Increasing equity of access to high-quality mental health services in primary care: a mixed-methods study

Article history

The research reported in this issue of the journal was funded by PGfAR as project number RP-PG-0606-1071. The contractual start date was in August 2007. The final report began editorial review in August 2012 and was accepted for publication in February 2013. As the funder, the PGfAR programme agreed the research questions and study designs in advance with the investigators. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PGfAR editors and production house have tried to ensure the accuracy of the authors' report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Christopher Dowrick is a non-executive director of Mersey Care NHS Trust. Pete Bower has received funding from the British Association for Counselling and Psychotherapy for consultancy. Carolyn Chew-Graham was an advisory board member for Self Help Services, Manchester. Karina Lovell is a non-executive director of Manchester Mental Health and Social Care Trust.

Copyright statement

© Queen's Printer and Controller of HMSO 2013. This work was produced by Dowrick et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Background

According to the World Health Organization, in Western Europe over one-third of the population experience mental ill health at some time in their lives, with the majority coming into the diagnostic categories of anxiety and depression. 1 Mental health problems impose substantial emotional, social and economic burdens on those who experience them, their families and carers and society as a whole. 2 People suffering from depression and anxiety find it much more difficult to hold down a job, and those in work are likely to have high rates of sickness absence: > 900,000 adults in England claimed sickness and disability benefits for mental health conditions in 2003, with particularly high claimant rates in the north of England. 3,4 Layard5 has calculated the resulting loss of output within the UK as £17B or 1.5% of gross domestic product. Much of this cost falls on the Exchequer, which loses in consequence roughly £9B in benefit payments to people with mental health problems and reduced tax receipts. If problems related to impaired work efficiency (‘presenteeism’) as a result of mental ill health are included, the calculated loss of output is considerably higher. 6

A wide range of clinical interventions,7,8 collaborative care,9 self-management10 and social and community initiatives11 have been shown to be effective in trials in improving outcomes for people experiencing common but disabling mental health problems such as depression and anxiety.

Underserved groups

Many people with high levels of mental distress are disadvantaged, either because care is not available to them at the right place or at the right time or because when they do access care their interaction with caregivers deters help-seeking or diverts it into forms that do not address their needs. 12

Groups with inadequate access to primary care include people from black and minority ethnic (BME) communities, asylum seekers, homeless people and adolescents with eating disorders. For example, women of South Asian origin in the UK have a high prevalence of depression and self-harm, often in the context of severe and persistent social difficulties, which only become apparent when they are in a crisis. 13 Asylum seekers are at greatly increased risk of mental health problems:14 they are 10 times more likely to have post-traumatic stress disorder (PTSD) than age-matched general populations15 and face substantial political, institutional, cultural and language obstacles to accessing primary care. 16 Mental health problems are up to eight times more prevalent among homeless people than in the general population. 17 There is a ‘cycle of reluctance’ of homeless people to use primary care services and of general practitioners (GPs) to accept them, and homeless-focused primary care services may still not access rough sleepers, who are at greatest risk. 18 Although anorexia nervosa has the highest mortality rate of any mental disorder,19 young people with eating disorders rarely present on their own account to clinical services. 20

Groups who receive inadequate help when they do access primary care include elders, people with advanced cancer, those at risk of long-term sickness absence and people with medically unexplained symptoms. Depression is common in older people, particularly those with chronic physical illness, but tends to be underdiagnosed and inadequately managed. 21,22 Depression appears to be a significant and persistent symptom for about one-quarter of patients with advanced cancer23 and may be an independent predictor of mortality. 24 People with mental health problems are at risk of long-term sickness absence25 but primary care currently has few strategies to identify or help them. People considered to have unexplained physical symptoms often present with mental health needs,26 but clinicians tend to disregard these and assert somatic agendas; many of these patients are sceptical about doctors' ability to manage their health problems. 27

We focused on these eight groups as exemplars of those who are underserved by primary care because we were able to deploy the existing scientific and clinical expertise of our research group with regard to these groups. We considered that lessons learnt here may have wider implications for other groups of people whose mental health problems are managed within primary care.

The extent of commonality of issues across these groups meant that a combined approach was likely to be the most effective. The Social Exclusion Unit's report on mental health confirms that people from these groups face overlapping barriers to getting their mental health needs addressed. 3 Women from BME communities, homeless people, asylum seekers and elderly people living alone, for example, often experience severe and persistent social difficulties. Engagement and communication are inherently problematic in the cases of adolescents with eating disorders28 and women from ethnic minorities. 29

These groups also share a crucial set of problems with regard to health literacy, that is, the capacity to obtain, interpret and understand health information and services in ways that are health enhancing. 30 Health literacy is in part related to lack of educational attainment, but importantly it is also about the stigma associated with seeking help for mental health problems. Health literacy is critical for patients' confidence in themselves, and in the services they seek (or often do not seek) to access.

Health literacy needs to work both ways. Health services have an obligation to be literate about the needs of the patients they seek to serve. Empirical work on primary care provision for hard-to-reach groups has focused on how to physically structure services, that is, whether primary care trusts (PCTs) [or the forthcoming clinical commissioning groups or (CCGs)] should commission services that focus solely on provision for specific groups (e.g. elders or people from BME communities), or whether mainstream general practices should be encouraged to provide more flexible care through the enhanced services route. However, we considered that the key issue here is not how services are physically structured or funded, but how people from these underserved groups can be enabled to access high-quality mental health services in primary care. 31

Access to health care

In the UK, policy interest in access to mental health care has largely been restricted to two major issues. Understanding the delivery of mental health care has often been predicated on the influential ‘pathways to care’ model. 32 This model identifies several ‘filters’ that exist between patients with mental health needs in the community and different ‘levels’ of care, and has highlighted the importance of the recognition of mental health problems by primary care ‘gatekeepers’33 and the difficulties of training professionals to improve recognition and referral. 34,35

The second major policy concern is the provision of non-pharmacological treatments such as psychological therapies. Although such treatments are generally more acceptable to patients than medication,36 access is far more restricted because of limitations in the number of adequately trained therapists who can deliver these treatments, and because of problems with cultural congruity between therapists and many prospective clients. 37

Developing interventions to improve access to mental health care is a policy priority and has led to two major policy innovations in the UK. The first involves changes to the contract under which GPs are paid: one target in the Quality and Outcomes Framework (QOF)38 involves payments to GPs to use standardised screening instruments to detect cases of depression in populations such as patients with long-term conditions. 39 The second is the Improving Access to Psychological Therapies (IAPT) programme, which has involved the recruitment of psychological therapists,40 the introduction of ‘minimal’ psychological interventions such as guided self-help and computerised treatments41 and the adoption of the ‘stepped-care’ model to maximise patient access to care. 42

These access innovations are based on assumptions that are not always made explicit. Importantly, there is a focus on supply-side factors, such as the availability of treatments, and on structural and organisational changes required to reduce or remove instances in which pathway ‘filters’ become barriers to care. There is generally less explicit consideration of demand issues and the factors governing the journey of the patient in need. Within health sciences, there is increasing recognition of the need to base quality improvement activities on rigorous theories and models that describe core concepts of relevance, and outline the mechanisms and relationships that underlie service delivery problems. 43,44

Proxy measures of access, such as service use or supply of health professionals, have been used, but a full consideration must include multiple dimensions such as adequacy of supply, barriers to access, effectiveness and equity for different population groups. 45

If services are available and there is an adequate supply, then the opportunity to obtain health care exists. The extent to which a population gains access also depends on financial, organisational and social or cultural barriers that limit the utilisation of services. Available services must be relevant and effective if the population is to gain access to satisfactory health outcomes. Availability and barriers have to be considered in the context of the perspectives, health needs and settings of diverse groups in society. Equity of access may be measured in terms of the availability, utilisation or outcomes of services. Both horizontal (equal treatment for equal groups) and vertical (unequal treatment of unequal groups) equity require consideration.

Aim and objectives

We decided to undertake a research and development programme to determine how people from underserved groups can make better contact with primary care and to find out how to ensure that they receive the help that they really want and need for their mental health problems.

The aim of the AMP programme was to increase equity of access to high-quality primary care mental health services for underserved groups.

We had six principal objectives:

1. to clarify the mental health needs of people from underserved groups, who have been shown to receive inadequate primary care mental health services

2. to identify evidence-based primary care services that such people consider relevant to their needs

3. to identify barriers to, and facilitators of, access by such people to high-quality primary care mental health services

4. to develop and test a portfolio of interventions that are credible and acceptable for particular underserved groups

5. to establish effective strategies for the dissemination of evidence about interventions for members of underserved groups with mental health problems that have demonstrable efficacy and acceptability

6. to begin to integrate into primary care effective and acceptable interventions for people with mental health problems from underserved groups.

We considered it important to address the following key issues:

• Complex problem presentation in primary care. The common mixture of physical and psychological symptoms makes assessment challenging, for example identifying depressive symptoms among older people or those with advanced cancers, or being aware of the frequency of primary presentation of physical symptoms among women from ethnic minorities.

• Current diagnostic categories may be irrelevant, unhelpful or disabling,46 for example the concept of medically unexplained symptoms, the difference between depression and demoralisation for patients with advanced cancers, and depression in older people with chronic physical disease.

• User-defined perspectives are crucial regarding definitions of unmet needs and how these needs are best understood. For example, qualitative work with patients with unexplained symptoms indicates that patients are unconvinced of the ability of doctors to treat psychosocial problems;47 and older people often view depression differently from health professionals. 48

• Engagement and communication problems. With adolescents with eating disorders, lack of engagement is an inherent part of the disorder, including difficulty describing thoughts and emotions, embarrassment and negative attitudes to adults in authority. 28 Women from ethnic minorities tend to seek professional care only when in crisis. 29 Homeless people perceive reluctance to treat on the part of service providers.

• Professional contexts. As currently configured, primary care services often do not meet the mental health needs of these patients. This may be because of the perceived stigma associated with attending mental health services, poor agreement between service and patient or client regarding treatment aims,49 poor outcomes of existing treatments or inaccessible or unclear pathways to care. In palliative care of advanced cancers, for example, professional training concentrates on physical care at the expense of psychological or spiritual dimensions. 50

We designed our programme to add value by combining clinical and methodological expertise. We built on Watters' three-dimensional model51 for exploring inter-relationships between macro-level institutional factors in the creation of mental health problems among these underserved groups and the interventions offered to them within or through primary care. We focused our efforts on targeting key clinical outcomes, exploring how to adapt service configurations to meet patients' needs, rather than changing patient presentation to fit in with existing services. We put patients' experiences and expressed needs at the centre of care, and addressed the practice and policy contexts in which these are located.

Programme plan

We designed the AMP programme to link knowledge about barriers to, and facilitators of, access for our exemplar groups with the perspectives of users, developing creative ways to meet the particular needs of these groups. We divided the programme into three overlapping phases (Figure 1).

FIGURE 1.

The AMP programme plan.

• Phase 1 (months 1–18) – understanding the problems and generating potential solutions. We undertook a general review of barriers to, and facilitators of, access to care, and interventions to overcome barriers, using a combination of literature reviews and qualitative research methods. We focused on user perspectives of mental health issues in relation to each of the underserved groups to understand whether or not there were any access issues specific to each group and how different issues affect them all. We then considered the results of our reviews and used a consensus process to develop exemplar interventions that took account of known barriers and faciliators while remaining sensitive to the needs, preferences and priorities of our exemplar underserved groups and other stakeholders. We describe this process in Chapter 2.

• Phase 2 (months 15–48) – testing potential solutions. We developed interventions designed to improve access, understanding and care by setting up a model of care, which we called the AMP Development Partnership. This model of care has three linked domains: increasing community stakeholder engagement, enhancing the quality of primary care and providing culturally sensitised well-being interventions. We developed interventions at the community, primary care and individual patient level. We used a quasi-experimental design to test the components of the AMP model, individually and jointly, in terms of their acceptability and credibility to service users and other stakeholders, their impact on the mental health of service users and their service implications. In Chapters 3–7 we describe how we implemented and evaluated the three components of the AMP model of care. We consider the implications of the combined AMP model of care for increasing access to appropriate primary care and well-being interventions and for reducing mental health stigma in the community.

• Phase 3 (months 43–60) – putting what works into practice. We developed and evaluated methods to disseminate our findings across the broad economy of primary health care, including focused implementation experiments at both policy and practice levels in different PCTs, stakeholder meetings and the creation of a dedicated AMP website. We describe this phase in Chapter 8.

Finally, we draw the elements of the programme together. We make recommendations for future research and consider the implications of our evidence for clinical practice.

Underlying perspectives

We began with the assumption that members of underserved groups are not passive sufferers of mental health problems but are people who interpret and respond to experiences and who are capable of mounting challenges to external forces bearing on them. 51 We considered the inter-relationships between macro-level societal and institutional factors in the creation of mental health problems among hard-to-reach groups and the ways of coping with and overcoming these problems, including interventions offered to them within or through primary care. We focused on key clinical outcomes, exploring how to adapt service configurations to meet patients' needs rather than changing patient presentation to fit in with existing services. We put patients' experiences and expressed needs at the centre of care and addressed the broader practice and policy contexts in which these are located.

We adopted a whole-system approach to identify the need for a range of credible and acceptable interventions, and the barriers to and facilitators of their development. This meant that we should examine diverse sources of evidence. 52 Quantitative evidence from randomised trials provides information about what works and for whom, but is of little help in explaining how or why. Grey literature, produced by service providers or advocacy groups, is a useful source of evidence for the current priorities of policy-makers and service providers, and for service innovations, although is sometimes lacking in rigorous evaluation. Qualitative sources of evidence are needed to find out why current practices may not work, and what might help to improve them. Published qualitative sources may not directly address key questions that we are concerned with here, namely problems with access; hence, there may be a need to turn to detailed interview transcripts, or to ask actual and potential service users and providers directly about specific issues. Stakeholders must be involved not only in answering research questions, but also in conceptualising the questions to be asked. It is then essential to find effective and valid ways of synthesising evidence to generate and test potential solutions.

The approach we have taken is presented schematically in Figure 2, to which the rest of this chapter refers. We conducted seven linked studies. We began with a conceptual scoping review. We undertook a structured scoping review of published quantitative literature and a meta-synthesis of published qualitative literature. We reviewed relevant grey literature and undertook a secondary analysis of qualitative data sets generated with actual and potential users of primary mental health-care services. We also conducted two stakeholder consultation exercises, generating new qualitative data. We carried out the literature reviews and stakeholder exercises in parallel, rather than in series, to give stakeholders the opportunity to inform the literature reviews and analysis of available qualitative data.

FIGURE 2.

Methods of evidence generation and synthesis. 52 Reproduced with permission from Dowrick C, Gask L, Edwards S, Aseem S, Bower P, Burroughs H, et al. Researching the mental health needs of hard-to-reach groups: managing multiple sources of evidence. BMC Health Serv Res 2009;9:226.

Identifying key concepts

We began with a conceptual scoping review to develop a map of key concepts concerning access to primary care, and to identify the range of current interventions that have been used to improve access to care (see Structured scoping review of published quantitative literature). This review was generic and not restricted to mental health; the intention was to capture the full range of relevant concepts and definitions. Initially, members of the research team were asked to identify key papers and books relating to access. This database was augmented by a search of electronic databases using a range of terms relating to access, combined with a search filter developed at the National Primary Care Research and Development Centre in Manchester, to identify conceptual and theoretical literature. Candidate interventions around access were developed, starting with the list of interventions developed by the Effective Practice and Organisation of Care (EPOC) group of the Cochrane Collaboration,53 and refined through focused literature searches and reflection on the developing conceptual map.

The major components of our conceptual scoping review have been described in the paper by Dowrick et al. 52 and are presented in Figure 3.

FIGURE 3.

Key concepts in access to health care.

We draw attention to six key overlapping concepts emerging from this review: recursivity, candidacy, identity and morality, graduated access, adjudication and cultural competence.

Recursivity refers to the ways in which illness behaviour is both enabled and constrained by the interactions that take place between individuals and health professionals in health service settings. Illness behaviour and personal illness management strategies are modified by service contact, as people learn to ‘fit’ into what is required of them by health professionals and their beliefs about what is legitimate illness and service use. 54 The ability of health professionals to communicate effectively with patients may reinforce or discourage health action in the future. 55

Candidacy describes how people's eligibility for health care is determined between themselves and health services. 56 The reaction to each illness episode draws on past experience, cultural expectations, norms and self-definitions. This reservoir of candidacy, an understanding of a right to care, is then shaped towards resolving the individual illness episode. Candidacy in relation to an individual illness episode should be seen as a special case of the individual's ongoing perceptions of their right to care, balanced against a fear of socially inappropriate behaviour. Candidacy arises from ongoing negotiation, influenced by a wide number of factors. Health services are constantly defining and redefining the legitimate objects of health services. In response, people in the community are also trying to make sense of this process.

Following determination of candidacy, individuals undertake navigation to gain a point of entry to health services – an ‘appearance’. This involves a series of psychological and cognitive competencies and resources, which include self-efficacy57 and health literacy. 58 These capacities are partly patterned by individual characteristics, but also through the interaction of those individual capacities and the health-care system.

Appearances can involve a number of different approaches, including appearing before health services through patient-initiated actions, or through ‘invitations’ (where people respond to health services) or ‘grabs’ (where candidacy is not under patient control, such as compulsory hospital admission). Screening for depression during routine medical consultations reflects aspects of both of these types. 39 Other work has highlighted the role of more diffuse approaches (described as ‘muddling through’) in the presence of ambiguous symptoms, in which the effects of close social networks are emphasised. 59

With respect to identity and morality, an orientation to candidacy brings several aspects of the sociological discourse on identity into play. Identity refers to an individual's sense of self, and is maintained in interaction with others. People are motivated to seek confirmation of their identity in interactions with others, and there is some evidence that negative experiences with regard to health services reflect in part threats to identity. 60 This may involve perceived stereotyping by health professionals, feelings of disempowerment and feelings that their subjective experience is ignored.

Another strand of the identity discourse deals with roles. If illness threatens competence to perform social roles, it may involve a fundamental challenge to self. For example, in women presenting with depression in primary care, perceived failures of competence in primary social roles were interpreted as a sense of duty to seek care,61 but framing help-seeking as a moral action coloured women's experiences of care, which in turn came to be seen in terms of ‘self-sacrifice’ and the ‘moral dilemma’ of accepting medication.

In terms of graduated access, the concepts of recursivity and candidacy suggest that access to care is best seen as an interaction between organisational factors operating in health services combined with people's perceptions of those services and their individual social circumstances in seeking care. 62 One way of changing access is to modify service provision options at the point at which people want to access services. This may involve ‘graduated access’ to services, including services that people can access without requiring face-to-face contact with professionals and which maximise the use of new technologies such as the internet.

Adjudication refers to professional judgements about the presentation of an individual for an intervention or a service, influenced by categorisations made by professionals with reference to current services and relationships. Traditionally, this has been dominated by work on recognition and diagnosis and the application of standardised diagnostic systems. However, wider considerations are also at play. There have been concerns that psychological therapy services for mental health have been preferentially delivered to certain populations, threatening equity of access. 63 As services are generally oversubscribed, they may be withheld or rationed in ways that may not be made explicit, or inappropriate contact may be negotiated to preserve a relationship with a patient.

Adjudication leads to an offer (or non-offer) of a health service, which may be accepted or rejected. Current models of improving access, which focus on increasing throughput of patients and the efficiency of treatment delivery, have tended to assume that an offer will be sufficient, without necessarily giving enough attention to the likely response. The response of a patient will depend on the ‘fit’ between the offer and his or her identity. Primary care access can be seen as a distinctive field of activity and habitus,64 a set of dispositions that generate practices and perceptions of the way in which people encounter it. As primary care presents a set of technologies and relationships, past experience of illness and service contact coalesce with immediate decision-making about use. The concept of recursivity captures how the response of the system to patients may reinforce or discourage future health actions.

With regard to cultural competence, mental health services in the UK have been criticised for not being responsive to the needs of racial and ethnic minorities. 65 Improving the experiences of such patients requires measures such as hiring practitioners who speak relevant languages and understand the backgrounds of particular patient groups, and a willingness of clinicians to take cultural and ethnic issues into account during their management of patients. 66 One criticism of the notion of ‘cultural competence’ is that it often focuses on particular BME groups and ascribes characteristics to individuals based on a crude group membership. 67 An alternative conception is that cultural competence requires clinicians to take into account the individual values, beliefs and practices of the patient (which may or may not reflect his or her membership of a group). In this way, cultural competence can be seen as a specific form of patient-centredness,68 in which the clinician ‘tries to enter the patient's world, to see the illness through the patient's eyes’ (p. 35). 69 It is also important to pay careful attention to the ways in which cultural context affects our understanding of the nature of clinical evidence. Kirmayer70 has noted that Western diagnostic and conceptual frameworks may be culturally determined, and that cultural communities may have ‘ways of knowing’ that do not rely on the kinds of observational and experimental measures and methods that characterise evidence-based practice.

Gathering evidence

We used these concepts to inform our investigation into the mental health needs of people from underserved groups and how best to meet them, drawing evidence from six separate sources.

Structured scoping review of published quantitative literature

Aim

The purpose of our structured scoping review was to assess the extent and broad results of published quantitative evidence concerning the effectiveness of candidate interventions in improving access to care.

Methods

The population under review was patients with mental health problems in primary care. This included diagnosed disorders and non-specific problem categories such as ‘psychosocial problems’ and ‘stress’. Interventions included the range of candidate interventions that have been used to improve access to care identified in the scoping review. Comparisons included no treatment, usual care and other candidate access interventions. We expected that most published quantitative evaluations of the effectiveness of interventions would focus on effectiveness and cost-effectiveness outcomes and that benefits of access would be implied rather than stated. The focus was on identifying the amount of evidence for each group, the range of interventions that had been trialled and the broad pattern of the results.

The review began with a search of the Cochrane Database of Systematic Reviews (CDSR) and Database of Abstracts of Reviews of Effects (DARE) for previous reviews of candidate access interventions. This was followed by a search of MEDLINE, EMBASE, Cumulative Index to Nursing and Allied Health Literature (CINAHL) and PsycINFO using a range of terms relating to candidate access interventions, combined with a randomised controlled trial (RCT) and primary care filter. Searched were conducted between January and April 2008. The search strategy is provided in Appendix 2. The full texts of abstracts identified by the search were obtained and eligibility was judged by two reviewers. Papers for which eligibility was difficult to judge were assessed by other members of the research team and disagreements were resolved by discussion.

The resources available for the structured scoping review of published quantitative literature were insufficient to allow a full systematic review of all interventions in all populations groups (i.e. comprehensive searches, full assessment of quality and formal synthesis). Rather, the review was designed to identify the amount and type of quantitative evidence relating to access interventions in each of our exemplar groups and to summarise the broad magnitude, direction and consistency of effects.

Results

In the first search we identified 82 reviews and selected 23 as potentially relevant. 71–93 In the primary database search 764 abstracts were checked, which resulted in 105 quantitative outcome studies94–198 (some studies related to more than one exemplar group). Figure 4 shows the flow of studies through the review process.

FIGURE 4.

Flow of studies through the review process.

Interventions were categorised as:

• application of existing mental health interventions in exemplar groups

• modification of existing interventions to make them more acceptable to exemplar groups

• access innovations, that is, interventions designed to increase the number of patients from exemplar groups receiving care.

Some interventions involved more than one category or more than one exemplar group, for example an intervention designed to deliver an existing psychological therapy, which was delivered using an innovative method, or an intervention delivered to elderly ethnic minorities. Examples of access interventions are provided in Table 1.

TABLE 1 - Examples of access interventions

Study	Intervention	Process intervention is designed to affect
Bhugra and Hicks105	Educational pamphlet to influence help-seeking attitudes in South Asian women	Candidacy
Michael et al.156	Community health workers (CHWs) used popular education to identify and address health disparities in Latino and African American communities in a metropolitan area in the USA. CHWs met regularly with community members to identify health needs and rank intervention priorities	Candidacy, navigation, receipts
Cornelius et al.122	24-hour crisis management hotline with a mobile support team and a residential care unit to reach indigent, urban African Americans. Service provides 24-hour crisis intervention and seeks to reduce the financial and social barriers to access	All
McCusker et al.153	A two-stage (screening and nursing assessment) intervention for older patients in the emergency department (ED) who are at increased risk of functional decline. Patients aged ≥ 65 years expected to be released from the ED to the community with a score of ≥ 2 on the screening tool were identified. The intervention consisted of disclosure of results of the screen, a brief standardised nursing assessment in the ED, notification of the primary care physician and home-care providers, and other referrals as needed	Adjudication and offer
White et al.193	Older adults were provided with internet access and received 9 hours of small-group training in six sessions over 2 weeks. Computers were available for continued use over 5 months and the trainer was available 2 hours per week for questions	Candidacy and navigation
Eisses et al.127	Many people with depression in residential care homes for the elderly do not receive treatment because their depression remains undetected. The intervention consisted of a training programme for staff and collaborative evaluation by staff and a mental health specialist of residents with possible depression	Adjudication and offer
Slesnick et al.181	Case management and individual therapy offered through a drop-in centre for homeless youth on substance use, mental health, housing, education, employment and medical care utilisation	Candidacy, navigation, receipts
Boiko et al.111	Audiotaped survey developed for those with low literacy levels. Subjects from 11 migrant camps completed a self-administered audiotaped survey in Spanish to diagnose major depression, substance abuse, panic and generalised anxiety, and domestic violence. Primary care clinics assisted in finding camps and provided follow-up treatment	Candidacy, navigation, receipts

Summary of evidence

Here we summarise the evidence from the scoping review, with reference to each of our eight exemplar groups.

Medically unexplained symptoms

There was inconsistent evidence concerning the effectiveness of training primary care providers to manage medically unexplained symptoms. Only one study showed a positive outcome184 and four reported no effect of training on patient outcomes. 99,145,162,173 Trials of collaborative care models reported benefits,150,188 as did trials of different types of psychological therapy [cognitive–behavioural therapy (CBT, reflecting interviews)]. 114,118,129,169

Advanced cancer

There was evidence that patients with advanced cancer could benefit from collaborative care models128,143 and from psychological therapy. 157,182

Ethnic minorities

There was reasonable evidence that ethnic minority patients could benefit from conventional psychological treatments and those modified for use with these groups. 94,101,104,112,116,117,131,132,142,157,158,161,170 There was limited evidence that training primary care providers in communication skills was effective in ethnic minority patients197 and that simple educational interventions were associated with benefits. 105,139 There was evidence that conventional collaborative care interventions were effective in improving mental health outcomes in ethnic minority patients. 125,128,159,160,192 There was some evidence that community outreach impacted on mental health outcomes. 103,156

Older people

There was reasonable evidence that a number of forms of psychological therapy (including CBT, reminiscence therapy, life review)96,102,119,134,157,195,198 and minimal psychological interventions (self-help, internet access, telemonitoring) could improve outcomes in older people. 106,135,140,143,144,175,176,185,193

Homeless people

There was some evidence that various forms of psychological treatment were effective in improving outcomes in homeless patients. CBT and supportive therapy were effective in reducing depression in women at a homeless shelter,131 whereas case management and individual therapy were effective in improving mental health, substance abuse and access to housing in homeless youths at an urban drop-in centre. 181 Other effective interventions were nursing interventions,186 psychiatric rehabilitation in street-dwelling individuals with psychiatric disorders177 and psychological therapy in cocaine-addicted homeless people. 146

Asylum seekers

There was some evidence that various forms of psychological treatment were effective in improving outcomes in asylum seekers. Training to improve primary care consultations about pain and health education and exercises improved outcomes in immigrant patients. 149,172 CBT and exposure were effective in the management of PTSD in refugees. 166 A group psychosocial intervention for mothers displaced by war was associated with gains in terms of mother and child mental health. 126 CBT and medication were effective in the management of PTSD in refugees165 and CBT was effective in the management of PTSD and panic among refugees. 137 Home visits by school nurses and bilingual teachers reduced depression in South East Asian refugees. 132 Family support and education groups were effective in improving social support and mental health in Kosovar refugees. 191

Sickness absence

There was limited evidence that various forms of psychological treatment were effective in improving outcomes in patients with sickness absence related to mental health. A social support intervention increased levels of satisfaction and attitudes towards employment for disabled African Americans and Latinos. 101 Solution-focused therapy was no more effective than routine care in the management of sick-listed employees. 163 A worksite stress reduction programme was related to improvements in depression and reductions in sick leave. 141 A CBT programme was more effective in reducing sick leave than usual care, but only in patients on short-term sick leave. 151

Adolescents with eating disorders

There was limited evidence that adolescents with eating disorders benefited from psychological interventions. Internet-delivered interventions for eating disorders in adolescents were associated with benefits. 123,133,136,167,196 Motivational enhancement therapy increased readiness to change in eating disorders in adolescents. 124,130 Self-help interventions98,190 and school-based programmes showed preliminary evidence of benefits in the management of eating disorders in adolescents. 171,189

Conclusions

Current evidence on interventions to improve access to care suggested that many patients in underserved groups can benefit from existing interventions, either provided as they are or modified to make them more acceptable. The evidence is particularly well developed for the elderly and ethnic minorities.

At present it is not know whether modifications to interventions to make them more acceptable are associated with increases in effectiveness or in overall levels of access. Although some quantitative research on innovations in access has been carried out, the current data are very limited. This may reflect the fact that evaluating the effect of a service innovation on overall levels of access is complex and not always amenable to a conventional RCT.

Meta-synthesis of published qualitative literature

Aims

By identifying commonalities across disparate published qualitative sources, meta-synthesis works in a similar way to meta-analysis of disparate published quantitative sources in generating a more powerful level of evidence in relation to a particular issue. This qualitative review examined the experiences of and attitudes to mental health treatment of identified hard-to-reach groups in their social contexts to shed light on participants' understandings of the processes of access and their experiences of the health system.

Methods

Search terms generated from known papers and previous research were combined with a previously developed qualitative research filter (see terms 89–115 in example search strategy for MEDLINE in Appendix 3), tested and adapted to run across MEDLINE, CINAHL, EMBASE, PsycINFO, Applied Social Sciences Index and Abstracts (ASSIA) and Web of Knowledge. Databases were searched on 15 May 2008. The resulting abstracts were assessed for relevance and a subset of papers was then extracted. The British Sociological Association's quality criteria199 were used to generate a definitive set of papers for the final synthesis. Key findings were extracted to a standardised proforma and were synthesised within and across groups using the classic lines-of-argument approach proposed by Noblit and Hare. 200

Results

Twenty-one papers were identified as suitable for detailed analysis: two each for advanced cancer201,202 and adolescents with eating disorders,203,204 one each for homelessness205 and employment problems,206 two for older people with depression,48,207 six for medically unexplained symptoms,208–213 three for asylum seekers214–216 and four for BME communities. 29,217–219 Figure 5 shows the flow of studies through the review process.

FIGURE 5.

Flow of studies through the review process. BSA, British Sociological Association.

Summary of evidence

Details of our findings from this analysis have been published in the paper by Lamb et al. ,220 of which the key features are as follows.

Engagement in mental health treatment requires significant personal investment, including of time, material and emotional and social resources, and poses a threat to wider social identities with regard to friends, co-workers and neighbours. These investments in treatment are proportionally higher in deprived communities, where individual resources are low and the stigma attached to mental ill health is high. An understanding of these ‘costs’ by patients and GPs underlies decisions to seek and offer help, but is not explicit in the consultation.

A variety of self-management strategies are used across exemplar groups to maintain social function. These include denial, concealment, reframing and resilience. This could be functional or dysfunctional depending on circumstances. As the limits of self-management strategies are reached, patients withdraw from social networks, focusing their resources on primary roles (e.g. close family and work roles). Overinvestment in these strong bonds results in a sense of insecurity as peripheral roles in weaker networks are neglected. Withdrawal from social networks could be chosen, inflicted or circumstantial.

For example, one study illuminated how some homeless mothers had exhausted the capacity of their social networks, resulting in withdrawal and a lack of confidence in others. 205 The inevitable focus on their children and their role as mother clashed with shelter living, which deprived them of the space and resources for performing the maternal role. Women from South Asian communities could be cut off from their familial and social networks if they failed to conform to the rules and roles of those cultures or the expectations of their partner or family. 29,218

Decisions about whether or not to consult, and the appropriateness of treatment, are often based on attributions of aetiology: mind, body or social. Mind and body may be seen as secondary attributions, used when the inability to adequately perform social roles is legitimated by the sick role. Social problems relate to changes in, diminution of or incapacity to perform – or to be seen to perform – previous social roles that formed their identity.

Conclusions

The person providing the intervention has to demonstrate a willingness to achieve a comprehensive understanding of how the service user/patient views him- or herself in relation to his or her world and presenting problems to effectively engage him or her in care. However, the focus of treatment needs to be much broader, to include an emphasis on re-engagement with the wider social world and its inherent and often complex problems.

Self-help may be counterproductive in some groups as a focus on further investment within the narrow perspective of the primary role (which may be inherently dysfunctional, e.g. being an ‘anorexic’) may be unhelpful as it further isolates the person from reintegrating with the world.

People require interventions that can offer opportunities to fulfil new roles, which build on their existing strengths and help to increase confidence. A challenge of treatment is to address patients' and families' unrealistic expectations, sometimes brought about through their engagement with multiple value systems (especially in BME communities), while not posing a threat to their standing, position and safety within their community.

Depending on the nature of the problem in question, cultural and gender matching may or may not be the most appropriate option for presentation with mental health problems. The traditional ongoing GP relationship was reportedly valued. However, the ability to make informed choices and the diversity of available gatekeepers are paramount to increasing access for these groups. Respect for, and interest in, culture (wanting to find out, rather than knowing) appear to be more important than understanding, although some service users considered this to be important.

Pressure to consult in family pairs or units, and the use of family interpreters, limit opportunities for effective mental health treatment in orthodox settings such as the general practice consultation.

Secondary analysis of qualitative data sets

Aim

The aim of this study was to gather evidence from service users about key issues arising from the scoping review in relation to access to primary care mental health services. This analysis was complementary to the meta-synthesis (see Meta-synthesis of published qualitative literature) in that it allowed access to existing qualitative data (the full text of interview transcripts) rather than the published findings of qualitative studies with only illustrative material presented.

Methods

Qualitative studies contributing data for secondary analysis were sampled by convenience and selected according to the team's judgement on their relevance to our research objectives.

Data comprised anonymised transcripts of semistructured and in-depth interviews with white British, black Caribbean and South Asian adult residents of north-west England. The interviews were conducted within seven different qualitative studies (one unpublished) during a period from 2001 to 2007 (Table 3). 47,48,219,341–343

TABLE 3 - Data sources for secondary analysis

Study	Exemplar ‘hard-to-reach’ group	Year of data collection
Reeve et al.341	People with advanced cancer	2004–5
Gask et al.342	People from BME (South Asian) groups	2007
Edwards and Gabbay343	People with risk of long-term unemployment	2001–2
Chew-Graham et al.48	Older people with depression	2005–6
Chew-Graham et al. (unpublished data)	Older people with depression	2007
Edge and Rogers219	People from BME (back Caribbean) groups	2001–2
Peters et al.47	People with medically unexplained symptoms	2004–5

As the original studies were designed to answer various research questions related to common mental health problems and mental well-being, transcripts that were judged by the primary researcher as irrelevant with regard to our interest in exploring access were discarded after discussion. The next step was a random selection of transcripts from each data set.

Out of the 111 transcripts of the original studies, 19 were discarded as they contained no relevant discussion and 92 were retained. Random selection of transcripts from each data set resulted in 33 transcripts being selected for initial comparative analysis with the remainder available for triangulation.

The process of secondary analysis was based on the methods described by May et al. 344 Initial analysis using a new conceptual framework345 (Figure 6) was conducted by the secondary researcher, who was naive to the original research findings. Initial findings were validated by reanalysis by or analytical comments from primary researchers. The next step was cumulative comparative analysis in which the initial findings could be complemented by additional theoretical sampling from the larger data set. Final summaries of findings for each study were sent to primary researchers for validation. Finally, a condition comparative analysis considered similarities and differences between findings for each study group. The overall approach to analysis was embedded in the interpretative research paradigm with orientation towards narrative and case study approaches.

FIGURE 6.

Analytical model: a heuristic for researching access to primary mental health care. 345 Reprinted from Social Science & Medicine, Vol. 72, by Kovandžić M, Chew-Graham C, Reeve J, Edwards S, Peters S, Edge D, et al. Access to primary mental health care for hard-to-reach groups: from ‘silent suffering’ to ‘making it work’, 763–72, 2011, with permission from Elsevier.

Summary of evidence

We have published the main findings on commonalities between groups in a paper by Kovandžić et al. 345 and some of the findings related to specific groups included in this study in a paper by Chew-Graham et al. 346 Here we outline the main findings, illustrated with some quotations from the data. Our analysis points to an extensive overlap in access-related issues across the observed groups, along with some particularities. The analysis confirms the hypothesised existence of five ways in which problems of access to primary mental health care are clustered across the analysed groups (see Figure 6). Reflecting the ‘taste of data’ we named these clusters as (1) ‘silent suffering’, (2) ‘in between suffering, coping and seeking help’, (3) ‘navigating networks of services’, (4) ‘negotiating acceptable treatment’ and (5) ‘making it work’.

1. ‘Silent suffering’ is characterised by participants' descriptions of periods of psychological distress that were not formulated into mental health problems.

2. Even if suffering is formulated in ways that could lead to formal help-seeking in primary care, people may hesitate to do so. As one older person said:

   Honestly GPs are hopeless. In my opinion GPs don't know mental health.

3. Once the attempt of placing demand is enacted, it is still not certain that the relevant health professional will be reached. Here is an example of the problems experienced by a patient with medically unexplained symptoms:

   So I rang up after a fortnight and I asked the receptionist for an appointment to see Dr [. . .] regarding my knee, the results of my knee. She said ‘I can tell you Mr [. . .], there's nothing wrong with your knee’. So this is the receptionist. I said ‘Excuse me but have you become a doctor over night?’ I said ‘I know there's something wrong with my knee, you don't need to tell me, what do you think I'm coming to the doctors for’. So that was the time I decided to go private.

4. Access to effective and acceptable help is not guaranteed even if a person actually gains access to a relevant health professional. This domain of the problems of access relates mostly to the concept of concordance,347 but also to availability of choices.

5. Even if access to treatment is fully gained, the work of recursivity identified earlier implies the need to look at the factors influencing satisfaction with the quality and personal perceptions of the effectiveness of treatment to secure future access. A South Asian woman described this process as follows:

   If you were here and I was depressed, severely depressed where I wasn't on medication and I was going completely mental I would be sitting here so quiet that you would have to do all the talking [. . .]. But now [after taking medication] I've come to a point where I'm getting my confidence back where I can approach people or talk to people freely and not hesitate so that would help [joining a therapy group].

Conclusions

Regardless of some particularities, there is an extensive overlap in access-related issues across the underserved groups. The main barriers to improving equity of access to primary mental health care could be formulated as multiplied stigma and related lack of effective information. Even if these barriers did not exist, access would remain problematic for underserved groups unless available services were more acceptable and ‘in tune’ with their needs and preferences. Our analysis indicates that developing acceptable services is the main facilitator of equitable access. To achieve this the services need to be pluralistic, adaptive, holistic, resonant with users' needs and socially conscious.

Synthesising evidence

We synthesised the evidence gained from these multiple and diverse sources using a consensus process349 to develop candidate interventions that would take account of known barriers while remaining sensitive to the needs, preferences and priorities of our exemplar hard-to-access groups and other stakeholders. We focused on mental health-related services that have the potential to be commissioned by PCTs, including those delivered by non-statutory organisations.

Our synthesis of these sources of evidence produced key emergent findings, including:

• ways in which (mental) health problems are formulated by people from underserved groups

• risks attached to engaging with mental health interventions offered through primary care

• limitations in access to existing psychosocial interventions and

• better understanding of the dimensions of access.

Generating solutions

Our procedures for generating the candidate interventions involved a series of interactions between the research team and local stakeholders, actual and potential service users from our exemplar groups and a panel of national and international academic experts in the field of primary care mental health (see Figure 2).

1. We began this process with a tightly structured day-long workshop for the entire AMP research team at which we considered the totality of evidence generated from the various sources indicated earlier and used facilitated small-group techniques to generate an initial ‘long list’ of candidate interventions.

2. We then produced a working document as a basis for consulting with the wider health-care community. This document proposed a set of linked interventions in three domains:

   • enhanced community engagement, to raise awareness of the potential benefits of interaction with primary care

   • increased sensitivity of primary care teams to the mental health needs of people from hard-to-reach groups

   • the design and implementation of a range of sensitised psychosocial interventions.

3. Members of our stakeholder group and our panel of experts were invited to comment on the working document, using an electronic proforma.

4. The research team then formed into three subgroups to work up detailed proposals for each of the proposed domains, taking account of service user, stakeholder and expert comments. We then integrated the working groups' reports into a single document.

5. We invited our panel of experts to subject the detailed integrated proposal to a formal peer-review process and, in parallel, held a series of consultations with service users and primary care teams to gather views on the utility and acceptability of the proposal.

6. Finally, we submitted the full proposal for formal ethical review.

To illustrate how these processes have worked, we describe how we focused our intervention strategy on specific hard-to-reach groups – older people and people from BME communities who are experiencing depression, anxiety or medically unexplained symptoms – and why we chose particular sorts of interventions.

Our starting point was to target groups who are least likely to access primary care (such as BME communities) and those who received suboptimal care if they do obtain access (such as older people). Our scoping review of psychosocial interventions identified the largest number of positive outcome studies in these two groups. Findings from our meta-synthesis and our secondary analysis highlighted differing ways in which mental health problems may be expressed, with a strong emphasis on physical manifestations of suffering; hence, we realised the importance of including physical symptoms. We received convergent advice from our local stakeholders, panel of experts and participating PCTs that older people and BME groups should be given highest priority.

As an example of the processes we employed, the rationale for our choice of the form and content of psychosocial interventions to be tested is derived from a synthesis of our multiple sources of evidence, including our programme team meetings. This is summarised in Table 5.

This summary of evidence shows how stakeholders, service users and the qualitative literature all emphasised the need for health-care organisations and health professionals to promote antidiscriminatory attitudes and behaviours. Reviewing the different sources of evidence about increasing access to interventions, we found that the content of psychosocial interventions needs to be presented in ways that are culturally acceptable, that adjust for physical difficulties and that decrease social isolation. It also became clear from the meta-synthesis, the secondary analysis and the stakeholder perspectives that there is a need to incorporate service users' own explanatory models, language and metaphors within the psychosocial interventions. Stakeholder and service user perspectives indicated that the interventions should be available in a number of delivery modes and in both health-care and community settings. 350 None of these sources on their own would have provided sufficient information to allow a credible range of psychosocial interventions to be developed.

Discussion

The AMP programme is designed to generate knowledge and capacity to increase equity of access to high-quality mental health services in primary care. The first steps to achieving this are to find out what high-quality services exist and what the barriers and facilitators are to their successful implementation for people from underserved groups. In this chapter we have explained how and why we gathered and then synthesised evidence from multiple sources to understand the problems and generate potential solutions.

Introduction

In response to our synthesis of evidence, linked to the findings from our methods for generating potential solutions – as described in the previous chapter – we considered it essential to address three distinct but overlapping domains:

• the world beyond primary care, to address the processes that occur before service contact (candidacy and navigation)

• the interface with primary care, to address the process by which services and patients agree on appropriate access to care (appearance, categorisation, adjudication, offer and recursivity)

• the acceptability of interventions available in that setting and the likelihood that they will be attended and used as expected (receipt). 348

We therefore proposed a new multifaceted model, designed to improve access to primary care mental health services for people from underserved communities. The model is presented in Figure 7. The three core components of the model are:

• Community engagement. Community engagement provides the opportunity to improve well-being and access for individuals and the resourcefulness of a system as a whole. It aims to meet local needs and presents an opportunity for reclamation of identity and agency, and includes the option of improved access to sensitised primary care services.

• Primary care quality. To improve the quality of the patient experience when members of hard-to reach groups access primary care, primary care teams need to increase their competence in understanding and responding to the differing ways in which members of underserved groups present suffering and in encouraging them to access relevant services.

• Psychosocial interventions. There is evidence for the effectiveness of psychosocial interventions for many underserved groups. Interventions must be tailored to meet the personal and communal needs of those who may benefit from them.

FIGURE 7.

Multifaceted model of care.

We deployed this model to create the AMP Development Partnership, within which the AMP team sought to work with community, practice and policy ‘champions’ to generate self-sustaining change at a neighbourhood level.

The three aims of the partnership were to:

1. increase equity of access to mental health services through primary care

2. improve the quality and responsiveness of those services

3. test the hypothesis that these changes can best be achieved by delivery of a new complex intervention.

As we explained in the previous chapter (see Generating solutions), in this phase of the AMP programme we decided to focus our attention on two underserved groups: elders and people from BME communities.

Intervention sites

In concordance with initiatives from our collaborating PCTs, we used a locality-based approach to deliver our interventions. We implemented and evaluated interventions in four localities, selected on the basis of the following criteria:

• City: two localities in Liverpool and two in Manchester.

• Population size: each locality had a population in the region of 25,000, currently considered as the optimum size for delivering the full range of integrated primary care services.

• Range of exemplar groups: we needed to ensure access to as many of our exemplar groups as possible while working to promote existing PCT and Department of Health priorities in relation to the mental health of older people and BME communities.

• The Index of Multiple Deprivation (IMD) 2007 (the Department for Communities and Local Government commissioned the Social Disadvantage Research Centre at the University of Oxford to produce the IMD): this is a local area-level measure of deprivation, with overall scores calculated from seven domain indices for income, employment, health, education, barriers to housing and services, crime, and living environment. Lower-layer super output areas (SOAs) are geographical entities introduced after the 2001 census. They contain an average of 1500 residents.

We focused on four specific localities. In each city one locality would enable involvement of ethnically diverse and economically deprived local communities. The other would involve more homogeneous communities in terms of people belonging to the ethnic majority, but they were equally disadvantaged in terms of higher than average employment insecurity and prevalence of chronic diseases.

The two Liverpool localities were Croxteth and Norris Green and Picton and Granby (Figure 8). Croxteth and Norris Green (wards 10 and 18 on Figure 8) are predominantly white British neighbourhoods, which in recent years have become synonymous in the public's view with gang violence. The PCT report that there are high levels of deprivation and worklessness in this area. According to standardised mortality rates (pooled by gender and age) for all wards in Liverpool, Croxteth is in the most deprived quintile for circulatory diseases and Norris Green is in the most deprived quintile for all cancers. 358 Henceforth, we refer to this locality as Croxteth.

FIGURE 8.

Liverpool intervention localities.

Picton and Granby (also known as Princes Park) are wards 20 and 21 in Figure 8. They are located in Liverpool's inner core and lie in close proximity to both the city centre and Wavertree Technology Park. The locality is characterised by ethnic diversity (> 40 languages in use), relative deprivation and extensive urban/housing reconfigurations within a number of urban regeneration projects. The majority of housing in the area is high-density pre-1919 terraces, and there is a large BME population (22.5% for Picton and 37.5% for Princes Park), including a substantial Somali community.

This locality includes the city's main centre for asylum seekers. Tenure patterns differ from national and local trends and are characterised by high levels of rented accommodation. Princes Park ward contains around 10 SOAs: seven of these are classified as belonging to the most deprived percentile of all SOAs in the country. Picton contains around 13 SOAs: three of these are in the most deprived percentile and seven are in the 2–5% most deprived. Approximately one-quarter of the population of each ward has some kind of long-term illness that limits their capacity to work. 359 Henceforth, we refer to this locality as Picton.

The two Manchester localities were Wythenshawe and Longsight (Figure 9). Wythenshawe is Manchester's largest district, a massive housing estate that was created in the 1920s as a garden city where an overspill population could be rehoused from industrial Manchester. It was originally divided into five local government wards: Baguley, Benchill, Northenden, Sharston and Woodhouse Park. In 2003, the ward of Benchill was abolished and its former territory was divided between the wards of Northenden, Sharston and Woodhouse Park.

FIGURE 9.

Manchester intervention localities.

The IMD 2007 score for Baguley ward was 46.82 and it was ranked 15 out of 32 across Manchester for deprivation. 360 In 2006, the Office for National Statistics (ONS) recorded that 9.3% of the population was from non-white ethnic groups. 360 According to the Manchester Residents' Survey of 2007, 21% of Baguley residents reported that they were not in good health,360 compared with 17% across Manchester as a whole and 7% across England. The unemployment rate in 2008 was 4.3%, higher than the average in both Manchester (3.9%) and England (2.5%). 360

Longsight is a residential suburb situated < 3 miles south of Manchester City Centre, which stretches from the City Centre through Ardwick. It has a very ethnically diverse population. According to the 2001 UK census, 52.7% of the population was from non-white ethnic groups. 360 This includes 36.1% who described themselves as Asian or Asian British (including 24.6% Pakistani and 7.2% Bangladeshi) and 9.1% who described themselves as black or black British. In 2006, the ONS recorded that 59% of the population was from non-white ethnic groups. 360 This included 40.9% who described themselves as Pakistani, Bangladeshi or Indian, 6.8% who described themselves as black Caribbean or black African and 3.9% who described themselves as being of mixed race. The remainder were from other ethnic groups, including 2.1% who described themselves as Chinese. The religious affiliation of its residents is one-third Muslim and around one-third Christian. Longsight is home to several mosques, some temples and a gurdwara, plus churches of varying denominations.

The IMD 2007 score for Longsight was 49 and it was ranked 13 out of 32 across Manchester for deprivation. 360 According to the Manchester Residents' Survey of 2007, 13% of residents reported that they were not in good health. 360 This compared with 17% across Manchester as a whole and 7% across England. The unemployment rate in 2008 was 4.5%. 360

Evaluation procedures

Evaluation was a crucial component of the AMP Development Partnership. It was conducted as a coherent process, interlinking the three intervention streams.

The partnership represents a complex, multilevel evaluation, based on the theoretical assumptions that intervening at three levels would be mutually reinforcing and thus more effective than intervening at one or two levels. However, we did not know whether or not this would actually be the case. We therefore allocated different degrees of intensity of intervention in order to have a no-intervention comparator for each of the three streams and hence the ability to explore the extent to which certain interventions reinforce others (Figure 10). This approach was designed to enable us to assess the extent to which access to – and quality of – services were affected by each intervention stream, considered separately and jointly.

FIGURE 10.

Scheme for allocating interventions. Reproduced with permission from Feld R, Woodside DB, Kaplan AS, Olmsted MP, Carter JC. Pretreatment motivational enhancement therapy for eating disorders: a pilot study. Int J Eat Disord 2001;29:393–400.

Within this allocation scheme we used mixed quantitative and qualitative methods to evaluate process and outcome, including:

• the use of quantitative information including routine and project-specific data

• interviews and focus groups with key actors

• ethnographic observations and case comparisons.

We used a framework approach for analysis of qualitative data as this is well suited to an environment in which multiple researchers are gathering information from differing sites. 361,362 An initial coding framework was developed and transcripts were checked against the framework to ensure that there were no significant omissions. Using the qualitative software package MAXQDA (see www.maxqda.com; accessed 20 July 2012), codes in each interview were examined across individual transcripts as well as across the entire data set and allocated to the framework. Using the constant comparative method of analysis,363 broader categories were used for linking codes across interviews. Data were interpreted and analysed within the framework to distil, interpret and structure component statements about the intervention.

We provide further details of our evaluation methods within the descriptions of each of the three intervention streams in the following three chapters.

We received ethical approval to implement the AMP Development Partnership from the Salford and Trafford Research Ethics Committee (reference 09/H1004/67) with subsequent approval for the AMP well-being interventions being granted by the North West 8 Research Ethics Committee (reference 10/H1003/38).

To synthesise our findings and establish our key emergent conclusions, we organised a writing week for the whole of the AMP team, which took place in Warrington in January 2012.

Aims and approaches

Our intention was to develop a simple reproducible community engagement model that could deliver on the aims of the intervention in a way that authentically engaged local stakeholders in the design and delivery of the community engagement interventions while providing for the wider needs of the programme.

Our scoping study (see Chapter 2, Identifying key concepts) stressed that access involves a dynamic and often protracted set of processes and decisions, involving not only patients and health practitioners but frequently also a wide range of stakeholders. For deprived communities the perceived personal ‘costs’ and ‘benefits’ for the patient or his or her family of engaging in treatment may involve different parameters from those in operation in more affluent communities. 220 Improving access for underserved groups could involve addressing any of the barriers in the pathway from patients recognising that they may need help to them seeking, negotiating and engaging in treatment. The nature, scale and impact of different barriers at the local level, as well as how we might best address them, could be understood only by engaging with the communities.

A community engagement approach needs to give communities a sense of collective ownership of the interventions, by involving them in both design and delivery. The role of the intervention team is to facilitate local action and local partnerships.

The community engagement strand of the AMP Development Partnership had four main aims:

1. to develop our knowledge of the range of understandings and attitudes about mental health and well-being in the community

2. to use this knowledge to tailor health literacy approaches to improve awareness of mental health issues in the community, including how, when and where to seek help

3. to address stigma and the acceptability of seeking help and identify the practical barriers to engaging in treatment

4. to develop local knowledge and networks required for primary care training and the psychosocial intervention.

In conceiving community engagement as an inductive problem-solving activity (see Figures 9 and 10), we expected these aims to be negotiated, augmented, refined and adapted to meet the needs of the local community.

We reviewed theory and practice in many different existing approaches to working with the community, which provided useful models and tools. 364,365 We drew pragmatically on a range of techniques that have previously been used in community engagement for addressing health issues. The University of Central Lancashire (UCLAN) model,366 built on participatory action research, is perhaps the best known in the UK. Although having a number of techniques and principles in common, developing a model for community engagement in a complex intervention context involved a different resolution of the principles underpinning participatory action research. A key difference between our approach and the approach of many others was the need to deliver tangible outcomes on a controlled range of issues at the local level within the intervention time frame. The community engagement element of the AMP model was then tailored specifically to meet the needs of community engagement in an intervention context. In doing so we turned to earlier precedents. Braithwaite et al. 367 developed a model for health promotion in minority communities that combined knowledge acquisition with organisation and development. Following Lewin,368 we conceptualised community engagement as a particular case of an inductive problem-solving process (Figure 11), taking place within the context of the research intervention.

FIGURE 11.

Community engagement as a problem-solving process.

Such cycles are routinely used in health service improvement and in policy development, for example in ‘plan–do–study–act’ cycles,12 in which they are conceived as an inductive spiral (Figure 12). 368

FIGURE 12.

Spiral of action research cycles. 368

Community engagement model

Our community engagement model involved four components, implemented in sequence:

1. information gathering

2. community champions

3. consultative focus groups (CFGs)

4. community working groups (CWGs).

Our approach emphasised involving the community in reflecting on the problems and issues, and producing considered action at the local level in the relatively short time frame available. We also recognised the importance of gradually building trust when working with underserved communities with complex needs. We worked through participation, involving local people and service providers in both negotiating and delivering on the agreed aims. Drawing on the wider traditions of action research, this can be seen as empowerment through action and delivery.

We recognised that community engagement had certain predefined aims and parameters within the overall remit of the AMP programme. We therefore looked for an approach that would allow us to engage with the agendas of multiple community stakeholders, find common ground for action over the intervention period and help to foster partnerships and alliances that could lead to considered progress on more difficult issues in the longer term.

Because information gathering (step 1) was required to inform the AMP primary care and psychosocial interventions, this was carried out in all four localities.

In accordance with the wider intervention–control design, active community engagement (steps 1–4) was carried out in two of the four localities. The randomisation process involved stratification by city and by type of community. The Liverpool intervention site was randomly selected as Croxteth, with a focus on the mental health needs of elders. In Manchester, therefore, the intervention locality was Longsight, where we worked with the South Asian community. Within the timescale of the entire AMP programme, we were able to allow 2 years for implementation of our community engagement model.

Information-gathering findings

In Picton and Wythenshawe, the information-gathering phase was not followed by community interventions from the AMP team. We therefore present findings from this exercise in these two localities as discrete entities.

Community intervention findings

The community interventions in Longsight and Croxteth differed. In Croxteth the community champion worked through his existing organisation, rather than by adopting the CWG approach. Whereas the community champion and the CWG in Longsight addressed a wider range of issues, the focus of the community champion in Croxteth tended towards influencing local strategic issues. This may have reflected, in part, differences in needs and, in part, the differing styles of the two community champions. Although there were common themes between the localities (see Discussion), we report the work undertaken in each locality as a separate case study.

Longsight

Information gathering

Researchers initially immersed themselves in the local area, walking through the streets, local shops, parks and market, making field notes and taking photographs. The field notes, based on observations, conversations and naturalistic enquiry,361 provided an overview of local expertise and knowledge and gave us important insights into the motivations and needs of strategic stakeholders.

The first stage of observation focused on the activities and the ‘feel’ of the area:

I pass the Himmat Support Centre, which has a bouncy castle in the garden. There are several South Asian women standing near the front door. An Asian man in a wheelchair is being wheeled out of the gate, and a black man stands close to the gate, shuffling his feet; he appears to have learning difficulties and is talking to an Asian man that may have been his carer. The centre is busy and, on the basis of this afternoon's events, seems to be utilised by a varied cross section of the community.

Researcher notes

The second stage involved understanding these local activities through the eyes of the participants:

I sat and talked informally to [. . .] a member of the community mental health team [CMHT], about his work in Longsight and [. . .] a football team for people with enduring mental health issues which he organises. He talked about his belief in the effectiveness of a community centre based approach to supporting people with mental health issues, and this was his motivation to become involved with [local church] and bring some of the CMHT's work to the centre.

Researcher notes

Working through the contacts developed during the observation phase, the go-along interviews often elicited more nuanced, personal or emotive understandings of the local area and people:

One woman said to me, ‘We used to have servants but now we are servants.’ She was so ashamed that her son was a taxi driver when in Pakistan they had chauffeurs.

Researcher notes

The field notes from these exploratory site visits fed into team debriefings and the development of a database of local contacts and stakeholders.

Community champion

The Longsight community champion had lived in the area for a number of years. A British-born Muslim, with family in Pakistan, she had previous experience of working on cross-cultural health issues in other localities. As community champion she organised and ran the CWG meetings.

Consultative focus group

We held four CFGs during the programme. Timings were tailored to the availability of attendees and the progress made by the CWG. We held four CFGs throughout the programme (Table 6). Timings were adapted to the availability of attendees and the progress made by the CWG. Details of attendees at the CWG have not been fully collated. The imam and community police officer from group 1, for instance, continued to participate regularly in the working group although they did not attend further CFGs.

TABLE 6 - Longsight CFG: participation

Attendee	Group 1	Group 2	Group 3	Group 4
Third sector	4	3	7	5
Third sector (Bangladeshi)	0	0	2	3
Police	2	0	0	0
Faith leader	1	0	0	0
GPs	2	0	1	0
Practice managers	2	1	1	1
Public health	0	1	1	1
Mental health counsellors (GPs)	0	1	2	1
Domestic violence counsellors	0	0	1	1
Teacher (secondary)	0	0	1	0
AMP research team	4	5	8	4
Community champion	0	1	1	1
Total	15	12	25	17

The agenda for each meeting tracked the problem-solving cycle (see Figure 11) and the emerging priorities and areas for action identified by the group are summarised in Figure 13. Researchers used field notes and recordings to produce a summary of the discussions and action points for the CWG.

FIGURE 13.

Longsight CFGs: evolving agendas.

Community working group

The Longsight CWG began in July 2010. Led by the community champion the group included local service providers, a public health representative, a community police officer, a local imam and the practice manager of one of the intervention practices, as well as members of the AMP research team. The group had a high level of cultural knowledge and experience specific to working within the Asian community of Longsight. They met once a month, working towards the priorities identified in the first CFG of improving understanding of mental health issues and where to seek help, resulting in the following outputs.

Output 1: mental health calendar

The group decided that producing a mental health-themed calendar would have significant benefits over leaflets and posters as it might be displayed in the house, particularly in the kitchen, traditionally an area where women meet in the South Asian community. It could provide a talking point, potentially broadening its impact, as it might reach members of the community who would not be comfortable taking mental health information into their own homes.

To ensure that the calendar was relevant to the community, consultations were carried out with local Asian women's and men's groups as well as with local imams. These groups felt strongly about including three main areas: jinn and black magic, forced marriages and domestic abuse.

With one page per month, the calendar introduces different themes and issues in mental health and well-being matched to the seasons (see Appendix 5). Themes are illustrated through appropriate quotes from Quranic texts developed with local imams. The text runs in parallel columns in English, Urdu and Bengali. The final poster page reviews key messages and has contact details for local services.

The 2011 calendar was launched in late December 2010. The community champion appeared on Asian Sound Radio to promote the calendar and the work of the group. The calendar was distributed through local GP practices, third-sector partners, including mental health groups, Pakistani and Bangladeshi community groups and mosques. The calendar was refreshed and updated for 2012 and distributed through the same outlets and through new partners from the CFG.

Output 2: Facebook (and Twitter) group – ‘Mental Health in South Asian Communities’

A priority identified by the third CFG was a mechanism to share up-to-date information on changes in local services, events and resources for those working with the South Asian community. The concept behind the Facebook group was that members could update their own information as and when it changed. The Facebook page could also serve as a repository for good mental health materials available in South Asian languages. Facebook was chosen as a medium as it was widely used, enabled linkage with existing groups and was low cost and sustainable.

A member of the CWG proposed this idea and has been instrumental in setting up and administering the Facebook group. The group was launched in January 2012, drawing initially on the work carried out for the calendar. It has generated interest from practitioners in and beyond Longsight. At the end of June 2012 the Facebook page was generating up to 15 hits per day; the Twitter feed had 32 followers with an additional 3345 linked through LinkedIn accounts.

Output 3: relaxation CDs

Another issue identified by the CWG was the long waiting times for counselling and mental health services in South Asian languages, particularly Bangla. Many members of the group were looking for resources and materials that could help patients and clients while they were waiting for treatment. After reviewing all of the materials they could access they found little that they considered fully appropriate and freely available. Members of the group suggested a CD tailored to the South Asian community, including breathing exercises and muscle relaxation. The group has been working together on the content. They will consult through relevant community groups to ensure that the CD is culturally appropriate and addresses the community's needs. When complete, CDs will be manufactured and distributed through local health and third-sector services. The CD will also be freely available to download from the Facebook group.

Croxteth

Information gathering

We gathered information from walking and driving around the area, from visits to well-being resources and from local media and the internet. We undertook 19 individual interviews and one focus group with older people.

We witnessed a pride in the area and enthusiasm for making Croxteth a better place. There were concerns about the lack of public amenities, especially transport. Safety was an issue, related in part to danger from traffic and in part to fear of violent crime. We found that Croxteth has relatively few well-being facilities and that the majority of third-sector organisations appear to be economically vulnerable. However, we were impressed by the informal networks between various services and the presence of ‘human hubs’ who provide a sense of continuity and stability. There is a desire among community workers for greater contact with GP practices and recent initiatives have included a small local farm and health trainers.

Problems in accessing well-being services included a lack of knowledge of well-being, a lack of motivation for those who are depressed, a reluctance to engage with groups, inadequate transport services, safety concerns, financial difficulties, stigmatisation and sensory impairments. Proposed solutions to these problems included a web-based database of well-being resources, more health staff and community policing, improved public transport, and services made more physically accessible and provided free at the point of use.

Community champion

The community champion in Croxteth was an older man who had lived and worked in the locality all his life. As an active part-time employee with one of the third-sector organisations in the area, he was familiar with many of the problems facing local residents. Likewise, he was already aware of many of the local resources and had existing links with some of the key stakeholders.

Consultative focus group

Four CFG meetings were held in Croxteth between March 2010 and September 2011. Table 7 shows the background of the participants at each meeting. Apart from members of the AMP team, members of third-sector groups were the most likely to participate consistently in the CFGs. The participation of the police, community health workers, GPs and PCT staff (including commissioners) increased over time. Local councillors and housing and business leaders each took part in one CFG, whereas faith leaders did not participate in any of the meetings.

TABLE 7 - Croxteth CFG: participation

Attendees	Group 1	Group 2	Group 3	Group 4
Third sector	6	6	9	4
Housing associations	0	0	1	0
Police	0	0	1	1
Local councillors	0	0	2	0
Business leaders	0	0	1	0
Faith leaders	0	0	0	0
GPs	0	0	1	3
Community health professionals	0	1	4	1
Primary care trust	0	1	2	3
AMP research team	6	5	5	4
Total	12	13	26	16

The agendas of the four CFGs are summarised in Figure 14.

FIGURE 14.

Croxteth CFGs: evolving agendas.

Outputs 1 and 2: ‘What's in L11’

Following extensive consultation with local third-sector organisations, and with the support of the AMP team and the CFG, the community champion produced two linked resource documents:

• a ‘What's in L11’ leaflet

• a well-being calendar for 2011 (see Appendix 6).

The leaflet and calendar both focused on the availability of, and access to, local third-sector and health resources of relevance to the mental health and well-being of older people in Croxteth and Norris Green.

These resources were distributed by the community champion, members of the research team, the local AMP well-being facilitator and the community police officer. Locations included primary care and community health centres, libraries, community centres, third-sector organisations and local pharmacies. Some 150 copies of the calendar and about 2000 copies of the leaflet were distributed.

The community champion also worked to engage the major local housing association and transport company in promoting the AMP agenda, the former through the inclusion of material in its regular resident newsletters and the latter through advertisements on local buses. Despite considerable effort, neither organisation was persuaded of the merits of pursuing these initiatives.

Strategic involvement

The community champion actively advocated the AMP community engagement agenda and promoted awareness of the mental health needs of older people through extensive networking with a variety of local and city-wide organisations. Local networking focused on fostering communication between the third-sector organisations represented in the ‘What's in L11’ literature. The city-wide organisations included the PCT's prevention and mental health group and Liverpool LINK Neighbourhood Open Space events. Here, his focus was on policy interventions. The strategic involvement of the community champion enhanced the opportunities for the AMP development agenda to influence PCT policy (see Chapter 8, Policy interventions).

Evaluation

Discussion

When reviewing our activity in light of the aims of our community engagement strategy (see Aims and approaches), we have evidence of a degree of success in all four domains:

1. Our information-gathering and community-mapping procedures enabled us to develop our knowledge of the range of understandings and attitudes about mental health and well-being in each of the four communities we were working with.

2. We used this knowledge to tailor health literacy approaches to improve awareness of mental health issues in the community. Working through community champions, CFGs and (in Longsight) a CWG we generated well-received outputs in the form of calendars and related material that addressed culturally relevant mental health issues and provided clear information on accessible resources.

3. Community members, clients and professionals considered that the materials and the processes of local engagement and partnership working developed through the programme could help address stigma among older people and members of BME communities, increase the acceptability of seeking help and identify (and overcome) practical barriers to engaging in treatment.

4. We were able to develop and utilise our local knowledge and networks to enhance the effectiveness of AMP trainingplus and our well-being intervention (see also Chapter 7).

The model that we adopted provides informative case studies of community engagement within the context of a complex intervention. The balance between community participation and working within the remit of the wider intervention differed between the two intervention localities. The strategic influence in Croxteth was successful in its own terms, but allowed for less active participation by local community organisations. The approach in Longsight relied on engaging with an active community, who recognised the value of the work and for whom the activities aligned with some of their own work.

AMP trainingplus

The underlying rationale of the AMP model is the need to promote primary care services that recognise and accommodate the various ways that users and their communities ‘frame’ common mental health problems. This put the initiative broadly in line with ideas of ‘patient-centred’ and ‘culturally responsive’ services. However, for a successful outcome we needed to improve (where applicable) both patient and provider experiences of services. Simply offering training to GPs, even if this includes work on skills as well as knowledge, does not lead to improvement in outcomes for patients with mental health problems (apart from, possibly, newly diagnosed depression). 376 Improvement of primary care ‘quality’ and organisational change are both important. 377 Any initiative needs to be seen as relevant to practices' everyday work, and flexible enough to respond to the challenges posed by the needs of an individual practice. 378

As we explained in Chapter 2, the AMP programme assimilated evidence on the nature of access problems for underserved groups, and views of how access might be more equitable. We identified five key messages relevant for primary care (Box 1).

These messages were incorporated into a syllabus checklist, which was the foundation of AMP trainingplus, designed to be a made-to-measure, flexible training package for primary care teams that would improve primary care quality, offering:

• a training programme tailored to fit with local needs and priorities

• practical support to review/address local organisational barriers to access, for example appointment booking, consultation times and interpreter services

• links to a wide range of internal and external resources, including community-based psychosocial interventions within the AMP trial.

We developed AMP trainingplus with three interlinked strands:

• Knowledge transfer, including a training component of up to six sessions, initially chosen from a menu of subject options. Our aim was to foster a space within which team members could reflect on practice and learn from colleagues.

• Systems review: we offered practices intensive observation (up to 1 week) to identify organisational and structural features that may impede or promote access by underserved groups. This was centred on reception areas and appointment booking systems.

• Active linking: we offered to raise awareness of other relevant organisations and resources that had been mapped and logged by the AMP team.

The knowledge transfer component was based on Grol's model,379 which emphasises the need to combine expertise from specialists with coproduction of ‘new’ knowledge within the practice. Grol noted that no single approach or focus is superior in all settings or for all problems, with different problems and settings needing individualised exploration of what would work to improve care. He does, however, offer guidance, including the need to bring expertise from specialists to the front line of care (e.g. to improve treatment of depression); promoting knowledge transfer through organisational changes and service integration; the importance of defining clear targets for change; the need for exploratory work to understand specific settings and target groups; and the importance of positive attitudes to change and a culture of learning and collaboration in target setting.

AMP trainingplus was developed for all practice staff, both those who have clinical contact with service users and those who have incidental contact with service users. We intended it to be practice-centred and responsive to the needs of each practice and to include topics such as consultation skills (detection of mental health problems, negotiating diagnoses, initial management, psychosocial skills, behavioural activation) and topics of broader relevance such as cultural competence, exploration of personal attitudes and values and working with interpreter services. The particular emphasis and specifics would depend on the local priorities determined by practices and primary care teams.

We identified the following intended outcomes:

• increased staff awareness of, recognition of and respect for diversity

• a change in consultation or encounter style and content, including communication, listening and negotiation, use of the evidence base, models of illness, use of interpreters and use of motivational techniques

• a change in consultation or encounter outcomes, including referral, signposting, prescribing and satisfaction

• a change in processes and systems, for example booking systems, use of interpreters, and referral to psychosocial interventions and social prescribing.

Engaging with practices

A vital part of AMP trainingplus was the need to embed the training in the locality within which each practice was situated. We needed to identify the scope and particularities of the issues to be addressed in each locality. We therefore undertook a process of gathering evidence and understanding its local relevance and identifying and logging third-sector resources, a procedure that we called community mapping. We have described this in Chapter 4 (see Information-gathering findings).

Working with our PCT partners, we identified four primary care teams (or practices) in two areas of each PCT based on mapping populations of underserved groups (see Chapter 3, Intervention sites). We needed to recruit practices to become involved in the AMP programme as either intervention or control practices. Our aim was to recruit two practices in each of the four localities selected to participate as intervention sites, with two others acting as control sites. We randomised the practices in each locality to the intervention arm or the control arm.

For the intervention practices, we sent letters of invitation to named senior GP partners and to practice managers along with written information explaining the purpose and nature of AMP trainingplus (see Appendix 8). Members of the research team then attempted to contact the named GP (and other GPs in the practice) and the practice manager to ascertain interest in the study.

When a practice expressed an interest in the programme, we arranged one or two pre-meetings to introduce the AMP staff and programme and answer questions about involvement in the study.

Topics covered in the initial meeting included summarising the results from phase 1 of the AMP programme; introducing community mapping (and community engagement for those localities where this was part of the intervention); negotiating the possibility of conducting ethnographic observations in the practice and arranging convenient slots for ethnography based in the reception area; and negotiating the possibility of an initial training session at which we would present the AMP model and the results of the community mapping and practice ethnographic observations, and explore local applicability and the practice's learning needs. We emphasised that there was no minimum commitment required on the part of the practice.

Practices were advised that the AMP team would facilitate practical arrangements for the first and any subsequent sessions (rooms, equipment, refreshments, etc.). We offered financial remuneration to cover the time that staff spent participating in AMP training and activities, which could include payment for locum cover or reimbursement for the use of the out-of-hours service.

At a later stage we also approached the two practices in each locality that had been randomised to act as control sites. We sought to meet with members of the practice team to introduce the programme of work and the aims of phase 2 of the AMP programme. In this meeting we aimed to negotiate the possibility of GPs and practice nurses using agreed computer codes to record the management of people with identified mental health problems; we also used this meeting to introduce the AMP well-being intervention and referral guidelines (see Chapter 6, Testing the intervention).

We also sought to discuss these two additional items with the eight intervention practices. We did this either as part of our knowledge transfer sessions or in separately arranged meetings.

Ethnographic observations

In this section we summarise the ethnographic observations that we conducted within the AMP intervention practices. The focus of the observations was on encounters between patients and receptionists at the front desk and encounters between staff within the practices. A more detailed account of this work may be found in the paper by Hammond et al. 380

Delivering AMP trainingplus

Evaluation

Discussion

Aim and objectives

In this arm of the AMP Development Partnership our aim was to develop and evaluate an evidence-based, feasible, acceptable and culturally sensitive psychosocial intervention for older people and people from BME communities.

To determine the content, delivery method, acceptability and cultural sensitivity of an intervention targeting older people and people from BME communities, we sought to synthesise data from previous work streams to design the intervention protocol; and to develop and deliver a training package for mental health workers to effectively deliver the intervention protocol.

We then conducted an exploratory trial in which we sought to:

• test the intervention protocol, exploring the delivery of the intervention as well as adherence and acceptability

• estimate key parameters for a definitive trial of the intervention protocol, through examination of recruitment rates and a comparison of outcomes in patients receiving the intervention and those receiving usual care.

Developing the intervention

The methods used to develop the intervention are summarised in Figure 15.

As noted in Chapter 2 (see Generating solutions), we held a 1-day synthesis workshop with the entire research team. The aim of the workshop was to synthesise the sources of evidence noted in Chapter 2 and produce key findings about the ways in which (mental) health problems are formulated by people from underserved groups; the costs attached to engaging with mental health interventions offered through primary care; and the limitations in access to existing psychosocial interventions.

We drew up a matrix of results, with each row of the matrix detailing one of the key intervention design issues that we wished to address, and the columns referring to the results from each individual data set (see Chapter 2, Generating solutions). This matrix was used as the platform to derive the key principles to be incorporated into the intervention. These included:

• evidence-based psychosocial interventions

• focused on both psychological issues and social issues

• aimed to improve depression and anxiety

• aimed to decrease social isolation

• working with patients' explanatory models

• destigmatising.

Following our synthesis, the intervention work-stream team drafted the details of the intervention and developed the training materials. The focus of the psychosocial intervention was on older people in Croxteth and Wythenshawe and people from BME communities in Longsight and Picton. Within the BME communities we focused on people from South Asian backgrounds in Longsight and people from Somali backgrounds in Picton. These groups were identified as being of high priority by both our stakeholder consensus exercise and the local PCTs. They include those less likely to access primary mental health care (BME communities) and those likely to receive substandard care when they do access primary care (older people). Our initial scoping review of psychosocial interventions (see Chapter 2, Structured scoping review of published quantitative literature) demonstrated the largest number of positive outcome studies in both of these groups.

Focus groups

To ensure that our intervention was acceptable and culturally sensitive, we undertook 11 focus groups with (potential) service users and service providers in each of our four localities (Table 14). South Asian and Somali service user groups were conducted either through interpreters or (in Manchester with two of the three groups) by a researcher who spoke Urdu, Hindi and Punjabi.

TABLE 14 - Focus groups for psychosocial interventions

Focus group	Location	n	Gender	Ethnicity	Age range (years)
Elders	Croxteth	11	Female 9, male 2	All white British	49–79
Practitioners (older persons services)	Croxteth	8	Female 5, male 3	All white British	28–62
Elders	Wythenshawe	5	Female 5	All white British	49–66
Practitioners (older persons services)	Manchester University	10	Female 9, male 1	White British 9, white (other) 1	30–59
South Asian	Longsight	17	All female	Pakistani 14, Bangladeshi 1, Indian 2	24–65
South Asian	Longsight	14	All female	Bangladeshi 14	22–55
South Asian	Longsight	9	All male	Pakistani 7, Indian 2	35–65
Practitioners (South Asian community)	Manchester University	17	Mixed	Mixed – majority South Asian	Not recorded
Somali	Granby	10	All female	Somali	25–70
Somali	Liverpool	10	All male	Somali	61–80
Practitioners (Somali community)	Liverpool community centre	4	Male 2, female 2	Mixed	Not recorded

We present the main emergent themes for older people and South Asian and Somali people.

Older people

Flexibility of provision

This included the need for longer sessions, more sessions, short waiting times, home visits, group sessions as well as individual sessions and the ability to respond to patient choice. Practitioners supported signposting for management of practical issues. They were also positive about group work for elders in community settings.

Stigma

The shame historically associated with mental health issues meant that elders were reluctant to talk about them outside the family. Therefore, deciding to access services relating to mental health might be difficult.

Overcoming barriers

When they needed to look for help beyond their immediate circle of contacts, elders identified methods of accessing the services that could help them, which supported multiple points of access, bypassing primary care if necessary. They did not always know what was available and what they were entitled to. Leaflets in community centres might help to inform elders of the existence and details of the services.

There was concern that GPs often fail to recognise psychological difficulties in their elderly patients or believe that there is nothing that they can do in the face of multiple problems. Elders identified receptionists as people who could obstruct access to their GP and were not always approachable. The perceived power wielded by receptionists was described vividly by one focus group participant who said, ‘Receptionists think they are God’, and added: ‘Before, it was like getting through Gestapo trying to get through a receptionist.’ This reinforced the importance of being able to bypass primary care.

South Asian people

Access

As with elders, stigma was a key issue. Participants recommended not naming interventions as ‘mental health’ programmes but rather as ‘well-being’ or ‘positive health’ programmes. There were barriers to access related to culture, religion, fears about confidentiality and difficulties with language. Use of interpreters could be problematic. Participants thought that accessing the psychosocial interventions should not be complex. Patients could be self-referred as well as being referred by friends, family or the health-care providers. Interventions should be provided by people who had expertise, good interpersonal and communication skills and the ability to develop trust and confidentiality. The place where the psychosocial intervention was delivered carried more significance. Professionals and service users both expressed preferences against home-based care. Service users preferred places that could be accessed without feeling self-conscious or embarrassed.

Social and practical elements

Participants thought that psychosocial interventions should empower people with skills and tools that they could take away and use in their daily life, including art therapy, jewellery-making, card-making, group-based exercise sessions, informative sessions on health and intergenerational conflicts, and leisure trips. There was also enthusiasm for exercise sessions.

Type of therapy

Some preferred individual therapy, whereas others seemed more positive about group therapy as a more effective method of destigmatising mental health. Few service users were aware of CBT and problem-solving. Most participants thought that patients should be able to start a new therapy or switch between therapies as appropriate.

Somali people

Content

Professionals thought that the design of interventions should be user-led, as professionals do not have enough understanding of the needs of the population. There was little concept of ‘mental health’ as an issue, or of counselling/talking therapies as a way of treating mental health, as these were generally considered to be the realm of the GP.

For talking therapies, Somali group members preferred a private weekly format to a group setting. They also thought that an effective intervention would be socially oriented, not focused on mental health, because of associated stigma.

Access

Professionals and members of the Somali groups both thought that the intervention should be community based. There were mixed views about how help would be accessed through doctors. A male group member said: ‘If we are sick, we go to the doctor. But, I don't know where to go with stress.’

FIGURE 15.

Developing the well-being intervention.

The well-being intervention

The well-being intervention focused on brief cognitive–behavioural strategies aimed at decreasing anxiety and depression and social isolation. The title ‘well-being’ was designed to maximise engagement and reduce stigma and to enhance linkage with our community engagement interventions. We incorporated a patient-centred interview and shared problem statements, goals and well-being plans. The intervention was delivered by well-being facilitators.

Participants were offered an initial patient-centred assessment session with a well-being facilitator and collaboratively devised a well-being plan. The well-being plan specified desired health- or social-care changes based on self-identified goals. Significant emphasis was placed on the patient as the ‘agent of change’, incorporating patients' previous experiences and coping strategies into the intervention and addressing stigma, expectations and illness trajectory to better engage patients. Once the goals had been identified, participants chose up to three ways to obtain support to achieve them: individual sessions with their well-being facilitator; group sessions with other participants; or direction (signposting) to appropriate public or third-sector services in their locality (Figure 16).

FIGURE 16.

The well-being intervention.

Testing the intervention

Our objectives were to test the acceptability of, adherence to and delivery of the intervention protocol and to estimate key parameters for a definitive trial of the intervention protocol, through examination of recruitment rates and comparison of outcomes between patients receiving the intervention and those in usual care.

Well-being outcomes

The well-being quantitative analysis involved two feasibility trials in two populations with separate randomisation sequences, so the data were described and analysed separately. The small numbers in the feasibility studies mean that caution must be exercised when assessing differences between groups.

Recruitment and randomisation

In Croxteth and Picton we took referrals for 12 months from late September 2010. In Longsight and Wythenshawe we took referrals between late November 2010 and 31 December 2011.

In the elders group, 84 patients were referred. Fifty-two referrals were by GPs, seven by primary care mental health teams, four by voluntary organisations and 10 by others; 11 were self-referrals. Thirty-seven patients (44%) met the inclusion criteria and agreed to be randomised. At 20 weeks, 33 (89%) completed follow-up (96% in the intervention group and 71% in the control group).

In the BME group, 39 patients were referred. Of these, 20 were referred by GPs, two by primary care mental health teams, five by voluntary organisations and four by others; eight were self-referrals. Twenty (51%) met the inclusion criteria and agreed to be randomised. At 20 weeks, 16 (80%) completed follow-up (79% in the intervention group and 83% in the control group).

The Consolidated Standards of Reporting Trials (CONSORT) diagrams are shown in Figures 17 and 18.

FIGURE 17.

The CONSORT flow chart for the elders sample.

FIGURE 18.

The CONSORT flow chart for the BME sample.

Baseline sociodemographic characteristics of the elders sample are shown in Table 16. As with most primary care mental health trials, most respondents were female, with a mean age around 64 years, nearly three-quarters were retired or otherwise not working and most were not involved in community activities. Levels of distress were relatively high, with a mean CORE-OM score of around 20 and a mean PHQ-9 score of 18.

Statistical significance testing of baseline differences is not recommended in trials generally and is inappropriate with small numbers when large differences between groups are likely to be not statistically significant because of limited power. Visual inspection of the data indicates reasonable comparability in terms of gender, age and ethnicity, but differences in employment and community engagement. Importantly, baseline levels of distress on the CORE-OM (primary outcome) and depression were comparable, although there were differences in other health measures.

TABLE 16 - Sociodemographic characteristics of patients included in the AMP feasibility study: elders sample

SD, standard deviation.

All health scales except for the EQ-5D are scored so that a high score is indicative of poor health.

PHQ-9 scores of 10+ are usually indicative of clinically significant depression, and scores of > 20 are indicative of more severe major depression.

CORE-OM scores of ≥ 25 are usually considered indicative of severe distress.

Characteristic	Well-being intervention (n = 23)	Usual care (n = 14)	Total (n = 37)
Sex, n (%)
Male	6 (26.1)	3 (21.4)	9 (24.3)
Female	17 (73.9)	11 (78.6)	28 (75.7)
Age (years), mean (SD) range	65.64 (10.29) 50–84	60.70 (8.16) 53–78	63.77 (9.73) 50–84
Ethnicity, n (%)
White British	20 (87.0)	13 (92.9)	33 (89.2)
White other	2 (8.7)	1 (7.1)	3 (8.1)
Mixed	1 (4.3)		1 (2.7)
Employment, n (%)
Working full- or part-time	5 (21.7)	5 (35.7)	10 (27.0)
Retired	12 (52.2)	3 (21.4)	15 (40.5)
Not working or unemployed	6 (26.1)	6 (42.9)	12 (32.4)
Engaged in community activities, n (%)
Yes	3 (13.0)	6 (42.9)	9 (24.3)
No	20 (87.0)	8 (57.1)	28 (75.7)
Baseline measures, mean (SD)a
PHQ-9b	17.70 (5.14)	19.00 (5.35)	18.19 (5.19)
GAD-7	15.91 (4.31)	13.20 (6.12)	14.88 (5.16)
WSAS	26.00 (9.48)	22.86 (13.35)	24.81 (11.03)
EQ-5D	0.29 (0.36)	0.31 (0.45)	0.30 (0.39)
EQ-5D health state (0–100)	44.65 (19.59)	53.93 (25.81)	48.16 (22.27)
CORE-OM (mean × 10)c	19.85 (6.72)	20.00 (6.90)	19.91 (6.69)

Baseline sociodemographic characteristics of the BME sample are shown in Table 17. All participants were female, with a mean age around 40 years, with most from Pakistani or Bangladeshi groups. In total, 20% had a degree and nearly one-third were involved in community activities. Levels of distress were relatively high, with a mean CORE-OM score of around 25 and a mean PHQ-9 score of 19.

Visual inspection of the data indicates reasonable comparability in terms of age and education, but differences in ethnicity and social engagement (although these are very sensitive to the small numbers in the groups). Importantly, baseline levels of distress on the CORE-OM (primary outcome) and depression and the other health measures were comparable. It is notable that the PHQ-9 scores in the BME and elder samples are comparable, but the CORE-OM scores were much higher in the BME sample.

TABLE 17 - Sociodemographic characteristics of patients included in the AMP feasibility study: BME sample

SD, standard deviation.

All health scales except for the EQ-5D are scored so that a high score is indicative of poor health.

PHQ-9 scores of 10+ are usually indicative of clinically significant depression, and scores of > 20 are indicative of more severe major depression.

CORE-OM scores of ≥ 25 are usually considered indicative of severe distress.

Characteristic	Well-being intervention (n = 14)	Usual care (n = 6)	Total (n = 20)
Sex, n (%)
Male	0 (0)	0 (0)	0 (0)
Female	14 (100.0)	6 (100.0)	20 (100.0)
Age (years), mean (SD) range	38.92 (9.29) 25–56	43.02 (14.50) 21–58	40.15 (10.87) 21–58
Ethnicity, n (%)
Pakistani	6 (42.9)	4 (66.7)	10 (50.0)
Bangladeshi	6 (42.9)	2 (33.3)	8 (40.0)
Punjabi Indian	1 (7.1)	0 (0)	1 (5.0)
Somali	1 (7.1)	0 (0)	1 (5.0)
Qualifications, n (%)
Degree or higher degree	3 (21.4)	1 (16.7)	4 (20.0)
Eighth grade	0 (0)	1 (16.7)	1 (5.0)
Sixth grade	1 (7.1)	0 (0)	1 (5.0)
Fifth grade	1 (7.1)	1 (16.7)	2 (10.0)
Third grade	1 (7.1)	0 (0)	1 (5.0)
A levels	0 (0)	2 (33.3)	2 (10.0)
O levels	4 (28.6)	0 (0)	4 (20.0)
High school	1 (7.1)	0 (0)	1 (5.0)
None	2 (14.3)	1 (16.7)	3 (15.0)
Other	1 (7.1)	0 (0)	1 (5.0)
Engaged in community activities, n (%)
Yes	3 (21.4)	3 (50.0)	6 (30.0)
No	11 (78.6)	3 (50.0)	14 (70.0)
Baseline measures, mean (SD)a
PHQ-9b	19.36 (3.20)	19.50 (1.76)	19.40 (2.80)
GAD-7	18.07 (3.52)	17.83 (1.17)	18.00 (2.97)
WSAS	28.85 (8.11)	26.83 (8.35)	28.21 (8.01)
EQ-5D	0.21 (0.36)	0.20 (0.49)	0.20 (0.39)
EQ-5D health state (0–100)	17.14 (12.51)	17.50 (11.29)	17.25 (11.86)
CORE-OM (mean × 10)c	25.01 (5.86)	24.92 (4.56)	24.98 (5.38)

The 20-week outcomes

Elders

The 20-week outcome data for the elders is shown in Table 18. Caution must be exercised in interpretation of outcome data in a feasibility study as the small numbers mean that baseline differences can occur through chance and power to detect differences is limited.

The results suggest that the group receiving the well-being intervention improved compared with the group receiving usual care.

The results are plotted in Figure 19 using a standardised mean difference (effect size) measure. The largest effects are seen for CORE-OM and PHQ-9. The small numbers mean that these results are not statistically significant, but the magnitude is fairly substantial compared with that seen for other psychosocial interventions in primary care. 392 These results are not adjusted for any potential baseline differences.

TABLE 18 - The 20-week outcome data for the elders sample

Outcome	Well-being intervention, mean (SD), n	Usual care, mean (SD), n
CORE-OM	14.32 (8.37), 22	19.71 (8.58,) 11
GAD-7	11.41 (6.96), 22	12.27 (8.17), 11
PHQ-9	11.82 (8.05), 22	16.55 (6.25), 11
WSAS	18.20 (11.84), 22	23.55 (14.30), 11
EQ-5D	0.40 (0.44), 22	0.27 (0.44), 11

FIGURE 19.

Forest plot of the effects of the well-being intervention on older patients. CI, confidence interval; SMD, standardised mean difference.

Black and minority ethnic groups

The 20-week outcome data for the BME sample is shown in Table 19. Again, caution must be exercised in the interpretation of these data as the small numbers mean that baseline differences can occur through chance and power to detect differences is limited.

The results are plotted in Figure 20 using a standardised mean difference (effect size) measure. The effects are smaller than those seen in the elders sample. The largest effects are seen for PHQ-9 and EQ-5D, for which the impact of the intervention was similar to that seen in comparative data from primary care. 390 These results are not adjusted for any potential baseline differences.

TABLE 19 - The 20-week outcome data for the BME sample

Outcome	Well-being intervention, mean (SD), n	Usual care, mean (SD), n
CORE-OM	20.98 (7.40), 11	20.32 (6.32), 5
GAD-7	13.45 (4.53), 11	14.00 (4.64), 5
PHQ-9	13.99 (4.91), 11	16.00 (6.44), 5
WSAS	22.30 (11.93), 10	24.40 (10.45), 5
EQ-5D	0.35 (0.46), 8	0.21 (0.38), 5

FIGURE 20.

Forest plot of the effects of the well being intervention on the BME sample. CI, confidence interval; SMD, standardised mean difference.

Evaluation

Discussion

We developed and evaluated a psychosocial intervention tailored to older people and people from BME communities. We synthesised data from previous work streams to design the intervention protocol, refined this with help from focus groups and developed a training package for mental health workers. We then conducted an exploratory trial to test the intervention protocol.

We found that the well-being intervention could be delivered as planned. It was generally acceptable to patients and there was evidence of adherence. The majority of patients preferred one-to-one sessions, although the options for group work and signposting were valued.

We investigated parameters for a definitive trial of the intervention protocol through examination of recruitment rates and a comparison of outcomes between patients receiving the intervention and those in usual care. Evidence from our quantitative analyses indicated that patients offered the well-being intervention showed greater improvement than those offered usual care, with the largest effects seen for reduction in depressive symptoms. The magnitude of difference was comparable with that seen for other psychosocial interventions in primary care. 390 Our evidence of effect sizes will be helpful in estimating sample sizes for a future definitive trial.

Despite considerable effort on the part of the AMP research team, our recruitment rates were lower than anticipated. Recruitment to primary care trials in the UK is routinely problematic, and is particularly difficult for mental health trials. 393–395 Therefore, determining recruitment rates and the most effective recruitment strategies are critical steps in the development of a definitive trial. However, available evidence to support recruitment strategies is sparse. 396

In other mental health trials we have found that successful recruitment has been achieved by mass-screening lists of registered primary care patients,397 which results in a large numbers of recruits, although the overall response rate tends to be low. Although it might be possible to adopt this strategy for older people, who do tend to register in primary care, our knowledge of the barriers to recruitment identified in earlier stages of the AMP project made such an approach problematic. Mass screening would also be less likely to reach underserved members of BME communities who may not have registered as patients with primary care teams or, if they have, may not be readily identifiable from practice lists. Therefore, the current recruitment rates should be seen in a context in which one of the most effective strategies was not available.

The relatively high number of self-referrals was a positive finding, although many of these did not translate into randomised participants. In Chapter 7 we discuss the reasons for and implications of this finding.

Recruitment would have been enhanced if the design of the evaluation had allowed the inclusion of more practices, as we had numerous enquiries from potential participants who were not registered with one of our 16 practices. Recruitment would also have been enhanced by having a better knowledge of those practices that members of the Somali and South Asian communities tended to register with.

Many trials report initially poor levels of recruitment. Recruitment over a longer period (e.g. 18 months) would enable GPs to become used to the referral process and the service, and to receive positive feedback from their patients, potentially encouraging further referrals. It would also help to simplify the eligibility criteria. For example, in Croxteth and Picton, certain postcodes were initially specified. This caused problems and was soon changed so that a patient's postcode made no difference to his or her eligibility. Providing explicit information about the referral process during the AMP training sessions enabled receptionists and administrative staff to know exactly what was required (in one practice two referrals went astray and resulted in the GP losing confidence with AMP). Arranging for the GPs to meet the well-being facilitators at an early stage may also help – as, for example, in Croxteth – so that they know who will be working with their patients.

Cultural issues were of great relevance. On the one hand, it is important to ensure that well-being facilitators are seen as culturally appropriate by the local communities. This applies as much to older indigenous populations as to particular BME groups. On the other hand, our experiences may support an argument for reducing the cultural specificity of well-being interventions. There is a trade-off between specificity and potential sample size (most evident for the Somali community). Reducing specificity would also limit the risk of a particular community seeing itself as stigmatised, rather than prioritised, by the offer of an intervention. However, it is also clear that for the South Asian participants cultural specificity was valued and seen as important. Different strategies may therefore be needed to engage particular communities.

It should be noted that attrition rates at the first follow-up were not unusually high, despite the difficulties experienced in engaging these groups. Minimising attrition is essential to ensuring that baseline comparability provided through randomisation is maintained, and that the sample size achieved at recruitment is not significantly reduced in the main analysis. The study suggests that, although there is still work to be done in encouraging these groups into trials, their experience once recruited may not lead to particular problems of retention. However, it is possible that the low levels of recruitment reflect engagement of a very select group of patients in the trial and the low level of attrition might be a particular characteristic of that group; the low level of attrition might not generalise if wider participation is achieved.

Introduction

We now consider the extent to which the different elements of the AMP Development Partnership may have impacted on each other to increase equity of access to mental health services through primary care and improve the quality and responsiveness of those services.

As we demonstrated in Chapter 2, two main barriers to equity of access are lack of information and multiple forms of stigma, whereas a major facilitator is the communicated availability of acceptable mental health services. 346 In Chapter 3 we provided our rationale for implementing a complex intervention model with three interlinked components (Figure 21).

FIGURE 21.

Multifaceted model of care.

In Chapters 4–6 we have explained how each component of the model was developed and assessed. The next challenge was to develop ways of evaluating this model, capturing both process and outcome and enabling the interactions between levels to be assessed.

If we take Figure 21 as a metaphor for the possible mechanisms of the AMP Development Partnership, we can see it as a set of interlinked cogs within a gearing system. A gear is a rotating machine part having cut teeth, or cogs, which mesh with another toothed part to transmit torque. Two or more gears working in tandem are called a transmission; they can produce a mechanical advantage through a gear ratio and thus may be considered a simple machine.

One way of looking at this model is to assume that the community engagement cog starts the process in motion, driving the primary care cog, which in turns drives the psychosocial intervention cog. However, because all cogs are interlinked, it is possible that activity in any one can influence the activity and speed of any others, and that concurrent activity in two could substantially affect the speed of the third. Conversely, lack of activity in one cog could reduce or stop the transmission of torque elsewhere in the system.

Improved responsiveness in a system like this can have several different outcomes, and hence can be viewed from several different perspectives. It would be possible to examine community-level outcomes, for example changes in the level and experience of mental health stigma within local communities, or changes in the extent of self-help activity at local community level. Given that the focus of our programme was on increasing equity of access to high-quality mental health services through primary care, we decided in this report to focus on the two most directly relevant outcomes: the extent to which the integration of the AMP model affected access to psychosocial interventions sensitive to the needs of underserved groups; and the extent to which the model impacted on the quality of mental health-care delivery within primary care overall.

Access to the well-being interventions

The assumption underlying the combined AMP model was that increased access to the well-being interventions would be related to the intensity of the other two AMP interventions. Therefore, we would expect access to the well-being interventions to be higher in the localities that were offered the AMP community engagement programme and for patients registered with practices that participated in AMP trainingplus; and highest for those who were in a position to benefit from both community engagement and practice training activities.

Consequently, we focus on patterns of referral and recruitment into the well-being interventions.

Referrals to the well-being interventions were more likely to come from the two localities that were offered the community engagement intervention. Figure 22 shows that there were a total of 80 referrals from these localities (50 in Croxteth and 30 in Longsight) compared with a total of 43 from the other two localities (nine in Picton and 34 in Wythenshawe). Not all referrals came through the practices. Referrals did not appear to be influenced by whether or not practice teams had been offered AMP trainingplus. There was no difference between the combined number of referrals from the practices offered AMP trainingplus and the combined number of referrals from the practices not offered the training intervention.

FIGURE 22.

Referrals to the AMP well-being intervention by locality and practice. Numbers in rows are of practice referrals. Numbers in columns are of locality referrals.

Recruitment to the well-being interventions, in contrast, was associated with the offer of AMP trainingplus, but not with community engagement. As we can see from Figure 23, 41 (72%) of the 57 participants recruited were registered with one of the practices offered the AMP training programme, whereas only 16 (28%) were registered with a practice not offered the training. Thirty-two (56%) of the recruited participants came from one of the two localities that had received the community engagement intervention, whereas 25 (44%) came from one of the other two localities.

FIGURE 23.

Recruitment to the AMP well-being intervention by locality and practice. Both column and rows numbers are of patients registered with practices.

Synthesised analyses of our evaluation interviews provide a set of explanations for these divergent quantitative findings, which can be grouped under the headings of awareness, acceptability and authority.

Awareness of the existence of the AMP well-being intervention appeared to be more influenced by the community engagement intervention than by the practice training intervention. This may be partly because all of the 16 AMP practices were invited to refer patients to the well-being intervention, regardless of whether or not they were invited to participate in AMP trainingplus. The absence of a community engagement intervention in Picton, combined with the fact that many members of the Somali community gather information by word of mouth rather than through written materials, were likely to have contributed to low levels of awareness of the well-being intervention in that locality.

There appeared to be similar levels of acceptability at community and practice levels. There was evidence of considerable enthusiasm among many community organisations for the psychosocial intervention.

A participant with a community role with a housing trust in Wythenshawe told us:

The first problem [with depression] is people don't know that they've got it. They just think they're going through a low period in their life and they're just sad or . . . a turn of events: you lose your job . . . but I think the great thing that you're doing is you're trying to catch them before they go into that severe depression stage, and work on them mild to moderate before they're at that severe stage. You know, where they feel like they can't do anything. You're sort of hitting them mild to moderate where, you know, they're feeling low and it's about changing their mindset. And I work with other organisations with cognitive behavioural therapy and it's about making them aware of how they're feeling and changing that into a positive. So I think it's a great idea.

A service commissioner in Liverpool told us:

I think about the whole thing . . . so the AMP is improving access to mental health and primary care and then you have got the well-being facilitators . . . and . . . really exciting and well done for the bit that says in our, you know, here are the . . . facilities, here are the local things that we have in our local community . . . which for us in a practical world is really helpful. Really helpful for GPs, really good to say, listen here's, you know, here's what you've got . . . Then they have got the well-being facilitators . . . and the people who offer the therapy sessions – I think they are the same people.

Acceptability was enhanced by word-of-mouth reports from participants in the well-being intervention, who subsequently encouraged friends to self-refer.

However, other community stakeholders, particularly in Picton, were concerned about the potentially stigmatising effects of an intervention targeted at a specific ethnic minority. A Somali British woman told us:

Yes definitely you're making them very, very isolated because you're going after them as a group, I could see it now where this project would be like running after Somali, stop it's a well-being project here, don't you want to get involved, and they're like why do we need the well-being project, but if you stopped like ten people at the door who were a mix of people, multicultural people and you said to them this is this project is available I really think you should be interested, would you put your name down to just come and see, then you're more likely to get more people actually wanting to attend so I think that's a better way of doing it.

Concern was also expressed by potential service users that self-referral to the well-being intervention could result in a negative reaction from the primary care team, and the risk of being denied access to future primary health care.

Although practice teams were generally aware of the availability of the AMP well-being intervention, their views of its acceptability to the practice team varied. Variations in practice perspectives on its acceptability appeared to relate to three factors: workload prioritisation, which was usually dependent on the enthusiasm or otherwise of the practice manager; ease of navigation, which was particularly problematic in Picton where several GPs reported that their practice record systems made it difficult for them to readily identify patients from Somali backgrounds; and variable trust in the perceived efficacy of the intervention.

A sense of authority was stronger in primary care than among community organisations or potential self-referrers. GPs in particular are used to making decisions about whether or not and when to refer on to other agencies, whereas members of community organisations often do not feel sufficiently responsible or powerful to do so.

A community worker in Wythenshawe told us:

I suppose when it first started we thought the GPs were going to initially be doing that [referring].

A project worker in Liverpool said:

If there was anyone that I was sort of concerned about, I always discuss it with the vicar. And, you know, if there's anything we can do in that way then by all means, but I wouldn't be making decisions by myself; I don't think it would be part of my role.

General practitioners who had been offered AMP trainingplus were thus in a particularly strong position to make effective referrals insofar as their sense of authority was combined with a greater awareness of the types of patient who would be likely to need and benefit from a culturally sensitised psychosocial intervention.

Mental health care within primary care

When thinking of how the community engagement and psychosocial components of the AMP Development Partnership affected the primary care element, we should not use the metaphor of cogs and gears too rigidly, to imply that these were fixed entities. Rather they represent a dynamic set of processes. The components or strands did not routinely have set or rigid boundaries, much as threads are not always visible or separable within an overall fabric. They were defined rather by ongoing negotiation of boundaries between and within the three strands, so that they shaded into each other.

We should note that this complex interplay was to a large degree prescribed by the model itself and the underlying intention to promote opportunities for fruitful dialogue. Such interplay was effected in localities by (and depended on) individual research staff, health workers, etc., moving virtually, and sometimes physically, between community, primary care and well-being ‘spaces’. Movement in such cases was modulated by many factors, including assigned roles and priorities and perceived relevance in respect of the process and content of ‘other’ strands. It was also constrained by such roles and priorities: as a GP from Practice B noted, involvement in AMP-related meetings came ‘at the expense of time actually seeing patients’.

There were four main ways in which the primary care strand (AMP trainingplus) was affected by the other strands of the AMP model: community mapping, the well-being intervention, CFGs and understanding the AMP model.

Conclusions

Referrals to the well-being intervention appeared to be enhanced by our community engagement strategies, whereas recruitment was more likely for patients registered with practices that were offered AMP trainingplus. These differences were related in part to a greater sense of enthusiasm among community groups, in contrast to a greater sense of authority bestowed on and perceived by GPs.

The finding that receipt of a well-being intervention could have an impact on participation in community activities emphasises the potential fluidity of roles with the AMP model,372 especially beyond the medical zone of care. It also links to our findings from phase 1 (see Chapter 2) on the value of helping others as a way to support personal recovery.

With regard to the impact of the other elements of the AMP model on primary are, our analysis suggests that our community mapping activities were seen as strongly positive from a primary care perspective; the offer of the well-being intervention was initially welcomed but was seen by many GPs to contain various difficulties in terms of opportunities for recruitment; and there were clear, though varying, benefits from the interaction between practice training and participation in CFGs. It was not clear that GPs and other members of primary care teams fully grasped the implications of the overall AMP model; however, it was also unclear whether, or to what extent, that mattered in terms of their involvement and commitment.

Further analysis is needed on the impact of the primary care and well-being intervention elements on the community engagement element of the AMP Development Partnership. Further research is needed to clarify and extend the tentative conclusions that we have drawn in this chapter. These matters are discussed further in the next chapter.

Introduction

In phase 3 of our programme we moved from experimentation to implementation. We used evidence gained from phase 2, together with our knowledge of and experience in the diffusion of innovation377,399,400 to inform implementation strategies at three complementary levels:

• we developed an educational intervention, a web-based resource pack on ‘how to do it’, linked to local stakeholder meetings

• we undertook policy interventions at local and national levels

• we initiated service interventions to implement key aspects of the AMP development model in differing clinical contexts.

In this chapter we describe these initiatives. We then make recommendations for future research in this field and reflect on the implications of the AMP programme for policy and practice.

Educational intervention

Our aim in creating a dedicated AMP website was to describe what we did within the AMP Development Partnership to inform and change professional attitudes and behaviour. Our intended audience was the range of specialist and non-specialist mental health staff working in primary care, and also academics and policy-makers in a position to influence practitioners.

Working with an information technology specialist, we describe the key elements of the interventions, which we have described in detail in the previous chapter of this report. The home page introduces the AMP model and explains how and why we developed it. The next pages present the three elements of the model and explain our evaluation procedures. We then give pictorial and verbal descriptions of the four intervention localities. We have included pages for study findings as they emerge and for further information, including AMP-related publications. There is also a members area and frequent opportunities for site visitors to print or comment on particular pages and topics. The website can be viewed at www.amproject.org.uk (accessed 10 June 2013).

The AMP website went live on 9 June 2012. We launched it through our national and international academic and policy networks. We highlighted it within our presentations of the AMP programme at two stakeholder events in Croxteth and Picton in mid-June. Both meetings were held in community venues relevant for the study target groups. These were attended by 25 members of third-sector organisations, three primary care practitioners, five primary care commissioners, two city councillors and others including housing association managers and study participants. We also linked it to the Facebook and Twitter groups set up by the Longsight CWG (see Chapter 4, Community intervention findings).

We also held two stakeholder events in Manchester in September 2012. We invited service users and providers, including patients recruited to the well-being intervention, local residents and service providers who participated in go-along interviews, all four practice teams in each area, participants in our focus groups and working group, people we met during the community mapping phase, and others representing organisations working with older people and the South Asian community. We presented key findings from all three strands of the AMP programme and two patients spoke at each meeting to share their experiences of receiving the well-being intervention. The meetings included questions and discussion.

Feedback on the website has been very positive. By 31 July 2012 there were 613 page views on the website, by 90 unique visitors. An international primary care leader commented: ‘Beautiful “human” stories. Impressive how you have found a consistent integrative line to make it possible to specify and diversify according to local conditions.’

We will keep track of web-related activity to identify where, when and how our AMP website influences attitudes and practice in the field of primary care mental health.

Policy interventions

In this section we describe the impact of the AMP programme on policy at both local and national levels.

Local policy impact

The AMP programme was set up from the start to encourage a close working relationship with local policy stakeholders. The funding for the programme was directed through Liverpool Primary Care Trust and we included senior clinical managers from both Liverpool Primary Care Trust and NHS Manchester as programme principal investigators. Our explicit intention was to maintain dialogue and interdependence, to enable rapid implementation of key findings as they began to emerge from the phase 2 experimentation stage of the AMP programme.

We now illustrate the impact of this relationship by describing how it enabled the AMP team to have a major influence on a review of primary care mental health policy in Liverpool.

Ongoing engagement

There was early and sustained engagement between the AMP team and representatives of Liverpool Primary Care Trust at different levels. At the managerial level, the programme was supported by the PCT's medical director and by its research and development (R&D) lead. During phase 1 and phase 2, we also had active liaison with the trust's director of stakeholder engagement, public health neighbourhood manager and community development workers, as well as members of its social prescribing group and its teams for integrated, mental health and public health commissioning and for delivering race equality.

We found the following benefits from this engagement as phase 2 of the AMP programme unfolded:

• There were early synergies between the AMP development model and the trust's strategic direction, which is built on a neighbourhood management model. 401,402

• Our community engagement activity in Croxteth, especially the CFG meetings and the active role of the AMP community champion (see Chapter 4, Community intervention findings), helped to strengthen relationships between the trust and third-sector organisations. In so doing it embedded key features and messages of the AMP development model, especially the active linking function.

• AMP trainingplus activities raised awareness about the AMP well-being intervention and the important role of third-sector organisations in providing emotional, practical and social support for underserved groups in primary care settings.

Liverpool primary care mental health and well-being strategy

Our ongoing negotiations and discussions with PCT commissioners, as research findings emerged from the evaluation of the AMP development model, provided empirical evidence to support the trust's evolving strategy for primary mental health and well-being. This culminated in the active involvement of members of the AMP team in planning and delivering a series of three workshops between May and July 2011.

These workshops were designed to develop a coherent pathway for mental health service provision (both within and without primary and secondary care) with local buy-in from service providers. The workshops were well attended, with energetic participation from 99 third-sector organisations and specialist psychological and secondary care mental health providers.

The agreed pathway that emerged from these workshops closely followed the AMP development model (Figure 24). It includes multiple entry points and referral routes, an emphasis on collaborative care and the introduction of new mechanisms to encourage greater use of non-statutory service providers. Community resources are seen as central to the proposed model of care.

FIGURE 24.

Model of care proposed by the Liverpool primary care mental health and well-being strategy. CMHT, community mental health team; CW, community worker; EIT, early intervention team.

This pathway has subsequently been confirmed as being central to the primary care mental health strategy for Liverpool Primary Care Trust.

The aim of the strategy is to deliver a coherent and neighbourhood-based care pathway for primary mental health and well-being. Its core elements are:

• a practical offer to reduce social exclusion

• a social offer to reduce social isolation and promote health and well-being

• a psychological offer to provide access to evidence-based psychological therapies

• an integrated offer of primary and secondary care work together.

Changing contexts

It is important to be mindful of political, economic and organisational changes and their potential impact on the implementation of this agreed policy. Since the summer of 2011, major changes have occurred in the organisation of primary health care and in the amount of funding available to statutory and voluntary bodies. These have posed risks to the successful implementation of Liverpool Primary Care Trust's primary care mental health strategy:

• Following the passage of the 2012 Health and Social Care Act, Liverpool Primary Care Trust ceased to exist on 30 March 2013. 403 Many of its functions will be taken over by the newly formed Liverpool CCG. It is not yet clear how the CCG will approach commissioning for primary care mental health.

• A consortium of service improvement and development staff, service users and stakeholders, under the protectorate of the trust's Integrated Commissioning Team, currently act as a patient ‘watchdog’. The future of this group, including its stakeholder engagement function, is uncertain.

• The city council has gone through major workforce reduction, based on the need to save over £140M between 2011 and 2013. This major reduction in costs has put third-sector, non-statutory services at risk, including many of those identified as crucial to the delivery of the primary mental health and well-being strategy.

Service interventions

Recommendations for research

Our high-level conclusion from the AMP programme is that innovative and complex interventions aimed at service design need an innovative and complex combination of methods to enable robust assessment and evaluation. We consider that our combination of methods has been imaginative, original and fit for purpose.

Our work to understand the problems of increasing equity of access to primary care mental health services, described in Chapter 2, involved a combination of scoping reviews of published quantitative literature, a meta-synthesis of qualitative literature, a grey literature review, secondary comparative analyses of existing qualitative data sets and qualitative analysis of new interview and focus group material.

Our evaluation of the linked components of the AMP Development Partnership, described in Chapters 3–7, was based on a multilevel quasi-experimental design, which gave us the ability to test the component parts of the intervention separately and in differing combinations. It then involved the deployment and subsequent analytical fusion of comparative case studies, ethnographic investigations, qualitative analysis of new interview material (gathered from focus groups and individual respondents), descriptive quantitative analysis of routinely collected data and statistical evaluation of our pilot RCT.

Our evaluations have addressed the majority of the research objectives that we set for the AMP Development Partnership. They have also led to further questions, which, in combination, create a substantial agenda for future research into effective service redesign for primary care mental health services, with a focus on increasing equity of access for underserved groups.

Our community engagement model (see Chapter 4, Community engagement model) needs to be replicated and tested in other geographical areas, including rural as well as urban communities, and with a focus on differing underserved groups. It could also be usefully applied to other health priorities, for example cardiovascular disease or diabetes, for which inequities of access are known to exist.

Although we have argued that all four steps contained within our community engagement strategy were necessary to achieve maximum benefit, we cannot be certain that this was the case. Scaling up community intervention strategies across a considerably larger number of localities would allow for study designs that included some or all of the components, and would provide the opportunity to compare, for example, strategies with or without community mapping as a first step. It would also provide evidence to support or refute our assertion that a differentiation of function between CFGs and CWGs is the most effective means of delivering locally meaningful change.

Before AMP trainingplus (see Chapter 5, AMP trainingplus) can confidently be recommended as a template for quality improvement in primary care, it would benefit from replication with other primary care teams and across commissioning groups, perhaps with a focus on other underserved groups and different health priorities.

Although we consider that the initial ethnographic component was necessary to promote effective engagement, as we did not randomise at this level we do not know that this was the case. Similarly, although we found evidence of impact of – and synergies between – all three elements of the training programme, our study design did not allow for scrutiny of the independent effect of each element. Future large-scale studies should include quasi-experimental designs within similar primary care quality improvement initiatives to assess the separate and combined effects of these various elements.

Our quantitative measures of primary care outcomes were limited. Future studies should consider more detailed, systematic approaches to routine data collection than were possible in this programme.

Our design for a set of well-being interventions (see Chapter 6, The well-being intervention) with a focus on specific underserved groups had demonstrable validity and acceptability with most of the communities to which it was offered. Further qualitative investigation is needed to clarify the reasons why acceptability varied between different cultural groups, and to increase knowledge of cultural differences in understanding of mental health and attitudes towards help-seeking.

The apparent effectiveness of the well-being interventions, as assessed in our pilot trial, indicates that they are worthy of further investigation in larger RCTs. The results of the pilot trial will be helpful in determining sample sizes for future trials.

We are confident that psychosocial interventions can and should be revised to make them more relevant to the needs and expectations of differing cultural groups. However, further consideration needs to be given to two aspects. The first is whether or not the range of individual, group and signposting options is necessary for all populations, given that most participants opted for individual sessions with some signposting options. The second is how to deal with the need for cultural specificity of the psychosocial intervention without the perception of stigmatising of that population by publicly focusing on the nature of its difficulties. There are also difficulties in carrying out RCTs in populations with limited engagement in and understanding of research.

Our combined AMP model, with its simultaneous community, primary care and psychosocial interventions and its layered experimental design, is unique. We have demonstrated the internal added value of this combined approach in relation to its effects on increasing access to psychosocial interventions and enhancing the quality of primary care. Further analysis is needed to elucidate the impact on local communities in terms of increasing health literacy and reducing stigma, as well as to elucidate the optimal modes of ‘active linking’ between relevant local organisations. Work is also needed to estimate the economic impact of the AMP model in terms of direct costs and benefits of consequent changes in health-care use and indirect societal costs and benefits.

Although we have indicated the feasibility of the application of this model in related local contexts, before it can confidently be proposed as an effective method for increasing equity of access to high-quality primary mental health care it needs to be applied and rigorously tested in other settings. Our focus has been on underserved or marginalised groups in urban settings within a relatively prosperous country with a strong infrastructure of primary care services and a relatively well-developed range of existing psychological therapies. It is not clear how well our model would operate in countries where primary medical care is less developed or where there is little access to any psychological therapies. We have noted that third-sector organisations, crucial to the success of our community engagement strategies, are particularly vulnerable to change in economic circumstances; thus, their capacity is likely to vary depending on socioeconomic circumstances operating far beyond the control of research teams.

Our evaluation process was both necessary and useful, but it did carry risks and limitations. First, the inherent complexity of the process rendered analysis challenging. We needed to exercise careful judgement when drawing inferences across and between differing methodological perspectives, and to remain aware of the requirement to draw pragmatic conclusions of relevance to health-care planners and policy-makers. Second, it involved a perpetual balance or tension between achieving the core aim of the AMP programme – to increase equity of access to high-quality primary care mental health services – while maintaining sufficient flexibility to engage and maintain effective partnerships with our research stakeholders.

Although generalisation of our findings is permissible, it is essential to remain mindful of their contextual nature. The three elements of the AMP Development Partnership would have the same general structure if applied in other localities, or with a focus on other underserved groups, but the details of their content and processes, and hence their anticipated outcomes, would have been different. This would have been so within the confines of the AMP programme if our randomisation processes had led to the full community engagement strategy being offered in Wythenshawe and Picton instead of Croxteth and Longsight; if a different set of practices had been randomised to the offer of AMP trainingplus; or if our focus for the well-being intervention had been on other underserved groups. A fortiori, it will be the case for others planning to redesign primary mental health-care services and improve access for underserved groups elsewhere in the world.

Implications for policy and practice

Contribution of authors

All authors made a substantial contribution to the concept and design of the study or the acquisition of data or the analysis and interpretation of data, and drafted the manuscript or revised it critically for important intellectual content. All have given final approval of the version to be published.

Professor Christopher Dowrick (Primary Medical Care) coled the programme and the co-ordination of this report.

Professor Carolyn Chew-Graham (Primary Care) led the primary care element of the programme and the phase 1 secondary qualitative analysis and primary data collection.

Professor Karina Lovell (Mental Health Nursing) led the psychosocial intervention element of the programme.

Dr Jonathan Lamb (Research Associate) led the qualitative meta-synthesis, coled the framework analysis and took responsibility for drafting the community engagement element of the report.

Mrs Saadia Aseem (Research Assistant) contributed to the scoping quantitative review and the BME psychosocial evaluation.

Dr Susan Beatty (Research Associate) led the quantitative data gathering for the primary care evaluation and contributed to the other elements of the intervention.

Professor Pete Bower (Primary Care) led the scoping reviews and the quantitative analysis of the psychosocial intervention.

Dr Heather Burroughs (Research Fellow) coled the framework analysis and contributed to the community engagement interventions.

Ms Pam Clarke (Research Assistant) contributed to the initial stakeholder interviews, the community engagement intervention and the older people's psychosocial evaluation.

Dr Suzanne Edwards (Senior Research Fellow) supported the programme co-ordination and contributed to the community engagement and primary care interventions.

Professor Mark Gabbay (Primary Care) supported the development and management of the programme and critically reviewed the report.

Ms Katja Gravenhorst (Research Assistant) contributed to the primary care element of the intervention.

Mr Jon Hammond (Research Assistant) contributed to the primary care element of the intervention.

Dr Derek Hibbert (Research Associate) led the grey literature review and coled the primary care element of the intervention.

Dr Marija Kovandžić (Research Fellow) led the qualitative secondary analysis and contributed to the community engagement and primary care interventions.

Professor Mari Lloyd-Williams (Palliative Care) supported the development of the programme and critically reviewed the report.

Dr Waquas Waheed (Psychiatry) contributed to the community engagement and primary care elements of the programme.

Professor Linda Gask (Primary Care Psychiatry) coled the programme and the co-ordination of this report.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, CCF, NETSCC, PGfAR or the Department of Health.

National Institute for Health Research Programme Grants for Applied Research

Programme Summary

Aims and Objectives: To increase equity of access to high quality primary care mental health services for hard to reach groups. Objectives: to clarify mental health needs in these groups, identify relevant services and their barriers and facilitators, develop and test credible interventions, establish dissemination strategies and integrate new services into primary care.

Background and Rationale: Mental health problems impose substantial burdens patients, carers, and health-care systems. A wide range of interventions have demonstrable efficacy in improving the lives of people experiencing common but disabling mental health problems. However many people with high levels of mental distress are disadvantaged either because they are unable to access care (e.g. black and ethnic minorities, homeless, asylum seekers, adolescents with eating disorders), or because access does not lead to adequate care (e.g. people with unexplained symptoms, prolonged sickness absence, advanced cancers or older people with depression). Key research issues needing to be addressed include complex problem presentation and diagnostics in primary care, user defined perspectives on problem formulation, management and outcome, engagement and communication, and the professional context of mental health encounters.

Research Environment: Our team is strongly represented within the new National School for Primary Care, and key NHS bodies including the Mental Health Research Network and North West Primary Care Research Network.

Research Plans: The programme will link knowledge about barriers to access for our exemplar groups with the perspectives of users, and developing creative ways to meet the particular needs of these groups. Phase 1 (months 1 to 24): Understanding the Problems and Generating Potential Solutions. We will undertake a narrative review of barriers to access to care, and interventions to overcome these barriers, using a combination of systematic review, qualitative meta-synthesis techniques and interviews with service users and providers. We will focus on user perspectives on mental health issues in relation to each of the hard-to-reach groups, in order to understand whether there are any access issues specific to each group, and how different issues affect them all. We will then consider the results of our reviews, and use a consensus process to develop exemplar interventions which take account of known barriers while remaining sensitive to the needs, preferences and priorities of our exemplar hard-to-reach groups and other stakeholders. Phase 2 (months 19–48): Testing Potential Solutions. We will develop interventions in three domains: improving access (e.g. outreach, transport, accessibility, internet use), improving understanding (e.g. knowledge and attitudes of care providers, public understanding) and improving care (e.g. skills training, psychological interventions, self management, role of voluntary sector). We will test interventions in terms of acceptability and credibility to service users and other stakeholders, their impact on the mental health of service users, and their service implications and cost effectiveness. Phase 3 (months 43–60): Putting what Works into Practice. We will develop and evaluate methods to disseminate our findings across the broad economy of primary health care, including a focused implementation experiment in two PCT sites.

Projected Outputs and Dissemination Plan: Outputs will include resource packs with associated educational interventions. Outcomes: Improved knowledge, skills and case identification by care providers; new services sensitised to needs of hard to reach groups, with evidence of better access and outcomes; increased R&D capability of NHS staff. Impact: Increased equity of access to high-quality primary care mental health services.

Expertise of Research Team: We bring together internationally recognised mental health researchers from Universities of Liverpool and Manchester, with policy makers and managers from two Primary Care Trusts. We have substantial expertise in the range of methodologies necessary for applied health research, in service and diagnostic issues, and in the design and prosecution of trials and related interventions for our exemplar hard to reach groups.

Justification of Costs: We will integrate existing resource from our academic and NHS institutions with the resource directly associated with this programme, to provide an efficient and cost-effective mechanism for delivering on our aims and objectives. In order to achieve our objectives we expect to appoint a programme coordinator, two senior research fellows, four research assistants and a senior technician. Statistical and health economic expertise are important but not central to our objectives, and will be provided as required through consultancy arrangements.

Management and Governance: Our programme will direct by two chief investigators, with support from a Programme Management Group (meeting 6–8 times a year), an Advisory Group (meeting twice yearly) and an international group of experts. All aspects of the programme involving direct contact with NHS service users or providers will be the subject of applications for aproval to relevant Research Ethics Committees. The applicants include two Trust clinical directors and the lead of a service user organisation.

Patient and Public Involvement: Our programme ensures primary care user involvement at all stages, and hence will build strong academic–community partnerships. This includes:

1. (a) consultation with service users and their supporters to enable effective service redesign;

2. (b) collaboration with a service user as co-applicant, and others to be deployed to deliver interventions; and

3. (c) control: service user led self help clinics may be tested during the programme.

Community engagement

About 30 participants will be recruited (i.e. about seven or eight in each locality). Purposive sampling to include CFG invitees in Longsight and Croxteth. In Picton/Granby and Wythenshawe, sample from those stakeholders who would have been invited to CFGs if we had organised them: include primary care and third sector; and local residents as well as non-residents.

Primary care

Aim for two key informants from each practice including controls. Key informants could be a GP, practice manager, staff who've experienced the training, e.g. receptionists and/or practice nurses.

Contact

• One initial assessment.

• Eight face-to-face sessions over 9 weeks.

Method of referral

W030 was referred to the AMP programme through her GP. She stated that she found it hard to ask for help and so waited until she felt that was the only option. She would not normally have spoken to her GP about mental well-being concerns as she felt that the GP did not provide enough time for her and it was not something that should be spoken about to a GP. Since being involved with the AMP programme, W030 has discussed her concerns with her GP and is currently awaiting contact from the psychologist.

Initial contact

W030 was easily contactable by telephone and was willing to engage.

On-going contact

W030 expressed a wish to be seen at home. This worked well and she never missed or cancelled a session. Later she stated that if she had had to come to an external location she would not have come to the appointments. This was for a number of reasons relating to anxiety about going out, finding it hard to keep appointments and getting motivated to go out. For the final session we agreed that she would come to an external location as part of her therapeutic process. She kept to this appointment and this was a big part of her therapeutic process.

I believe that the flexibility afforded to her in being able to see her in her home was crucial to W030 continuing with the therapy.

Social engagement

W030 showed no interest in group work or organised social activities at first. By the third session she was keen to join the AMP group and felt that it would be good for her. She did not attend because of anxiety about new group situations but has since made moves to attend local community groups and feels that this will help her continue the work that we did in the sessions. For the final one-to-one session she came to the resource centre and took away information about the activities there; she had also been out more and discovered some other activities in the community that she wanted to engage in such as a coffee morning and IT lessons.

Other

W030 was stunned at her own progress during the therapy. She stated that her relationship with her husband had improved. He also reported this to me after the final session.

W030 said that she had made an appointment to see the smoking cessation nurse as she felt that she ‘had the strength’ to give up smoking now. She also reported a decrease in obsessive–compulsive disorder-type behaviours, tearfulness and panic.

Contact

• One initial assessment.

• Nine face-to-face sessions over 11 weeks.

• Eight group sessions over 8 weeks.

Method of referral

W019 was referred to the AMP programme through her GP.

Initial contact

W019 was initially difficult to contact. At the third attempt to contact her by telephone I managed to get hold of her and she seemed reluctant to engage, putting obstacles in the way of possible appointments. She agreed to a date 2 weeks from that telephone call but then cancelled on the day and said that she would call me back the following week. After not hearing from her for 2 weeks I recontacted her and asked her if she was still interested and she agreed to a date the following week.

In total, from the first attempt at contact it took 8 weeks and eight contacts/attempts at contact to complete the initial assessment.

W019 stated that this difficulty was because she was anxious about the unknown regarding the sessions and so found it hard to commit to attending. She felt panicky and was visibly shaking and nervous at her first session. She also thought that perhaps others needed the service more than her and wasn't sure if she was really suitable (i.e. not ‘in need’ enough). However, she decided to attend as she did want to speak to someone about her ‘troubles’.

It was clear from the beginning that W019 would benefit from this service and by the final few sessions she stated as much. I believe that she would never have attended had she received only one/two telephone calls or a letter.

Ongoing contact

One of the reasons why W019 stated that she had cancelled the first appointment was because she didn't want to go to the area in which I was located because of unpleasant memories of her past associated with that area. Therefore, I arranged to see her at her GP surgery, which she felt more comfortable with, and she subsequently committed to the sessions and attended each one.

Note: the extra work from my perspective of booking a room at the GP surgery was significant as the room could not be booked in advance and, therefore, I was required to telephone each week (sometimes more than once as I had to speak to the practice manager). I got the general impression that the practice was uncomfortable with me using its rooms. However, when W019 spoke to her GP, the GP was pleased to have the service on site.

Because W019 had decided to attend the group at the resource centre she agreed that we would transfer our sessions there so that she could have them straight after the group. This worked well and she attended every group and one-to-one session. This worked so well in fact that she has gone on to be a volunteer there and feels very safe and comfortable. The resource centre has become a safe haven-like environment for her.

In total, she received four sessions at the GP surgery and six sessions at the resource centre.

Social engagement

W019 engaged well in the group and, although feeling sceptical to begin with, states that she has gained a lot from the group. She has signed up to be a volunteer and would like to continue using Wednesday mornings to come to the resource centre to receive the social activity and engagement that has boosted her confidence, self-esteem and self-worth. Although relatively active generally, she has realised the importance to her well-being of getting out and engaging socially and so places great importance on continuing with this. I believe this is a result of the group work as well as the one-to-one sessions with AMP.

1. Opening

2. Views on reasons for accessing the intervention

Examine case notes in advance to establish what particular intervention the participant was offered, and whether they accepted it.

Aim: Participants will have been randomised to either the control or the intervention arm. The latter involves a variety of possible options, such as group sessions, individual sessions or signposting. Explore their opinions about what they were offered.

Those randomised to the intervention arm:

I think you went to see [insert name of well-being facilitator] who did an assessment. How did you find that assessment? Was it easy or difficult to talk about your problem(s)? Did you feel that [insert name of well-being facilitator] was interested in you and understood your problem?

After your assessment/after you saw [insert name of person who did the baseline interview]:

Did he/she give you any ideas about help that might be available? What kind of things were you offered?

Did any of them put you off straight away? Why? Did you find anything particularly appealing? Why?

What was it that influenced your decision to choose what you did? Did the facilitator persuade you to try something that you might not have thought of?

Can you talk a bit about [insert name of well-being facilitator] the well-being facilitator? (How much contact did you have with them? What did you think about their approach? Did you get on with them? Were they friendly? Did you feel that they listened to you properly? And understood what you needed? Did you look forward to seeing them? Do you think seeing the facilitator made you feel better?)

3. Experiences of receiving the intervention

Aim: Encourage the participant to provide a descriptive overview of the programme that they received. Encourage the participant to explore their opinions about the strengths and weaknesses of the programme. Primary areas of questioning: format, delivery, time and location.

4. Views on outcomes

Aim: Investigate the perceived impacts the intervention has had on those who have taken part.

How do you think that taking part in the well-being programme has had an effect on you personally (positive/negative)? Has it helped you? In what ways?

Can you give some examples of this? (Try and relate questions on outcomes to the specific circumstances and well-being issues of the individual in question)

Did you enjoy the sessions? Did you look forward to them? Were they better than you anticipated?

Are you satisfied with what you were offered? Would you recommend it to other people?

Was there anything you didn't like or would change?

5. Closing

Aim: Summarise some of the participant's main thoughts for clarity.

Ask if there is anything else that hasn't been covered; particularly other suggestions for how the programme could be improved.

Explain the next stages of the programme and how the interview data will be used in relation to them.

Is there anything that you'd have liked me to ask about but I didn't?

Do you have any further suggestions for how the well-being programme could be improved in the future?