Income-related inequalities in common mental disorders among ethnic minorities in England
- First Online:
- Cite this article as:
- Mangalore, R. & Knapp, M. Soc Psychiatry Psychiatr Epidemiol (2012) 47: 351. doi:10.1007/s00127-011-0345-0
- 277 Views
The relative prevalence of common mental health problems among different ethnic groups in Britain is one of the least researched topics in health variations research. We calculate and compare income-related inequalities in common mental disorders among ethnic groups in Britain.
Data from a nationally representative survey of ethnic minorities (the EMPIRIC survey) were used to calculate concentration index values to indicate the extent of income-related inequalities within and across ethnic groups.
Looking at income-related inequalities in common mental disorders within each of the ethnic groups, it was found that the burden of these disorders were greater for the lower income groups among the Irish, White and African Caribbean communities. Within-group inequality was less clearly defined for each of the three Asian communities: Indian, Bangladeshi and Pakistani. However, when the data were pooled and individuals were assigned income ranks in the pooled set (not within their own ethnic group), the relative position of those in lower income groups among the different groups was striking. The poor among the Bangladeshi, Pakistani and the African Caribbean groups clearly suffered both from low income and a greater burden of mental health morbidity than the other three groups. The effect of lower income is thus worse for the mental health of populations if they are African Caribbean, Pakistani or Bangladeshi than if they are White, Irish or Indian.
Inequality in mental health morbidity between and within ethnic groups is at least partly linked to income, and thus to employment and education. Tackling disadvantage and discrimination in these areas could help to tackle the challenge of mental ill-health.
KeywordsInequalityEquityEthnicityIncomeNeurosisCommon mental disorders
Concerns have long been expressed that morbidity is associated with low income and other dimensions of social exclusion [1–3]. There is specific evidence relating to mental health, linked to the wider issue and broader policy challenge of social exclusion of people with mental health problems [4–6]. Indeed, recent evidence points to a steeper income-related gradient in psychiatric morbidity compared to general health . Further long-term concerns relate to the complex links between psychiatric morbidity and ethnicity [8–10], and the relative prevalence of mental illness among different ethnic groups in Britain has been argued to be one of the most controversial issues in the health variations field. In fact, such concerns are common in other industrialized countries as well where there are substantial ethnic variations in population. Recent studies from countries such as the USA [11–15], Canada , the Netherlands [17–19] and Norway  bear evidence to this. Not surprisingly perhaps, it has been suggested that mental health has to be understood in its broader socio-political context [21, 22]. Thus, the need for research in this area cannot be emphasised enough. This paper focuses on one element in this spectrum of issues: to examine the links between prevalence, socioeconomic position and ethnicity in Britain. By calculating values of the concentration index, a standardised tool for measuring inequalities recommended by the World Bank and now widely employed in health policy discussions [23, 24], we compare differences in inequality between groups and with previous studies of health inequalities more generally.
Measuring inequality in mental health
The concentration index was developed to measure inequality which is related to socioeconomic position. The index is defined with reference to a concentration curve which measures, in its graphical form, the cumulative percentage of a sample or population ranked by living standards (hereafter shortened to ‘income’) on the horizontal axis, and the cumulative percentage of the health variable on the vertical axis. The curve is usually drawn so that the origin of the graph represents lowest income and worst health, and points on the graph show what cumulative percentage of health corresponds to what cumulative percentage of income. If there is no income-related inequality in health, the concentration curve is a 45° line through the origin, and the concentration index takes the value zero. The index takes a negative value if there is disproportionate concentration of poor health among poorer people, and a positive value if there is disproportionate concentration of poor health among richer people.
If it is suspected that some health problems are correlated with (say) age and gender, and that these factors are unequally distributed across income groups, then it is advisable to examine the standardised distribution of health, adjusting the concentration index for those possible influences. We have previously offered a detailed discussion of the index, standardisation and estimation methods in a mental health context .
Data and variables
Data for the present study were drawn from the Survey of Ethnic Minority Psychiatric Illness Rates in the Community (EMPIRIC) 2000, a cross-sectional survey of adults aged 16–74 years belonging to African Caribbean, Indian, Pakistani, Bangladeshi, Chinese or Irish ethnic groups, living in private households in England . The sample for the EMPIRIC survey was drawn from informants to the Health Survey for England (HSE) 1999, which focused on minority ethnic groups, and also white adults from the Health Survey for England 1998. The sampled adults had agreed, during the HSE interview, to be re-contacted. Ethnic origin in the HSE 1999 was self-defined (using the Census classificatory system) except for the Irish group which was defined as born in Ireland or with a parent born in Ireland. The White group in the HSE 1998 was defined using the same census classificatory system as in 1999.
The study sample of 3,565 was representative for each of the ethnic groups included in the survey. The data had been weighted in the HSE 1999 to remove the imbalances created by the use of different probabilities of selection in postcode areas and to boost samples of the minority ethnic groups to make the data representative of the groups included. In EMPIRIC, additional weights were applied to all cases to adjust for non-response at the follow-up stage (from HSE).
The EMPIRIC survey included assessments of neurotic symptoms and disorders using the Clinical Interview Schedule-Revised (CIS-R)  and of psychotic symptoms using the Psychosis Screening Questionnaire (PSQ) . Other questions covered social functioning, use of services, social networks, carer support, discrimination and harassment, and overall health-related quality of life using the SF12 Physical and Mental Health Summary Scales [28, 29]. After fieldwork was complete a large number of variables and entire modules from HSE 1998 and HSE 1999 were merged with the EMPIRIC data set. The modules included self-reported general health, self-reported long-standing illness, GHQ scores, social support and demographic variables including country of birth, dwelling type, tenure, social class of head of household, income and education.
The analyses in this study focus on common mental disorders only. The measure of morbidity used is a CIS-R score of 12+ (described as ‘cases’ in the EMPIRIC report) .
The equivalised (for household size) household income variable used in our analyses was a readily available derived variable in the EMPIRIC data set.
We conducted two types of analysis. First, we used the concentration index approach to examine income-related inequalities in mental health for each of these ethnic groups separately: Is mental health morbidity more common for poorer people within an ethnic group? Second, we examined the effect of the relative income distribution of these ethnic groups in a pooled sample on the distribution of mental health morbidity: Do some ethnic groups suffer both from low income and a greater burden of mental health morbidity than some others?
The report of the EMPIRIC survey gives a comprehensive description of the people included in the survey  and here we provide just a very brief summary.
Distribution of mental illness by ethnic groups
Indicators of mental illness
Total CIS-R scores (mean)
Cases (CIS-R 12+) (%)
100% (N = 614)
Income-related inequalities in mental health problems by ethnic groups
These differences in the prevalence of mental disorders between ethnic groups do not tell us if there are inequities in the distribution of ill-health within and across groups. That is, they do not indicate whether people in certain ethnic and income groups experience higher levels of illness compared to the rest of the UK population, or whether any of the observed variation in morbidity is due to confounding demographic or other variables. We therefore sought to study the distribution of mental health problems by income quintiles within and between ethnic groups taking into account the demographic composition of income groups as well as any other socioeconomic variables with which they may be correlated.
Income mean and range for each ethnic group
Equivalised household income (£)
Analysing morbidity by income groups using the concentration index approach enables us to understand if the pattern of mental disorders is associated with individuals’ socioeconomic status.
The concentration indices
In presenting within-group inequalities in mental health, sub-samples of each of the five ethnic groups are analysed separately. Individuals belonging to a particular ethnic group are ranked according to their income and the distribution of ill-health by income quintiles within that group is represented by the concentration index for that group. While this kind of analysis is interesting in itself, it is also relevant to examine inequalities between these groups in the wider context. Thus, in presenting inequalities in mental health from the pooled sample, all individuals belonging to all ethnic groups are pooled together while ranking them according to their incomes. The distribution of morbidity is then analysed using concentration indices for the separate ethnic groups with reference to the income ranks from the single (pooled) income distribution. These results represent the positions of the ethnic groups relative to each other both in terms of income and in terms of mental health variables. This second set of analyses is useful because even if the distribution of morbidity by income quintiles within a particular group may appear equal, the result could be rather different when measured against the income distribution of the whole population. The results are likely to be different because the number of persons in each income quintile from each of the ethnic groups represents a different set in the two types of analyses. For example, as can be inferred from the information in Table 2, many of those in the lowest income quintile from the White and Irish groups in the first set of analyses will not be in the lowest quintile in the second set of analyses using the pooled sample. Similarly, there is likely to be no one from the Bangladeshi and Pakistani groups in the highest income quintile in the pooled sample.
Within-group inequalities in mental health indicators
Among the groups that have inequality in mental health unfavourable to the poor, within-group inequality is highest for the Irish (−0.1370), followed by Whites (−0.1268), African Caribbeans (−0.1158) and Pakistanis (−0.0595). These standardised concentration indices (CI) are all statistically significant. The CI for Indians was not statistically significant. There was very little difference between the actual and standardised CI, suggesting that there is no strong age–gender effect on mental health and that almost all of the inequality in morbidity can be linked to socioeconomic status as measured by equivalised household income. Only in the case of the Bangladeshi group was the standardised index lower than the unstandardised index, and then only slightly (0.1712 compared to 0.1878), suggesting that some of the inequality in mental health observed may be due to the demographic composition of this group. Use of other control variables in addition to the demographic variables did not produce any significant changes in the results for standardised indices.
Inequalities in mental health between ethnic groups in the pooled sample
These results are reflected in the concentration indices which are negative for Bangladeshis (−0.4456), followed in decreasing size order by the Pakistani (−0.3576) and African Caribbean (−0.0936) groups. They clearly reveal the huge differences in mental health problems experienced by those in lower income groups in these three ethnic groups when compared to the other three ethnic groups which have positive CIs (White = 0.1912; Irish = 0.1224; Indian = 0.0368).
Concentration indices for ‘CIS-R12+’ for ethnic groups within-groups and in the pooled sample
While the influence of social class and economic differences on the health of individuals still dominates many policy and academic discussions, age, race and gender are increasingly recognised as important. It is clear now that social differentiation is important, rather than just absolute levels of material and social resources, as are the dynamics and character of the social environment .
At the same time, we cannot ignore the evidence that socioeconomic disadvantage is a major contributing factor for ethnic inequalities in health [12, 31–34]. Socioeconomic patterning of health is believed to be present within ethnic groups in most industrialized countries. There is no inherent link between being an ethnic minority individual and a greater risk of morbidity; those in better socioeconomic positions generally have better health. The overall impression, therefore, is that across ethnic groups, across countries, and across outcomes, socioeconomic factors contribute to ethnic inequalities in health in industrialized countries .
Inequalities in general health have been prominently placed on health policy agenda for some decades, have been re-emphasised by the government in England , and have been examined in a number of robust studies. There have also been investigations of the extent of inequalities in rates of mental disorders. However, relatively little attention has focused on inequalities in mental health within and between ethnic groups. Given the complex links between ethnicity and mental health, including concerns about institutionalised racism in service responses, such neglect is unfortunate.
Previous research has established the link between prevalence of mental disorders and socioeconomic position in general [7, 36–39], and also that there is a difference in prevalence rates between ethnic groups [25, 40–43]. Much of the attention focussed on ethnicity and mental health has concentrated on psychosis  and in particular on the higher rate of incidence of psychosis among young men from some Black groups , and on system responses to their symptoms and needs . In particular, there is concern about higher-than-average hospital admission rates, and at high rates of compulsory admission under the Mental Health Act for young African Caribbean, Black African and Other Black males [46, 47]. The few exceptions to this trend in recent years are: a study comparing prevalence and treatment of psychological problems among Black African and Black Caribbeans with White English , a study examining prevalence of common mental disorders among diverse groups of Black African origin  and a study comparing prevalence of depressive disorders between people of Pakistani origin and white Europeans in the UK . The focus in this paper has been different: on common mental disorders and the patterns of association between prevalence and income, and the analysis includes all major ethnic groups in Britain.
The main aim of the work in this paper was to examine the extent to which differences in prevalence within and between ethnic groups could be linked to differences in economic position in society as measured by income.
Limitations and strengths
To our knowledge only one previous study has attempted to address the issue examined in this paper. Brugha et al.  used data from the Psychiatric Morbidity Survey 2000 which did not include many people from ethnic minority groups (only 4% of the total sample) to assess the risk factors associated with low income and poverty and the impact on prevalence of mental illness. The small sample size led the authors to be cautious in drawing firm conclusions from their analyses.
Another notable work in this field was by Nazroo [8, 9]. Using First National Survey of Ethnic Minorities (FNSEM) data, Nazroo  compared the prevalence of depression, suicidal thoughts and non-affective psychosis among ethnic minority groups and found that, compared to the White British group, for each of these outcomes the Caribbean group had higher rates, the Indian/African Asian and Pakistani groups had similar or slightly lower rates, and the Bangladeshi and Chinese groups had considerably lower rates. However, this pattern was not consistent for men and women. For the three Asian groups, rates of neurotic disorders were low for women. For the Indian and Pakistani men, the rates were higher than for the White British. For the Bangladeshi men prevalence was similar to the White British. Even among the African Caribbeans, men had higher rates of prevalence than women for psychosis. Rates of depression were higher for the Caribbeans compared to the White population, contrary to previous findings. There was no suggestion of higher rates of suicidal thoughts among the south Asian women. Among the socio-demographic factors contributing to a higher risk of mental illness, in addition to ethnicity, only social class showed relatively uniform effect across ethnic groups. Social class was inversely related to mental health for all outcomes. Being married or cohabiting appeared to increase the risk of depression for the Caribbean group, while it reduced the risk for the White and south Asian groups.
In a related study Nazroo  looked specifically at the relationship between social class and mental health of ethnic minorities and found that the Indian or African Asian group showed a clear socioeconomic gradient in the risk of both indicators of mental illness and the expected gender difference; women having higher rates than men. However, Pakistani and Bangladeshi groups showed neither the gender difference nor the socioeconomic gradient. For the Caribbean group, although the annual prevalence of non-affective psychosis was higher compared with the White group, the difference was not as great as the three to five times higher rate that treatment statistics had previously suggested. The difference was entirely the result of the higher rate among Caribbean women compared with White women. Rates for Caribbean men were the same as those for White men. Nazroo concluded that there was no evidence to support propositions suggested in some earlier studies that hospital admissions for first-onset schizophrenia are particularly high among Caribbean men born in Britain  and among young Caribbean men . Contrary to evidence from treatment statistics for depression, he found that the Caribbean group had a 60% higher rate of depression than the White group. The difference was greater for men, with Caribbean men having twice the rate of White men. Socio-economic gradient was apparent in the Caribbean group for both of the mental health outcomes.
The analyses offered in the present paper aimed to take these results forward. One advantage available to us was a dataset (EMPIRIC) built on a large national survey that targeted people belonging to African Caribbean, Indian, Pakistani, Bangladeshi and Irish ethnic groups, which could then be supplemented with data on a sample of White adults. Sample size was reasonable overall, and the dataset was representative for each of the ethnic groups included.
Although the EMPIRIC survey contains rich information on a large number of people from ethnic minorities with mental health problems—indeed, it is largest such British survey—it was cross-sectional (as is the Psychiatric Morbidity Survey that it parallels). It is therefore not possible to draw inferences about whether any of the observed associations represent causality. In particular, the association between socioeconomic position and prevalence could run in either direction. At least two causal hypotheses, social causation theory and social selection, have been put forward. The former suggests that higher rates of mental ill-health among people in lower social classes are due to greater exposure to environmental and social stress, such as living in poverty and deprivation in disintegrated and isolated communities characterised by high crime rates. The latter hypothesis suggests that social class is affected by mental ill-health.
Our analyses looked only at people with common mental health problems, and not at people with psychoses or other more severe mental health problems. One reason for this focus was the available sample size, with relatively small numbers of people with psychosis included in EMPIRIC. Another reason was the fact that data came from a household sample and therefore missed those people with mental health problems who are treated long-term in hospital or who live in staffed housing facilities, where the most common psychiatric problem is psychosis . Our analyses also employed a fairly aggregated approach to the classification of ethnicity, again because of sample size considerations, and may therefore have masked some inequalities within an individual ethnic group.
One strength of the approach was to measure the extent of income-related inequalities in mental health among each of the ethnic groups in Britain using the standardised concentration index approach. This not only gives us an understanding of how unequal is the experience of mental health problems for different ethnic groups, but also allows comparisons with studies of other populations and of other conditions.
Summary of findings
The degree of inequality in mental health morbidity that is unfavourable to the lower income groups is different for each of the ethnic groups in Britain. The burden of common mental disorders is greater for the lower income groups among the Irish, White and African Caribbean communities, whereas such within-group inequality is not clearly defined for each of the three Asian communities: Indian, Bangladeshi and Pakistani.
Inequality in income range between ethnic groups results in inequalities in mental health morbidity in society in such way that some groups are more disadvantaged than others: they have both a higher proportion of people in lower income groups and also a larger share of the mental health morbidity among the poor.
The wide-ranging Equalities Review (2007) noted the importance of looking at equality rather than, say, poverty: ‘Government policies that aim to tackle poverty—by ensuring that poor people are lifted above the established threshold—might have no impact on the gap between groups’ [51, p. 133]. The analyses in the present paper have likewise focussed on the inequality ‘distance’ rather than simply on whether people are above or below a particular material or economic threshold.
What steps need to be taken to address these observed inequalities in income-related risk of poor mental health? The Equalities Review  offers some pointers, setting out ten ‘steps’. The first is to agree a valid definition and vision of equality, and, second, to build consensus on the benefits of achieving it. Third, progress should be made towards achievement of equality, in both the long- and short-term. Regular monitoring of progress is needed. The fourth step is transparency about progress. Next, the review recommends targeting persistent inequalities. For example, there may be need for specific measures within a particular service, or re-consideration about how services are designed and accessed, or empowerment of disadvantaged groups to take action, and the implementation of measures to accelerate progress. The sixth and seventh steps are to simplify the legal framework, and to make individual organisations and leaders more accountable for delivering equality outcomes. A related step is to use public procurement and commissioning as positive means by which achievements can be made. Linked to that, step nine is to construct a stronger business case for equality in the independent sectors. Finally, the Equalities Review recommends a more sophisticated enforcement regime which embodies greater transparency, community action, special inspection and public ‘listing’ of recalcitrant bodies.
The 2004 report Focus on Social Inequalities (published by the Office of National Statistics)  showed clearly that even when improvements are made for everyone, inequalities persist. Health inequalities are a reflection of wider inequalities, for example, in income, housing and education that in turn are linked to inequalities in opportunities and aspirations. So, to get to the root of health inequalities, the Government must tackle those wider inequalities. They are rooted in the circumstances where people are born and live [35, p. 29].
In other words, preventive action on its own is unlikely to eradicate inequalities, but would certainly help to address some of the root causes of poor mental health, such as poverty, debt, social exclusion, stigmatisation and discrimination.
A second policy response could be specifically to tackle perceived stigma and discrimination, which remains a huge challenge for all individuals with mental health problems, but which could be especially challenging for people from some ethnic minorities .
Employment is a major source of income and of social status, and employment difficulties have been linked to greater risk of mental health problems. There is a lot of UK evidence that minority ethnic groups experience considerably greater levels of social and material adversity compared to their White counterparts. Madood et al. , for example, showed important differences between the socioeconomic positions of different ethnic minority groups. In another study, which examined the impact of unemployment on British Asians, the unemployed group was found to have lower levels of psychological well-being and self-esteem compared to those in employment [54, 55]. Consequently, in so far as there is still some racial discrimination in employment, tackling this policy area could potentially help to tackle the mental health problem.
The founding principle of the NHS is to provide healthcare to all, irrespective of race, gender, religious beliefs, age, sexual orientation, disability, socioeconomic or other grounds. However, it has not necessarily recognised the greater health need of individuals who suffer disadvantage on these grounds or being proactive in promoting its services to these groups (p. 73).
‘Further enhancing the tools, incentives, accountabilities and leadership required to reduce health inequalities;
building and making the evidence base widely available;
ensuring all communities focus on reducing health inequalities and setting a new objective for post-2010; and
working more closely at Government level to focus on reducing inequalities in health’ (p. 10).
But it is also recognised that the Department of Health and the NHS ‘cannot reduce health inequalities in isolation from other public services’ (p. 10). Many of the causes of inequalities lie outside the health arena, linked for example to employment, education and taxation.
Conflict of interest