater social adversities over the life course are not reflected in higher levels of depression and anxiety in adulthood, that is, in disorders more commonly linked to difficult social conditions and experiences.

We are currently investigating these hypotheses in a newly established cohort study of around 4,000 young people aged 11 to 16 years, sampled from ethnically diverse and economically deprived neighbourhoods in South London (the REACH study, https://www.thereachstudy.com/).

Threat, hostility and violence

There is growing evidence in general that contexts and experiences involving high levels of interpersonal threat, hostility and violence specifically increase risk of psychoses. For example, using data from the E‐Risk study, Arseneault et al90 found that bullying and maltreatment, but not accidents, during childhood were associated with later psychotic (anomalous) experiences at age 12. Further, in analyses of data from our case‐control study of adversity and psychoses, we found that the strongest effects were for childhood exposures and adult life events that involved severe threat, hostility and violence91.

In relation to migrant and minority groups, these findings fit with the evidence of particularly high rates of psychosis among refugees (exposed, by definition, to extreme threat) and of effects for discrimination, especially involving violence, as detailed above. This points to a more specific formulation of the relevant social exposures: i.e., exposure – over the life course – to threat, hostility (including discrimination) and violence, especially in contexts of poverty, disadvantage and isolation (e.g., in areas of low ethnic density).

OUTCOMES

In some early reports, it was hypothesized that the course and outcome of psychosis among migrant and minority ethnic groups would be more benign and better92. There were two possible reasons to expect this. First, if misdiagnosis is a factor, then this should be reflected over time in fewer individuals experiencing the continuously symptomatic course that more often characterizes schizophrenia. Second, if the high rates are a consequence of social conditions, i.e., more reactive and less neurodevelopmental, with more positive and affective symptoms, this should again be reflected over time in fewer people experiencing negative and continuous symptoms.

There have been far fewer studies on course and outcome of psychoses among minority ethnic groups than on incidence, and the findings are mixed, perhaps partly because of methodological differences93. In brief, some suggest better outcomes, some suggest no difference, and some suggest worse outcomes.

This noted, our recent report on long‐term course and outcome by ethnic group in the AESOP sample raises intriguing and troubling possibilities94. In what is the largest long‐term follow‐up of an ethnically diverse cohort of individuals with a first‐episode psychosis, we found strong evidence that outcomes – clinical, social, and service use – were substantially worse for patients of Black Caribbean ethnicity, and worse or no different for patients of Black African ethnicity, compared with White British. Disparities, it seems, extend to outcomes.

Perhaps more intriguing, and relevant to this discussion, is that differences in clinical outcome were, at least in part, accounted for by differences in social disadvantage at baseline. That is, the poor clinical outcomes among Black Caribbean patients – in our data – were in part a function of high levels of social disadvantage at baseline. This raises the possibility that the effects of social disadvantage persist and further impact on course of disorder. It was clear, moreover, that these disadvantages persisted over the follow‐up. For example, of those who were unemployed at baseline, only three (out of 54; 6%) Black Caribbean and one (out of 21; 5%) Black African patients were employed at follow‐up.

STRUCTURAL VIOLENCE

A critical, but rarely made, point is that the social conditions and experiences considered in this paper are not randomly distributed in populations; they are socially structured. Higher levels of poverty, discrimination and threat in minority populations stem from long‐term historical processes that, in predominantly White societies, have systematically marginalized and excluded those from minority groups, creating systematic barriers to education and economic opportunities, to wealth and upward mobility, to living in more prosperous areas, and to positions of power. In other words, entrenched social structures and practices, at root, determine the differential exposure of ethnic groups to the socio‐developmental risks that, we argue, underpin the high rates of psychotic disorders – and subsequent poor outcomes – that have been repeatedly reported across diverse contexts over the past 60 years and more.

This is an example of what Galtung95 termed structural violence – i.e., social structures and institutions harming the health of populations by creating barriers to resources that enable individuals to meet fundamental developmental needs. From this perspective, high rates of psychoses in some minority ethnic groups are, fundamentally, a political issue. As such, our analysis further underscores the importance of policies and community strategies to reduce ethnic inequalities across all domains (e.g., education, employment, income, physical health, mortality) and counter discrimination and racism, in all its forms.

IMPLICATIONS

The necessity of political action and community level initiatives noted, there are also important potential implications for the structure and delivery of mental health services. Alongside studies of rates of psychotic disorders in migrant and minority ethnic populations, there is a substantial body of research showing that individuals with a psychotic disorder from the populations with high rates experience more negative and coercive pathways to and through mental health services96, 97, 98. We found, for example, that higher rates of compulsory admission among Black Caribbean patients in the UK, compared with White British, persisted over a 10‐year period following a first episode94. This is a deeply troubling picture: high rates of psychoses, poorer outcomes, worse and more coercive experiences of services.

There may be many reasons for differences in interactions with services, and they have been intensely and, at times, acrimoniously debated. This is a profoundly important issue.

Understanding is essential to responding effectively to reduce levels of coercion and to improve experiences of care. Our conclusion that more from minority ethnic groups develop psychoses against a background of poverty and disadvantage, with high levels of discrimination, threat and hostility, suggest one possible reason for – and consequent strategy to change – problematic interactions with services. In so far as mental health services are predicated on an illness model, in which responses and treatments are primarily focused on individuals and their symptoms, the social conditions and experiences that, for many, lie at the roots of their distress are marginal to the clinical exchange.

To be sure, many services, particularly early intervention services, do adopt more holistic approaches; but in practice attention to social circumstances and interventions to improve these are ad hoc and inconsistent. If we are right, that there is a predominantly socio‐developmental pathway to psychosis, then by focusing primarily on symptom management, usually with antipsychotic medication, mental health services systematically fail to fully address the underlying problems that, in many, drove onset and that continue to impact on outcomes.

This will disproportionately be the case for minority groups, among which – again, if we are right – more individuals develop psychoses against this developmental background. Indeed, when asked, service users from minority ethnic groups point to multiple social stressors linked to social disadvantage to explain the reported high rates of psychosis99. Perhaps, then, a reorientation of services to ensure systematic attention to the social histories and worlds of those with a psychotic disorder, particularly from minority groups, will address what matters most to many patients, and from that facilitate engagement and reduce the necessity for coercion. This is not to dismiss medication or other interventions. It is, rather, to suggest comprehensive assessment of social needs and perhaps bespoke and enhanced packages of social interventions, where indicated.

CONCLUSIONS

It is now sixty years since the first reports in the UK of high rates of psychoses among migrants from the Caribbean. In the time since, there have been numerous studies that have replicated and extended these initial findings to other populations and other countries. There is considerable diversity in the incidence of psychotic disorders by minority ethnic group. The current evidence, albeit relatively thin, points to adverse social conditions and experiences, possibly particularly those that involve threat, hostility and violence, as the primary determinants of these variations. The specific clusters of social factors that are relevant may vary for first and subsequent generation migrants.

We have proposed a socio‐developmental model, in which greater exposure to social risks across the life course accounts for the high rates of disorder in some groups, making the additional important point that these risks are socially structured. There is much research to be done to test and develop this model. But, most importantly, there is continued urgency to use what evidence we have to develop social policies, public health initiatives, and mental health services to tackle the interlinked problems of high rates of disorder, poor outcomes, and worse experiences of services among some of the most disadvantaged ethnic groups in our societies. Sadly, we are no further forward in meeting this challenge than we were twenty years ago.