Patients from Black and minority ethnic groups in the UK are generally perceived to have a poor experience of mental health services. However, robust comparative information in this area is scarce. Inside Outside highlighted the need for a national strategy to address the mental health needs of Black and minority ethnic groups (Department of Health, 2003). Delivering Race Equality in Mental Health Care outlined an action plan for tackling ethnic inequalities, one of its goals being increased satisfaction with services among patients from Black and minority ethnic groups (Department of Health, 2005). The Department of Health's standards require equity in access to services for minority groups and include a national target for improvements in patient experience as measured by national, validated surveys (Department of Health, 2004).

The Healthcare Commission coordinates a national programme of patient experience surveys on behalf of the Department of Health. This paper analyses ethnic variations in patient experience as reported in the 2004 and 2005 surveys of 26 625 and 25 143 users of community mental health services respectively across all National Health Service (NHS) mental health and primary care trusts providing mental health services in England. Surveys of mental health service users on this scale are unprecedented and offer a unique opportunity for analysing ethnic differences. This paper builds on previous analyses of ethnic variations in patients’ experience of NHS services (Commission for Health Improvement, 2004; Healthcare Commission, 2005a , 2006).

METHOD

The 2004 and 2005 postal questionnaire surveys of users of community mental health services included all 81 NHS mental health trusts and primary care trusts providing mental health services in England.

The surveys were approved by the Multi-Centre Research Ethics Committee for Scotland.

The questionnaire was developed following a review of the published literature on surveys of mental health service users (although there had been very few community-based surveys) and of the survey tools currently in use by NHS mental health trusts. Telephone and face-to-face interviews were carried out with mental health professionals and voluntary sector organisations to identify the issues they thought important to include. The information was used to construct a topic guide for use in focus groups with mental health service users, including those from Black and minority ethnic groups, in different parts of England. The results were used to construct a draft questionnaire, which was tested in cognitive interviews for face validity, comprehensibility and salience with people with mental health problems. Following consultations with an advisory group, including members of the Department of Health's Mental Health Task Force and service user group leads, the questionnaire was piloted before the surveys were launched.

Trusts were given detailed written guidance on sampling methods and advice on sampling was available from the NHS Surveys Advice Centre. To construct the sampling frame, trusts were asked to compile a list of service users aged 16–64 years on care programme approach (CPA; standard or enhanced) who had been seen within the 3 months prior to each of the surveys. They were asked to ensure that all separate CPA lists were combined and that any lists not held electronically were included. Trusts were asked to exclude service users that had been seen only once overall, current in-patients, those who had had no contact with NHS mental health services in the past 3 months and those that did not have a known UK address. Furthermore, prior to the survey a number of reviews of the quality of CPA lists were carried out; one of these is included in the Mental Health Survey Development report (Reference Osborn, Reeves and HowellOsborn et al, 2004). Full details of the sampling instructions can be found at http://www.nhssurveys.org/docs/MH2005_Guidance_v1.pdf

For each of the surveys, trusts were asked to take a simple random sample of 850 service users from their population lists. Detailed instructions on doing this using the Rand function in Excel were provided. The sample was not stratified by CPA level or by any other variable, because it was considered more important to minimise the risk of trusts making sampling errors by keeping the sampling instructions simple. Non-responders were sent up to two reminders.

The analysis is based on the national data-set for each of the two surveys. Although the sample was designed to be restricted to service users aged 16–64 years, some trusts included those over 64 years; these records were excluded from the analysis. The questionnaire used the 16 ethnic categories in the 2001 population census in England and Wales conducted by the Office for National Statistics (ONS). For the analysis, the ONS census ethnic categories were grouped as: White British; White Irish; White Other; Mixed (White–Black Caribbean, White–Black African, White Asian, Mixed Other); Asian or Asian British (Indian, Pakistani, Bangladeshi, Asian Other); Black or Black British (Black Caribbean, Black African, Black Other); Other (Chinese, Other).

For the 2004 survey, we analysed ethnic differences in experience of using services. The questions were grouped into domains of experience, developed jointly by the Department of Health and the Healthcare Commission with advice from the Picker Institute Europe, as follows: access and waiting (access); safe, high-quality, coordinated care (coordination); better information, more choice (information); building relationships (relationships).

Details of which questions were used to construct each of the patient experience domain scores are given in the Appendix. Domain scores from these surveys are used in the Healthcare Commission's performance assessments of trusts providing mental health services. They are also used at national level by the Department of Health to monitor progress against its Public Services Agreement target with the Treasury regarding annual improvements in patient experience.

Patients’ responses to questions were scored between 0 and 100, reflecting the gradient of negative–positive experience of services. Higher scores reflect a more positive experience than lower scores. Mean scores for each domain and an overall mean were derived for each ethnic group by aggregating scores for users’ responses to individual questions.

Multiple regression analysis was used to examine ethnic differences in responses after controlling for age, gender, employment status, living alone, self-reported mental health status, admission to hospital in the previous year for mental health reasons, detention under the Mental Health Act 1983 in the previous year and trust of treatment. Mean domain scores by ethnic group were estimated for a baseline group of respondents using the model coefficients, to illustrate the differences between different ethnic groups. Care programme approach status (i.e. standard or enhanced) was not known for 20% of respondents, hence it was not included in the model.

For the 2005 survey, we analysed ethnic differences in patients’ access to services and treatments, based on binary responses to individual questions. The White British group was used for comparison. Logistic regression was used to examine ethnic differences after controlling for selected independent variables. Adjustment was made for the same variables as in the 2004 survey, with two exceptions: the question on living alone was not included in the 2005 questionnaire and information on CPA status was more complete in the 2005 survey and hence was included in the analysis.

Results by ethnicity are presented after forcing all covariates into the model. Conclusions from models reduced using a stepwise approach to contain only significant covariates were highly similar. Statistical analysis was conducted using STATA version 8.0 for Windows. Details of the surveys, questionnaire, domain score methodology and trust-level results are available (Healthcare Commission, 2004, 2005b ).

RESULTS

A total of 27 398 and 26 555 service users responded to the 2004 and 2005 surveys respectively, with overall response rates of 41 and 40% respectively. It was not possible to calculate response rates for ethnic groups with any degree of accuracy because, although self-reported ethnicity was available for 97% of respondents to both surveys, ethnicity was grossly under-coded in the sample records. In the 2004 survey, 50% of the 67 179 sample records did not have an ethnic code and in a further 2% self-reported ethnicity of survey respondents did not match ethnicity in the trust record; in the 2005 survey, 37% of the 66 948 sample records did not have an ethnic code and in a further 1% self-reported ethnicity in survey respondents did not match ethnicity in the trust record. Moreover, it is not known whether there were any ethnic-specific biases between where ethnicity was or was not recorded in trust records. Finally, in trust records ethnicity was often coded with a general term such as White/Black/Asian rather than one of the 16 ONS categories used in the survey questionnaires. For these reasons, any comment on response rates by ethnicity has to be very tentative. Based on the partial information available, the indications were that response rates were lower among minority ethnic groups: in the 2004 survey they were 33, 32 and 45% in the Asian, Black and White groups respectively, and in the 2005 survey the corresponding figures were 30, 30 and 41%.

In both the 2004 and 2005 surveys, 10% of respondents were of Black and minority ethnic origin (i.e. excluding White British; 2745 and 2559 respectively; Table 1). The ethnic composition of respondents was similar across the two surveys.

The characteristics of respondents were also similar across the two surveys (Table 2). In both surveys, respondents from minority ethnic groups, other than the White Irish group, were younger than the White British group. Female respondents outnumbered male respondents except in the Asian group in the 2004 survey and in the Black group in the 2005 survey. The proportion living alone (only available for the 2004 survey) was highest in the Black group, being almost three times greater than in the Asian group. In both surveys, the proportions in paid work were lowest among the Asian, Mixed, Black and White Irish groups. Self-reported mental health status was poor/very poor in about one-quarter of respondents in most ethnic groups in 2004 and in about one-fifth in 2005; in both surveys, the proportion was significantly lower in respondents from the Black group (15.2 and 12.1% respectively). The proportion of respondents with at least one hospital admission for mental health reasons, or detention under the Mental Health Act 1983, in the past year was higher among minority ethnic groups than in the White British group, the proportions detained in the Black group being more than double those in the White British group in both years. Minority ethnic groups, the Black group in particular, were also more likely to be on enhanced CPA.

For the 2004 survey, we analysed ethnic differences in patients’ overall experience of using services (classified by domains of patient experience) after controlling for the independent variables (Table 3). Compared with the White British group, patient experience scores were lower for the White Other group for the access and information domains, and overall. For the Asian group, scores for the information domain were lower than for the White British group, and scores for the coordination domain were higher. Scores for respondents from the Black group were higher than for the White British group for the coordination and relationships domains.

The regression analysis (Table 4) showed the extent to which the independent variables predicted patient experience scores. Although ethnicity was a significant predictor of patient experience for some ethnic groups for some domains, overall some other independent variables had stronger effects. Self-reported mental health status had the strongest explanatory effects across all domains and overall, with respondents in poor health responding more negatively. Increasing age was a significant, positive predictor of domain scores. Living alone, detention and hospital admissions in the past year were negatively associated with patient experience.

For the 2005 survey, we analysed ethnic differences in patients’ access to services and treatments, based on responses to individual questions and with the White British group as the baseline for comparison. Odds ratios from the regression analysis are given in Table 5. Overall, 84% of respondents had been in contact with mental health services for over a year and about half for over 5 years. After adjusting for the independent variables, the White Other, Asian and Other groups were more likely to have had a shorter duration of contact with mental health services (i.e. under 1 year) than the White British. Overall, 83.6% of respondents said they had seen a psychiatrist in the past 12 months. Except for the Other group, there were no ethnic differences from the White British group for being seen by a psychiatrist in the past 12 months or for being seen by the same psychiatrist at the last two appointments. Overall 57.9% of respondents said they had seen a community psychiatric nurse (CPN) in the previous 12 months. Compared with the White British group, a greater proportion of respondents from the Black group said they had seen a CPN. Just over half (55.3%) of respondents said they had also seen a health professional other than a psychiatrist or CPN in the past 12 months; the proportion was significantly lower in the Asian group than the White British group.

Overall, 92.6% of respondents said they had taken a medication for mental health problems in the past 12 months; no ethnic differences were observed after controlling for the independent variables. Overall, 40.6% of respondents said they had had some form of talking therapy (e.g. counselling or psychotherapy) in the past 12 months. Compared with the White British group, significantly lower proportions of respondents from the Asian and Black groups said they had received any form of talking therapy in the past 12 months. However, overall, 42.6% of respondents said they did not want a talking therapy. Among those who said they did want a talking therapy, overall 65.7% said they had received it. The odds ratios for those receiving talking therapy among those who wanted it were below 1 for all minority ethnic groups except for the Other group. The results did not reach statistical significance but this could reflect sample sizes as most of the confidence intervals for the individual ethnic groups only just straddled 1.

Almost two-thirds (62.5%) of respondents said they had been told their care coordinator; significant proportions said they had not (27.6%) or they did not know (9.9%). Compared with White British respondents, the proportions that said they had been told their care coordinator were significantly lower among White Other and Asian respondents, and significantly higher among Black respondents. Over two-thirds (69.6%) of respondents said that it was less than 1 month since they had last seen their care coordinator, 19.0% said it was 1–3 months, with the remainder stating it was over 3 months. A higher proportion of the Asian and Other groups compared with the White British said they had last seen their care coordinator over 1 month ago.

When asked whether they had been given or offered a written or printed copy of their care plan, overall under half (44.9%) answered affirmatively, a similar proportion answered negatively (44.4%) and 10.7% said they did not know. Compared with the White British group, a significantly lower proportion of Asian respondents said they had a copy of their care plan. Overall, 47.1% of respondents said they had had a care review in the past 12 months; compared with the White British group, this proportion was lower among Asian respondents and higher among Black respondents.

When asked whether they had an out-of-hours number for the local NHS mental health service, 44.4% of respondents said yes, 46.5% said no and 9.1% said they were not sure or did not know. The proportions responding affirmatively were lower among Mixed and Asian respondents compared with White British respondents.

In summary, relative to the White British group, the main ethnic differences were for the Asian group, who responded negatively to several questions about access to community mental health services. No negative differences were apparent for the White Irish and Black groups. Respondents from the Black group were more likely than White British to say they had seen a CPN, had a care review in the preceding year and had been told their care coordinator. Overall, minority ethnic groups were less likely to have said they had received talking therapies in the past year.

Age, employment status, hospital admission, detention under the Mental Health Act 1983 and CPA status were stronger and more consistent predictors of responses than ethnicity. The strongest predictor was self-reported health status.

DISCUSSION

Ethnic differences in rates of mental illness, and access to and experience of mental health services, have been a focus of widespread and long-standing debate and concern in the UK. It is widely reported that Black and minority ethnic groups in England, especially African–Caribbeans, have adverse experiences of mental health services (Sainsbury Centre for Mental Health, 2002; Department of Health, 2003, 2005). Patient experience is increasingly recognised as being critical to service development and the provision of patient-centred care. However, there is a paucity of robust comparative research on the experiences of mental health service users, including those from Black and minority ethnic groups, as there have been few systematic, comparative studies with robust sample sizes.

APPENDIX A