Ethnicity and the Mental Health Act 1983

Abstract

Background

Black and minority ethnic (BME) patients are disproportionately detained under the Mental Health Act 1983. There has been no systematic exploration of differences within and between ethnic groups, nor of the explanations put forward for this excess.

Aims

To systematically review detention and ethnicity, with meta-analyses of detention rates for BME groups, and to explore the explanations offered for ethnic differences in detention rates.

Method

Literature search and metaanalysis. Explanations offered were categorised, supporting literature was accessed and the strength of the evidence evaluated.

Results

In all, 49 studies met inclusion criteria; of these, 19 were included in the meta-analyses. Compared with White patients, Black patients were 3.83 times, BME patients 3.35 times and Asian patients 2.06 times more likely to be detained. The most common explanations related to misdiagnosis and discrimination against BME patients, higher incidence of psychosis and differences in illness expression. Many explanations, including that of racism within mental health services, were not supported by clear evidence.

Conclusions

Although BME status predicts psychiatric detention in the UK, most explanations offered for the excess detention of BME patients are largely unsupported.

Over the past 20 years several studies have reported that a disproportionate number of patients from Black and minority ethnic (BME) populations within the UK are compulsorily detained under both civil and forensic sections of the Mental Health Act 1983 (Churchill et al, 1999; Bhui et al, 2003; Morgan et al, 2004). However, some studies have not found this overrepresentation, with some evidence that it may not apply to certain groups, such as people with a first episode of psychosis (Cole et al, 1995; Burnett et al, 1999). There is also evidence that detention rates may not be excessive for all ethnic minority patients. Rates for Asian patients, for example, lie between those for Black (Black Caribbean and Black African) and White patients (Audini & Lelliott, 2002). The presence of such inequalities in service provision is important to service users, service providers and policy makers. For service users and carers, traumatic experiences of detention and coercion can lead to long-term aversion to mental healthcare. From a clinical perspective, such negative experiences cause mistrust and resistance to intervention, with delayed help-seeking and the necessity for further coercion (Singh, 2001; Morgan et al, 2004).

Several hypotheses have been put forward to explain this excess. These can be broadly divided into patient-related and service-related explanations (Littlewood, 1986). Patient-related explanations include higher rates of psychosis (Bebbington et al, 1994), perceptions of Black and minority ethnic patients being at greater risk (Lewis et al, 1990) and poorer insight in this group (van Os et al, 1996). Greater stigma associated with mental illness within minority communities leading to delays in help-seeking and more severe symptoms at presentation have also been offered as explanations (Harrison et al, 1989). Service-related explanations have focused on inherent racism within psychiatry (Littlewood & Lipsedge, 1997) with associated ‘Eurocentric’ misdiagnosis (Fernando, 1988) and perceptions among Black patients of services being inaccessible and inappropriate (Cochrane & Sashidharan, 1996). There are two narrative reviews of such explanations (Littlewood, 1986; Spector, 2001), but a systematic and structured review determining the strength of evidence for the various explanations for this excess is lacking. We conducted a systematic review of all UK literature on ethnicity and detention to:

1. (a) examine the evidence for greater detention of Black and minority ethnic patients within psychiatric services in the UK;

2. (b) explore differences between ethnic minority groups;

3. (c) determine the full range of hypotheses put forward to account for any such excess;

4. (d) examine the evidence for these hypotheses within the literature.

METHOD

A literature search was undertaken of studies relating to the Mental Health Act in the UK, both civil and forensic sections, published between 1984 and April 2005. The following bibliographic databases were searched: Medline, EMBASE, PsycINFO, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), the Applied Social Sciences Index and Abstracts (ASSIA), the Health Management Information Consortium (HMIC), Web of Science, the Cochrane database, the System for Information on Grey Literature (SIGLE) and the National Research Register. The CD-ROM for the British National Bibliography was searched for relevant books. The electronic database search terms were divided into four sets: Mental Health Act terms; mental illness and forensic psychiatry terms; compulsory detention; and ethnic group terms. A combination of search terms from these sets was applied. Where Medical Subject Headings (MeSH) terms were available in the databases, these were exploded and combined. The bibliographies of relevant works were checked for articles missed by the initial search. Key review papers and published bibliographies in the area were also scrutinised for relevant studies.

Inclusion criteria

Studies had to fulfil the following inclusion criteria: publication in English; reference made to the use of compulsion to detain a person under the Mental Health Act 1983 in England and Wales; provision of original data relating to the Mental Health Act; and inclusion of two or more ethnic groups in the study.

The relevance of the literature was initially ascertained from the titles. N.G. and S.S. independently looked at the titles of the first 250 studies in the database searches and agreed on the relevance of all but one article. Discussion of this article led to an improved understanding of the criteria and N.G. then continued with the remaining articles. Where titles appeared relevant, abstracts or equivalent summary information were studied. Just over two hundred (n=210) hard copies of studies appearing pertinent from the abstracts were obtained. Further analysis of the full articles revealed that many of these did not fit the inclusion criteria and they were then excluded. Selected articles were read and the inclusion criteria applied independently by both N.G. and S.S. before the final selection was made.

Personal communication with experts

Once the articles for the review had been selected, 24 experts were sent the list of included studies and asked whether there were any further studies they could suggest. Five experts responded; however, their suggestions for additional studies had already been considered. One expert did not provide any studies, but expressed unhappiness that we had excluded case histories and therefore considered our review to be ‘invalid’. We did explain that this was a meta-analysis of data-based studies and by definition case studies could not be included.

Quality ratings

Literature quality was assessed using an adaptation of a scale (see data supplement 1 to the online version of this article) previously used in a similar review (Bhui et al, 2003). The resulting quality scores ranged from 0 to 14 and were divided into low (0–5), medium (6–10) and high (11–14). N.G. and S.S. rated five articles together to ensure consistent application of the scale and then the rest were rated independently. There was agreement on all but five studies, but discussion revealed that these differences were due to differing interpretations of the scale. Once this was resolved, complete consensus was reached on appropriate ratings for each study.

Data extraction

For meta-analysis, raw data were extracted independently by N.G. and S.S. Categories of explanatory evidence emerged as successive papers were studied; data regarding explanations were extracted independently by N.G. and S.S. and consensus was reached regarding categorisation of explanations. Explanations were recorded as presented in the original paper and no attempt was made to interpret the text to fit any a priori hypothesis. Only explanations relating specifically to ethnic differences in detention rates were included. For instance in papers discussing ethnic differences in admission rates in general rather than Mental Health Act detention rates specifically, explanations were not included in the results. Some explanations were difficult to categorise, such as poor adherence, which could potentially be assigned to more than one category; a judgement was made in these cases as to the most appropriate category. Study authors sometimes offered similar explanations but for different reasons, especially for complex phenomena such as delay in help-seeking among Black patients, which in turn might lead to more disturbed presentation with greater risk of detention. Some authors attributed this delay to lack of social support, whereas others attributed it to denial of illness. Such explanations therefore appear in more than one category. Perception of Black and minority ethnic patients as more violent or at higher risk was categorised separately from studies showing differences in clinical presentation between ethnic groups.

Level of evidence

Each study providing an explanation was scrutinised for the level of evidence for the explanation. Evidence was further categorised as primary evidence, secondary evidence or no evidence. Primary evidence was defined as direct evidence for an explanation provided by a study using its own data. This was further categorised as evidence at the level of an ‘association’ if the data demonstrated correlation between variables where confounders were not controlled and causal interpretations could not be made. An example would be studies where Black and minority ethnic patients were more likely to be detained but also more likely to be diagnosed with psychosis and it was not certain whether ethnicity or psychotic illness was the primary reason for the excess of detentions (especially if tests of association such as chi-squared tests rather than regression had been employed). Secondary evidence was defined as citations of other papers to support an explanation. These secondary citations were perused and key findings summarised, including (where possible) the strength of evidence for relevant conclusions drawn. A few authors discussed explanations for detention rates among Asian patients and these are distinguished from other explanations.

Analyses

Meta-analyses were performed where aggregate data of minority ethnic and White compulsorily admitted patients were provided. Pooled odds ratios were calculated for the overall comparisons using the fixed-effects model. The chi-squared test for heterogeneity was then performed to determine whether there was significant heterogeneity in the odds ratios between studies. For comparisons in which there was significant heterogeneity, four possible source variables for the heterogeneity were investigated. These were patient type (civil, forensic, mixed), episode (first episode, mixed episode), quality rating (high, medium, low) and year of publication. Pooled odds ratios and 95% confidence intervals are presented for studies within each grouping created by the categorical variables. Year of publication was categorised as studies from 1980s, from the period 1990 to 1994, from 1995 to 1999 and from 2000 onwards. Meta-regression was performed, plotting the log odds ratio for each study against year of publication, using appropriate weighting. All meta-analysis was carried out using Comprehensive Meta-Analysis version 2.2 for Windows.

One study (Goater et al, 1999) included three sets of data (at admission, year 1 and year 5), each of which reported different detention rates among Black and minority ethnic patients. Each set was treated as independent and included separately in the meta-analyses.

Terminology

In this paper the term ‘Black and minority ethnic’ is used to refer to participants of any ethnic group other than White. The term ‘Black’ refers to people of Black African, Black Caribbean and ‘Black other’ groups. The term ‘Asian’ is used for people originating from the Indian subcontinent (India, Pakistan, Bangladesh and Sri Lanka). Although all such terms have limitations and obscure important intra-group differences, this review is restricted by these terms as these are the most frequently used categories in such research.

RESULTS

Forty-nine studies met the inclusion criteria and were included in the review but only 19 provided raw data to permit meta-analysis. Table DS2.1 in data supplement 2 to the online version of this paper gives details of the 49 studies listed alphabetically by the first author. Research was mainly concentrated in major cities (71% of studies were from London, with 32% from the Institute of Psychiatry, the Maudsley Hospital or King's College). Most studies were cross-sectional and relied on routinely collected data. Some studies included both retrospective and prospective data; just over half used only retrospective data and a fifth were prospective studies. Sample size varied from 20 patients (Anderson & Parrot, 1995) to 31 702 admissions (Audini & Lelliott, 2002), and just over half (53%) included fewer than 120 patients. Few studies were hypothesis-driven and only 39% stated inclusion and exclusion criteria. No study included power calculations.

Figure 1 is a forest plot of the studies included in the meta-analyses, with odds ratios and 95% confidence intervals for each study on a horizontal plane and the pooled effect displayed with a diamond marker. Table 1 provides a summary of the meta-analyses of four main ethnic group comparisons: Black and minority ethnic (BME) compared with White; Black compared with White; Asian compared with White; and Asian compared with Black. Within these ethnic group comparisons and where there were sufficient data, subgroups such as patient types and illness episodes were also analysed.

Fig. 1 Forest plot of the Black and ethnic minority v. White comparison showing odds ratios and 95% confidence intervals for studies included in the meta-analysis. Goater et al (1999) is included three times in the analysis, hence n=21.

Table 1 Results of the meta-analyses: pooled odds ratios

Comparison	Number of data-sets	Odds ratio (95% CI)	P
BME v.White
Overall	21	3.35 (3.05-3.73)	<0.0001
Patient type	21
Civil	15	4.03 (3.37-4.81)	<0.0001
Forensic	2	2.29 (1.50-3.50)	<0.0001
Mixed	4	3.12 (2.72-3.59)	0.003
Illness episode	21
First episode	3	2.15 (1.55-2.98)	<0.0001
Mixed episode	18	3.53 (3.16-3.95)	<0.0001
Black v. White
Overall	21	3.83 (3.42-4.29)	<0.0001
Patient type	21
Civil	15	4.48 (3.71-5.41)	<0.0001
Forensic	2	2.45 (1.57-3.82)	<0.0001
Mixed	4	3.65 (3.14-4.29)	<0.0001
Illness episode	21
First episode	3	2.42 (1.74-3.38)	<0.0001
Mixed episode	18	4.06 (3.60-4.59)	<0.0001
Asian v. White
Overall	5	2.06 (1.60-2.65)	<0.0001
Patient type	5
Civil	4	3.42 (2.31-5.07)	<0.0001
Mixed	1	1.45 (1.04-2.00)	0.028
Illness episode	5
First episode	1	0.39 (0.113-1.37)	0.142
Mixed episode	4	2.21 (1.71-2.86)	<0.0001
Black v. Asian
Overall	5	2.25 (1.72-2.94)	<0.0001
Patient type	5
Civil	4	1.76 (1.18-2.64)	0.0006
Mixed	1	2.72 (1.90-3.88)	<0.0001
Illness episode	5
First episode	1	3.16 (0.87-1.45)	0.0800
Mixed episode	4	2.21 (1.68-2.91)	<0.0001

Ethnicity

Overall pooled odds ratios for BME compared with White (3.35, 95% CI 3.05–3.73, P < 0.0001) and Black compared with White (3.83, 95% CI 3.42–4.29, P < 0.0001) were similar. The odds for Asian compared with White (2.06, 95% CI 1.60–2.65, P < 0.0001) and Black compared with Asian (2.25, 95% CI 1.72–2.94, P < 0.0001) were both close to 2. Put slightly differently, compared with White patients, Asian patients were approximately twice as likely and Black patients approximately four times as likely to be detained.

Civil and forensic detentions

The pooled odds ratios of detention type showed that the excesses of BME (4.03, 95% CI 3.37–4.81, P < 0.0001) and Black (4.48, 95% CI 3.71–5.41, P < 0.0001) patients compared with White patients for civil detentions are greater than for forensic detentions (BME: 2.29, 95% CI 1.50–3.50, P < 0.0001; Black: 2.45, 95% CI 1.57–3.82, P < 0.001). The odds ratios differ significantly between the patient type groups for the Black v. White (P=0.031) and the BME v. White comparisons (P=0.017). The Black v. Asian comparison was non-significant (P=0.115) and although the Asian v. White comparison was statistically significant, this should be viewed with caution because only one forensic study was included.

Illness episode

There was also an effect for illness episode across different ethnic comparisons, with first-episode BME (2.15, 95% CI 1.55–2.98, P < 0.0001) and Black patients (2.42, 95% CI 1.74–3.38, P < 0.001) less likely to be detained than later mixed-episode BME (3.53, 95% CI 3.16–3.95, P < 0.0001) and Black patients (4.06, 95% CI 3.60–4.59, P < 0.0001).

Quality

Studies rated as high quality in both the BME v. White and Black v. White comparisons showed lower summarised odds than low- and medium-quality studies. This effect was statistically significant in the Black v. White comparison (P=0.03), but not in the BME v. White comparison (P=0.16).

Publication date

Overall the odds ratio decreased significantly with study publication date for both the BME v. White (P=0.001) and Black v. White comparisons (P=0.001). The Asian v. White comparison approached significance (P=0.06) whereas the Black v. Asian comparison was non-significant (P=0.55). There was a statistical correlation between higher quality and recency of publication (P < 0.01).

Explanations for the excess

Five categories of explanations emerged from the 49 studies included in the review. These were categorised as ‘patient-related’, ‘illness-related’, ‘service-related’, ‘culture-related’ and ‘patient–service interaction related’. Each category of explanation and literature offered to support it are presented in separate tables (Tables DS2.2–2.6 in data supplement 2 to the online version of this paper). The right-hand columns in each table describe the level of evidence offered for each explanation. Papers presenting evidence against that particular explanation are grouped at the end of each table.

Patient-related explanations

Patient-related explanations (Table DS2.2) included theories that higher rates of detention occur because Black and minority ethnic patients have higher rates of psychoses, are perceived as being at greater risk of violence and disturbed behaviour, have higher rates of comorbid drug use and have greater delays in help-seeking. Much of the evidence for these explanations came from secondary citations, with little primary evidence, especially for explanations such as comorbid drug use and delayed help-seeking. A few studies reported primary evidence that the effect of ethnicity could be entirely explained by an interaction between diagnosis and challenging behaviour. Some studies found that even when such variables were controlled for, BME status remained a predictor of detention.

Illness-related explanations explanations

Explanations in this category (online Table DS2.3) related to different illness expression in Black and minority ethnic patients, with more challenging behaviour or violence, association with offending behaviour, poorer adherence and greater denial of illness, all of which could account for higher rates of detention. Much of the evidence was of a secondary nature, with one study reporting no ethnic difference in clinical presentation of psychotic disorders.

Service-related explanations

Service-related explanations (online Table DS2.4) included the possibilities that excess detentions could be explained by under-recognition and misdiagnosis of mental illness in Black and minority ethnic patients, lower likelihood of referral to specialist services, greater contact with the police, and racial stereotyping and discrimination within both the mental health and the criminal justice system. There was some secondary evidence of underrecognition of psychiatric problems in such patients and the possible role of racial stereotyping.

Other explanations explanations

The other two sets of explanations, culture-related (online Table DS2.5) and patient–service interaction (online Table DS2.6), included a mixed set of explanations ranging from cultural differences in explanatory models of illness, stigma of mental illness in Black and minority ethnic communities, alienation from and mistrust of services due to negative perceptions and experiences, and unwillingness to seek help. Of all these explanatory categories, culture-related explanations had the fewest supporting citations. Negative perceptions of services, with mistrust and poor engagement, dominated the service–patient interface explanations, but there was lack of supportive primary evidence.

Overall, racial stereotyping, labelling and discrimination against Black and minority ethnic patients was the most often cited explanation and appeared in 15 papers (31%); this was followed by alienation, dissatisfaction, negative perceptions and mistrust of psychiatric services (in 26% papers), greater perception of violence (22%), higher rates of psychosis (22%), delay in help-seeking and poor social support (18%) and misdiagnosis, underrecognition of mental illness with lower referral rates to specialist services (16%). If the perception of Black patients as more violent or at greater risk is considered as part of the ‘racial stereotyping’ category, then this ‘race-based’ explanation was offered in 53% of the studies. There was no primary evidence provided by most studies to confirm any of these explanations, and some papers presented data that contradicted these explanations – for instance, some studies showed that the effect of ethnicity could be accounted for by an interaction between age, gender, diagnosis and challenging behaviour.

DISCUSSION

Excess rates of detention among certain Black and minority ethnic groups have been a major cause of concern for service users, health service providers and policy makers. Reducing ‘disproportionate rates of compulsory detention of BME users’ is a key aim of the government report Delivering Race Equality in Mental Health Care (Department of Health, 2005). Psychiatry and psychiatric services have been accused of being explicitly and implicitly racist both in service provision and diagnosis (Fernando, 1988; Littlewood & Lipsedge, 1997; Sashidharan & Francis, 1999; Sashidharan, 2001; Chakraborty & McKenzie, 2002). Excess detention of Black and minority ethnic patients is not only a clinically important issue, it is also politically charged and ethically contentious, requiring a cautious and balanced approach to research and interpretation of data.

This review confirms earlier findings of an excess of compulsory detentions among Black and minority ethnic patients (Churchill et al, 1999; Bhui et al, 2003; Morgan et al, 2004). However, our findings go further in identifying variations in detention rates between different minority groups, and also reveal differences between first and later illness episodes, and between civil and forensic patients, publication dates and research quality ratings. The finding that studies rated as high quality (a rating that included an assessment of degree of control of possible confounders) tended to report a reduced excess of detentions supports the hypothesis that at least some of the excess is accounted for by confounding variables. The reasons for differences between minority ethnic groups remain unexplored and warrant further scrutiny as to whether these are related to socio-economic, cultural or help-seeking differences between groups, or different experiences and perception of racism. Our finding that forensic detention rates for BME v. White and Black v. White comparisons were lower than the rates for civil detentions was unexpected, given previous results of the overrepresentation of BME patients in secure psychiatric care (Lelliott et al, 2001). However, meta-analysis results should be interpreted with caution as only two datasets were included for the forensic sections.

The increasing detention rate across time, with lower rates for first-episode illness, suggests that the relationship between Black and minority ethnic patients and mental health services deteriorates over time. Parkman et al (1997) found that although Black and minority ethnic patients had decreasing satisfaction with each hospital admission, whether the admissions were compulsory or not did not have an independent effect on patient satisfaction. The relationship between engagement, satisfaction and detention needs to be further explored in order to identify both general concerns and those specific to Black and minority ethnic groups, using longitudinal, mixed-methods studies exploring the process and experience of care and detention over time.

We found that racism and racial stereo-typing of Black and minority ethnic patients were the most common explanations offered for excess detentions, but without primary supportive evidence to justify these assertions. The second most common explanation was that these patients are alienated, mistrust mental health services and are dissatisfied with services. This also had little supporting evidence from the papers itself. Overall, few studies were hypothesis-driven or methodologically based on a testable theoretical or conceptual model. Even where ethnic differences were found, there was a disjunction between reported findings and proposed explanations, with no attempt to link or explore complex multidimensional interactions between variables.

One possible reason why explanations such as racism have become accepted as the ‘cause’ of excess detention is that authors of early papers that reported excess detentions speculated on several possible explanations for this new finding. Instead of robustly testing these hypotheses, subsequent research has presented these speculations as ‘evidence from previous research’. Although this often happens in scientific research, in politically sensitive and emotionally charged areas such as detention and ethnicity it is critical to distinguish fact from opinion and hypothesis from evidence. Racial discrimination undoubtedly occurs in British society and leads to much personal suffering and possibly also to mental illnesses (Bhui, 2002; Karlsen & Nazroo, 2002). Racism may indeed play a part in ethnic inequalities in mental healthcare, but this needs to be scientifically explored rather than accepted as the only cause of such differences (Singh & Burns, 2006).

Inclusion of publication dates in meta-analyses for the BME v. White and Black v. White comparisons shows a reduction in the excess of detention rate with later publication date. This can be interpreted in two ways. Either the excess rates for Black and minority ethnic patients have reduced over time, or with better control of confounders in later studies the effect of ethnicity is partly accounted for by confounding variables.

There is also an important issue of possible publication bias, in which research reporting significant differences between groups is more likely to be published, be cited by other authors and to produce multiple publications than research not finding such differences. The former studies are therefore more likely to be identified in systematic reviews, which potentially leads to bias (Sterne et al, 2001; Dubben & Beck-Bornholdt, 2005). It was noteworthy here that some studies not finding differences in detention rates did not attempt to explain this finding (Holloway et al, 1988; King et al, 1994; Harrison et al, 1999; Riordan et al, 2004), although this was in contradiction to much of the available literature. This suggests that statistically non-significant differences are perceived as less worthy of comment. Presumably, reporting and commenting on an absence of difference in rates was even less likely among authors whose main focus was not ethnicity and the Mental Health Act. This would mean their findings might not have been reported and therefore not included in this review and meta-analyses.

Internationally there is nearly twenty-fold variation in detention rates across Europe, with rates rising in England, Austria and The Netherlands (Zinkler & Priebe, 2002; Salize & Dressing, 2004). In The Netherlands immigrants from Morocco, Surinam and the Dutch Antilles have among the highest rates of psychiatric detention, but this excess is accounted for by the presence of more severe symptoms, risk behaviours, lack of treatment motivation and poor functioning in these groups (Mulder et al, 2006). Although there is no major difference in the attitudes of mental health workers and society with regard to the compulsory detention of people with mental illness across several European countries (Lepping et al, 2004; Steinert et al, 2005), it has been suggested that in England the mass-media-generated public concern about the dangers posed by the mentally ill, along with the high level of personal responsibility that psychiatrists are expected to carry, may influence decision-making and increase the tendency to detain (Turner et al, 1999; Szmukler & Holloway, 2000). A common ethical and legal framework is needed to harmonise these critical decisions and their outcomes across Europe.

Agenda for the future

In order to make studies comparable, there must be consistency in ethnic categories adopted and in their classification. We recommend using a formal standardised approach to classifying that should be adopted in future studies. In-depth, longitudinal, mixed-methods studies using both qualitative and quantitative techniques would improve understanding of patients’ experiences and their journey through the services, pathways to care and why compulsory admission is more frequently required in later admissions among Black and minority ethnic patients. Studies should be hypothesis-driven and also explore the process of application of the Mental Health Act. The true denominator for Mental Health Act studies is the population assessed for detention under the Act, not only the subgroup that is detained. Data relating to both assessment and detention should be routinely and centrally collected. Finally, as we have argued elsewhere (Singh & Burns, 2006), factors that contribute to excess detention even in the first episode of mental illness operate before presentation to mental health services. Hence, any potential solutions must go beyond the health sector and involve statutory as well as voluntary and community agencies.