Explanation of context, mechanisms and outcomes in adult community mental health crisis care: the MH-CREST realist evidence synthesis

Identified need

There is a drive in the UK to improve experiences of crisis care and to design services and interventions that are effective, timely and accessible to all those in need,30 including services that are able to respond with urgency that equals that of responses to physical health emergencies. 29 It is therefore vital to develop complex interventions from a theoretical understanding of the mechanisms that produce the desired outcomes and in which contexts these work best. The Crisis Care Concordat27 identified a need for crisis care to be developed across multiple agencies, including statutory and voluntary sectors; for crisis care to be largely community based; and for improved implementation across the UK, to avoid crisis care being postcode dependent. Attention has been focused on service providers and settings, but there is currently a lack of evidence about the mechanisms that underpin effective mental health crisis care and how these are activated to resolve crises across a range of contexts. A focus on underpinning theoretical development would enable commissioners to invest in a range of services designed to include the mechanisms that produce the best outcomes across service designs and providers.

Health services are under ongoing financial pressures, and inpatient care is not only undesirable in many cases, but is expensive37 and scarce. 33 Community-based crisis care presents an opportunity for cost-effectiveness, provided that interventions can successfully be developed, tested and implemented to enable improved prediction of outcomes. 46 The development of effective community crisis care may also help to alleviate pressure in the urgent care pathway, particularly in A&E departments.

Previous work

Prior to the commencement of the current study, the review group’s consultations with people who have accessed community crisis care [patient and public involvement (PPI)] showed that these services resemble a tangled web of overlapping services with complex referral routes and blurred functions. PPI participants endorsed a focus on community crisis services on the grounds that they are generally preferred; they provide the respite, information and support that people ask for; and they avoid the need to be away from home and family. These experiences echoed the published evidence in respect of mixed experiences of CRTs, and lent weight to the need to improve understanding of how services work to resolve crises. This could improve people’s ability to access the right care at the right time. 12

Patient and public involvement and engagement

Prior to commencement of the study, members of the public, people with lived experience of crisis care, their carers and other stakeholders in statutory and voluntary sector services were consulted to help formulate the general aims and direction of the research. This was achieved through telephone consultations with key voluntary sector service managers, attendance at team meetings of CRTs and a focus group with people who had recently accessed crisis services for themselves or a family member. Visitors from two voluntary sector services involved in pre-protocol PPI consultation continued their involvement through membership of the ESG.

Ethics approval

Following favourable review by an NHS Research Ethics Committee, approval was granted on 8 January 2020 [Integrated Research Application System (IRAS) reference number: 261486; Research Ethics Committee reference number: 19/YH/0347]. Approvals were obtained for recruitment and consent of members to the ESG and individual interviews (see project web page: www.journalslibrary.nihr.ac.uk/hsdr/TWKK5110).

Study design and rationale

A four-phase realist evidence synthesis design was developed, comprising (1) identification of initial programme theories (IPTs); (2) iterative group and individual interviews to prioritise, test and refine programme theories; (3) focused reviews of prioritised IPT ; and (4) synthesis to mid-range theory. The focus of the realist synthesis was to develop programme theories to explain how different elements of crisis mental health care work provide appropriate and effective responses to mental health crises. Figure 1 summarises the study design.

FIGURE 1.

Summary of study design.

Application of realist methods to explore complex interventions

The current study conceptualised crisis resolution services as heterogeneous, complex interventions, as defined by the Medical Research Council. 46 Complex interventions activate multiple human responses that interact in non-linear ways to produce highly context-dependent outcomes. 47 Realist review offers an optimal approach for exploring how and why complex programmes involving human actions and decisions, such as crisis mental health services, may or may not work, and to inform the theoretical development of intervention(s). 46,47

Theory testing

Outcomes of interventions are causally activated by multiple context-sensitive mechanisms. They happen not only because of what is done in an intervention, but also because of how people respond. 47,51,53 The realist approach offers a participatory method of synthesis that allows for, and indeed capitalises on, continual testing and refinement of emergent programme theories against empirical evidence, data from policy documents and primary data (e.g. from engagement with stakeholders). 49 The involvement of consumers of health care is central to this type of research. 54,55

A theory-driven approach strengthens the potential to inform commissioning, service design and delivery that are sensitive to context across diverse service designs and providers. 56 From the perspective of patients and the public, the theory testing provides a platform for future empirical testing of service designs, thereby improving access and experience of crisis care. This research aims to inform the ongoing development and evaluation of existing mental health community crisis services and interventions.

Programme and mid-range theories

Realist synthesis offers a lens through which a research team can make sense of what is occurring within a complex intervention, particularly in understanding the circumstances in which it is more or less likely to be effective through the identification and testing of programme theories. 47,49,57

‘Programme theory’ is conceived as a fairly concrete set of ideas and observations that explain how different elements of a specific intervention (or programme) interact to produce the intervention outcomes. In contrast, mid-range (or middle-range) theories represent intermediaries between programme theories and ‘all-inclusive speculations comprising a master conceptual scheme’. 56,58 Mid-range theory considers the theoretical and practical issues simultaneously, thus applying theory to practice. Mid-range theories are therefore useful frameworks to guide development of interventions and make generalisations about their application.

In the context of realist synthesis, these ‘minor working hypotheses’ are labelled ‘programme theories’ or explanations for how the programme or intervention works. Mid-range theories are often useful ‘as frameworks for understanding a problem or as guides to develop specific interventions’. 56 According to Pawson et al.,49 mid-range theories can be a product of the realist synthesis process. However, the world of health, social and behavioural research is so densely populated with mid-range theories that it is common to find confirmation of programme theory findings in existing mid-range theory. Although it has not yet proved possible to harvest these theories in a systematic way, realist synthesis harnesses theories identified via the expertise and knowledge of the review team and its experts, through theories referenced in included studies and through a body of theorisation published alongside included studies.

Overview of methods

As realist research is essentially iterative,49 decisions about how to proceed are informed by emerging insights into how best to explore the research question. These methods do not readily fit with a linear narrative and so the following description of methods aims to provide a transparent and clear account of the study components, rather than a strict chronology.

A preliminary search of the literature sought to identify IPTs about how UK crisis interventions and crisis services work. Expert stakeholders were consulted to refine the IPTs and to agree research priorities and the scope for the review. These processes resulted in three prioritised IPTs and were the focus for three focused reviews. Consultation with the ESG and individual interviews facilitated further sense-making, theory testing and refinement. A final search was undertaken to synthesise the programme theory components with mid-range theory. Pen portraits were used to formulate exemplars of service user journeys through the system, to convey generative causal logic, to connect the theory components with real-world experiences of stakeholders and to explain how mental health community crisis interventions work. Figure 2 illustrates the study process, highlighting how the focus contracted and expanded throughout the review.

FIGURE 2.

The scope and focus across the life of the synthesis.

Changes to the study plan, including in the context of the COVID-19 pandemic

This research commenced in September 2019 and was largely conducted during the COVID-19 pandemic. The pandemic precipitated ongoing adjustments to the project milestones and timeline: delaying ethics approval, prolonging uncertainties about the feasibility of face-to-face meetings and interviews, limiting options for recruitment of participants, restricting access to practitioners and service users, challenging continued engagement of stakeholders and resulting in unexpected absences among the research team. Hence, these challenges were addressed in the context of national restrictions on accessing practice areas and delays in communication due to the reduced capacity of many organisations. For example, the ESG was established before the start of the government restrictions. They met face to face in February 2020. Subsequent delays, postponements or cancellations and a switch to video link may all have affected the ESG contribution.

The original plan to interview up to 50 experts in crisis care from across the UK was modified. After postponement of interview recruitment, it was planned to include 25 participants; finally, 19 participants were recruited to 20 interviews. In addition, issues related to gaining ethics approval during the COVID-19 pandemic restricted recruitment to three NHS health trusts and one NHS ambulance trust in England. The planned six pen portraits was reduced to three, linked to each of the three prioritised programme theories. The original plans to support the ESG members to produce a short animation to share the findings with a wide audience has not been achieved owing to delays earlier in the project delivery and funding limitations.

Focus of the review

The research team focused the review59 using the definitions of crisis stated in the Crisis Care Concordat,27 which highlights the architecture of crisis services, summarised as follows:

Access to support before crisis point . . .

Urgent and emergency access to crisis care . . .

Quality of treatment and care when in crisis . . .

Recovery and staying well . . ..

Crisis Care Concordat27

The review focused on ‘Urgent and emergency access to crisis care’ and ‘Quality of treatment and care when in crisis’, specifically on initial contact and access via the crisis aspect of services, thereby excluding home treatment and onward referral.

Expert stakeholder group engagement, recruitment and membership

Following ethics approval via IRAS/the Health Research Authority, expert stakeholders (n = 15) were recruited through NHS trusts, the voluntary sector and service user/carer networks. The ESG members included professional and user/carer expertise. Two research team members (JT and NC) maintained contact with two voluntary sector organisations following the pre-protocol PPI, which aided engagement and recruitment of those with lived experience to the ESG. Michael C Ashman led and co-ordinated the recruitment of all stakeholders and co-chaired the group with Jill Thompson.

The ESG membership comprised six people with lived experience of using crisis services, a carer, a peer support worker, a consultant psychiatrist, a mental health nurse, an NHS crisis service manager, a mental health social worker with approved mental health professional (AMHP) experience, a voluntary sector manager, a mental health services commissioner and a mental health policy specialist.

Membership of the ESG reflected diverse community crisis care services, recognising that no single individual or group holds all knowledge about crisis care. 60 All were based in England at the time of the consultation. Many participants described dual roles: the peer support worker had (of necessity) crisis service lived experience, but many others had acquired lived experience, either personally or through carer responsibilities. Membership of the ESG remained largely stable throughout the study, with the following exceptions: one service user dropped out, commissioner expertise was covered by two people, and a voluntary sector manager was replaced by their new manager. Table 1 charts the composition of the ESG.

Expert stakeholder group consultations were held in February, September and December 2020 and April 2021. The first ESG meeting was face to face and was recorded through photographs and audio-recordings; subsequent meetings were via recorded video link.

Search for initial programme theories

The search for IPTs sought to find data to develop IPTs related to the architecture of crisis interventions and crisis services in the UK context. 61 During August and September 2019, an information specialist conducted the search using Google Scholar (Google Inc., Mountain View, CA, USA) (see Appendix 3 and Report Supplementary Material 1). 47 As programme theories may be located in the title, abstract and sections other than the results,62 the search strategy was designed to identify programme theories by screening full texts.

Eligibility criteria

Expert stakeholder group involvement in agreeing on the focus and priorities

In the first meeting, held face to face in December 2019, ESG members were introduced to the project and methods, and were provided with information about taking part and the opportunity to ask questions before providing written consent. One lived experience participant declined involvement in the ESG, but requested to be invited to take part in an individual interview.

Expert stakeholder group members worked in small groups of mixed expertise to use their expertise to make sense of, debate and rank the importance of the 11 embryonic IPT areas (see Box 3). This was achieved using the Diamond-9 consensus and prioritisation approach, which encourages participants to work together, developing and exposing thinking. 64 A template used for the exercise is shown in Figure 4 and a photographic image of the ESG group deliberations shown in Figure 5. Each of the programme theory areas were printed onto individual cards along with blank cards to facilitate new ideas from the ESG deliberations. The stakeholders organised themselves into four groups so that they could prioritise, reject, or amalgamate any of the 11 embryonic programme theory areas using a card sort process. Each of the four groups included experts by lived and professional experience. Expert views and rankings were aggregated across the groups to generate overall agreement on refined and prioritised IPTs.

FIGURE 4.

Diamond-9 prioritisation template.

FIGURE 5.

Photograph of ESG members deliberating IPT using Diamond 9.

The results of the ESG discussions and Diamond-9 process showed that the most important priority for the ESG members was ‘urgency of response’. The pivotal role of urgency and easy access was highlighted by one ESG member:

. . . if that’s wrong, everything goes out of line because things aren’t happening at the right time with the right people.

The theories, ranked in descending order of importance following urgency of response, were compassionate care, access to crisis care and treatment, culturally relevant care and treatment, and involvement. Sustaining professionals/staff was identified as the least important.

The initial theories ‘service configuration’ and ‘mature multi-agency approach and joint commissioning’ were not prioritised by the ESG. However, after discussion between the research team and the ESG, the research team retained the ‘mature multi-agency approach and joint commissioning’, considering it integral to the architecture of crisis services and fundamental to the research focus. Through an iterative process of discussion and deliberation between the research team and ESG members, a reduced and refined set of five IPTs were agreed. These IPTs were developed from the data extracted and recorded in the Microsoft Excel spreadsheet (see Appendix 1). These are outlined as a narrative summary in Table 2.

Defining the scope for the focused review

Following the RAMESES guidance on defining and refining the study scope,47,53 the ESG and research team jointly discussed the scope of the review and the selection of IPTs for focused review. First, an overarching focus on the initial stages of a person in crisis seeking help and securing support and the quality of the support provided (key intervention points in the architecture of crisis services as identified in the Crisis Care Concordat27,28) was agreed. Although the term ‘multi-agency’ was initially adopted, it was replaced by ‘inter-agency’ in the course of the team discussions because of the perceived synergy between agencies that work together effectively in crisis care. The two terms can be interpreted as broadly synonymous. 65

Figure 6 illustrates the process used to refine the scope of the study and identify the IPTs for focused review: from 11 embryonic IPT areas (see Box 3) to five IPTs (see Table 2) and, finally, resulting in three IPTs retained for focused review. The deliberations between the ESG and the research team resulted in amalgamation of ‘urgency of response’ and ‘access to crisis care’ into IPT 1 (crisis services can be accessed urgently) as they were felt to be inherently linked. The ESG and research team also combined compassionate and therapeutic care into IPT 2: care in a crisis is compassionate and therapeutic. The third and final programme theory combined multi-agency working and joint commissioning, amalgamated to IPT 3: inter-agency working.

FIGURE 6.

Figurative illustration of the decision-making process to prioritise and refine the scope of the review.

IPT 4, culturally relevant care (although highly relevant), was acknowledged to require consideration beyond the scope of the current study. Our initial searches revealed little published evidence focused on culturally relevant crisis care at a level that was sufficiently granular to provide adequate explanation. It was decided that IPT 4 would benefit from a future primary realist evaluation to generate new explanatory evidence. Data pertaining to cultural relevance, when identified as relevant to causal explanations, have been included in IPTs 1–3. IPT 5, involvement, was felt to be cross-cutting and has been reported when causally linked to concepts in focused reviews for IPTs 1–3.

Pen portraits

At the fourth and final ESG meeting, a pen portrait method was used to support theory development. Pen portraits offer an analytic method to focus large volumes of qualitative data while maintaining richness. 66 In ESG meetings, brief vignettes (see example in Report Supplementary Material 2) were used to structure discussions. These vignettes enabled the ESG members to use their expertise to relate the evidence from the included documents to real-world experiences. Discussions took a realist perspective that sought to explain and identify causal links.

Pen portraits were developed once the focused reviews were more fully developed, and provided an illustration from a real-world perspective of the causal link between mechanism and outcomes. They were constructed using synthesised data to map components of crisis interventions structured around the Template for Intervention Description and Replication (TIDieR)67 (see Box 10 and Appendix 2), together with context, mechanisms and outcomes, and to inform future empirical testing of the programme theory. The use of vignettes and pen portraits facilitated discussions with ESG members throughout the research, enabling theory to be presented to stakeholders in an accessible way, and to test and refine the understanding of CIMO in crisis services.

Rather than a description of any individual experience, the pen portraits provide examples of people experiencing crises, their circumstances and the crisis responses. Three pen portraits were developed (i.e. one drawn primarily from each focused review) to illustrate real-world examples of the theories in action; they were subsequently refined following ESG consultation and through the process of synthesis across focused reviews and mid-range theory. The draft pen portraits were discussed between research team members to ensure that the stories effectively linked to IPTs 1–3 (see Table 2) and interventions described in TIDieR Lite (see Box 10 and Appendix 2) before presenting them to ESG members. At the fourth and final online meeting between the ESG and the research team, ESG members were asked to sense-check three draft pen portraits, by moving between three subgroups. Discussions were recorded and reviewed by members of the research team (NC, KB and LS) and used to refine the pen portraits.

The discussions told us that the pen portraits were recognisable to ESG members, who endorsed them as exemplars of good crisis care. Some found them somewhat idealistic and were not convinced that the system could respond as the pen portraits described. Discussion of the final pen portrait versions is included in the three focused review chapters: Chapter 3, IPT 1 pen portrait: urgent and accessible crisis services; Chapter 4, IPT 2 pen portrait: compassionate and therapeutic crisis services; and Chapter 5, IPT 3 pen portrait: community crisis agencies work together.

At this final meeting, ESG members were invited to complete an ESG evaluation; the template for feedback and a summary of the evaluation results are in Report Supplementary Material 2, with a blog providing a personal account of participating in the ESG.

Sources for empirical testing

Ten academic databases were searched: MEDLINE, EMBASE, Web of Science Core Collection (Science Citation Index, Social Sciences Citation Index, Conference Proceedings Citation Index – Science and Conference Proceedings Citation Index – Social Science & Humanities), Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Cumulative Index to Nursing and Allied Health Literature, PsycINFO, Applied Social Sciences Index and Abstracts (ASSIA), ProQuest Dissertations & Theses Abstracts and Indexes (A&I), and Health Management Information Consortium (HMIC).

Search strategies combined mental health keywords with keywords for crisis and keywords for each programme theory, together with a highly sensitive and published geographical UK filter,68 using the AND Boolean operator. The filter was applied to all database searches except ASSIA, ProQuest Dissertations & Thesis A&I and HMIC. All searches were restricted to English-language studies. The search results were managed using EndNote (X9 version 3.3) [Clarivate Analytics (formerly Thomson Reuters), Philadelphia, PA, USA] (see Appendix 3 for a summary of search strategies and see Report Supplementary Material 1 for the full search strategies).

Theory-orientated searches

Theory-orientated, focused review searches were conducted from February to March 2020, with supplementary searches conducted in July 2020. The intention was to sample, rather than to identify an exhaustive body of literature. 69 A Google Custom Search (Google Inc.) was created to facilitate supplementary searching of grey literature sources across 32 UK websites (January–April 2020), with particular attention paid to NHS England publications, the Royal College of Psychiatrists and the Centre for Mental Health (see Appendix 3 for a summary of grey literature searches).

Cluster searching

The search strategy was developed iteratively and revisited at predetermined milestones, using different permutations and additional concepts. 69 As each focused review progressed, the lead author considered how each document contributed to the aims and objectives of the realist synthesis and to theory development. When relevant, lead authors deployed ‘cluster searching’, an innovative purposive sampling approach, to identify ‘clusters’ of related publications. This approach added to the conceptual richness and contextual thickness of the studies identified through topic-based searching. 69 Sibling (i.e. outputs directly linked to a single study) and kinship (i.e. papers sharing a common contextual/conceptual legacy) reports were sought to add richness while preserving rigour and relevance. 69 The research team pursued citation networks, using Google Scholar and Web of Science. The iterative approach enabled the research team to search for data beyond the literature particular to community crisis services, but also to test veracity in the context of UK community crisis interventions. The theory was continually tested and refined through expert consultation and individual interviews. The ESG also recommended records, based on its expertise. Searching continued until sufficient data were found (‘theoretical saturation’) to conclude that the programme theory components were coherent and plausible. 53,70

Screening

Retained records included primary research, reports, policy documents and expert opinion. Using EndNote (X9), retrieved records were screened for relevance by three members of the research team (LS, KB and NC). Full texts of the selected records were obtained and screened (see Figures 7–9). A list of the retained records from the searches for initial theory is provided in the Appendix 4.

Appraisal of relevance and rigour

A modified realist appraisal tool (Dr Justin Jagosh, director of the Centre for the Advancement of Realist Evaluation and Synthesis, visiting professor at the University of the West of England, Bristol, and a research fellow at the University of Surrey, personal communication, 2021) was used to appraise relevance and rigour across all the retained records (see Report Supplementary Material 3). Relevance was determined to be less substantial, medium or high. A document was considered highly relevant when the framing of the research and the research questions closely matched the review questions. Rigour was assessed by quality appraisal of the study methods, using the Mixed Methods Appraisal Tool (MMAT)71 (see Report Supplementary Material 3), in all documents where a methodology was available for assessment (J Jagosh, personal communication). 53

Data extraction

An analytical framework was created by the research team; the data extracted included the publication type, origin, type of service or intervention, and influencing factors (i.e. theoretical perspective). When it was possible to make inferences, the data extracted were attributed to context, mechanism or outcome. Data suggesting explanatory links between context and mechanism, between mechanism and outcome or between context and outcome were also extracted. Relevant data from the retained records were coded against the analytical framework (i.e. deductive coding) or by identifying new codes (i.e. inductive coding). 47,49,53 Data were extracted into a Microsoft Excel spreadsheet (see Appendix 1) and coded in NVivo version 12 (QSR International, Warrington, UK).

Recruitment

Purposive strategies were used to identify mental health service users, carers and professionals from diverse organisations including the NHS, local authorities, universities and the voluntary sector. Expertise not represented in the research team or the ESG was prioritised. Representation was sought from urban and rural areas, as well as mental health trusts, ambulance trusts and the police force. Study information was distributed to potentially eligible participants, who could contact the research team if they wanted to participate.

Sample

Following revision of the size and scope of the sample frame, in view of the 2020 COVID-19 pandemic, study recruitment targeted an updated intended sample size of 25. The final sample comprised 20 interviews, with 19 expert interviewees (one participant took part in two interviews). Table 3 details the sample profile, along with the corresponding interviewee codes.

Strategy for realist interviews

A bank of semistructured interview questions, formulated in accordance with a realist interviewing approach,72 was agreed within the research team (see Report Supplementary Material 4). Interview participants were asked to comment on the evidence and on how thinking in the research team was developing. Participants were selected for interview based on their expertise in one or more programme theories. All interviewers had access to all interview questions and could opt for relevant questions as the interview progressed (see Report Supplementary Material 4).

Participants were invited to take part via Microsoft Teams (Microsoft Corporation), Zoom (Zoom Video Communications, San Jose, CA, USA)73 or telephone, and were provided with a participant information sheet and consent form in advance. Verbal consent was audio-recorded before the interview; interviews lasted up to 1 hour.

Four research team members conducted the interviews (NC, LS, KB and JT). Interviews were recorded and saved to a secure, password-locked server accessible by the research team. The interview recordings were reviewed, and detailed notes transcribed by each interviewer and shared with the research team. Following discussion between the research team members, a deductive coding process was used to connect interview data with the IPTs in the focused reviews. Interview data were used to test the veracity of the IPTs, particularly where there were gaps in the published evidence, but also to strengthen real-world understanding of the architecture of crisis services and causal links between context, mechanism and outcome. Excerpts from the interviews are reported throughout the focused review chapters in the form of discussion and direct quotations. After the completion of 20 interviews, it was agreed that there were sufficient data to inform the programme theories.

Developing a mid-range theory

A final search was undertaken to link the three programme theories with mid-range theory. 61 Mid-range theories were identified relating to inter-agency collaboration,75,76 feeling in control,77 breaking point,78 a strengths perspective,79 compassionate care80,81 and protection motivation theory. 82

Structure of review chapters

Chapters 3–5 report the three focused reviews to test and refine IPTs 1–3 (see Table 2): IPT 1, crisis services can be accessed urgently; IPT 2, care in a crisis is compassionate and therapeutic; and IPT 3, community crisis agencies work together.

Chapters 3–5 are structured to follow the realist CMO convention. The context is reported first, followed by outcomes and unintended consequences, and, finally, mechanisms and conclusions. The IPTs are explained from three circumstances: the person in crisis and their family, front-line staff and the crisis care system. Each chapter concludes with an ‘if – then – leading to’ statement and a pen portrait to illustrate the IPT from a real-world perspective. Synthesis across the findings from IPTs 1–3 and mid-range theory is discussed in Chapter 6.

Approach to identification of studies for review

Three iterative database searches were conducted: the first used search terms related to urgency and access, the second search used terms related to waiting and delay and the third focused on mental health triage. The results of these searches were combined. After exclusions on the grounds of relevance to IPT 1, urgent and accessible crisis care, 33 records were retained [13 from academic databases, five from grey literature (websites and reports) and 15 from hand-searching]. Backward citation searching of 18 records identified in database and grey literature searches produced a further 14 records, of which five were retained. In total, 38 records were included. The flow chart in Figure 7 shows the process of identification of studies for this review. 48

FIGURE 7.

Search results of the focused review: IPT 1, urgent and accessible crisis services.

Retained records

Details of the 38 retained records for the urgency- and accessibility-focused review are provided in Appendix 4.

Settings

The research settings reported included all community mental health services,90,99,102,106,108–117 community crisis services,2,10,15,83,85,96–98,100 integrated services with the police,78,89,92 integrated services with paramedics,95 emergency departments93,94,103–105 and telehealth. 84,87,88,101,107

Focus

Included documents were focused on joint crisis planning,110–117 crisis response models,78,83,85,94,102 gatekeeping and referral processes,97,99,107 triage and decision-making,87,88,98,101,103 waiting times,103,105,106 service user perspectives of crisis services,2,10,15,78,96,100,104,109 carers’ experiences of mental health services,90 co-location models,78,89,95 interventions in rural areas,108,109 telehealth,84,87,88,107–109 experiences of black and minority ethnic groups accessing crisis services100 and the voluntary sector contribution to crisis services. 10 Companion documents containing findings from the same study focused on joint crisis planning112–114,117 and on telephone triage. 87,8

Appraisal of relevance and rigour

All included documents were appraised for relevance and richness using a modified realist appraisal tool (J Jagosh, personal communication) (see Report Supplementary Material 3). Documents containing primary research data were also appraised for rigour using the MMAT71 (see Report Supplementary Material 3). The combined appraisal of relevance and rigour identified that the evidence supporting IPT 1, urgent and accessible crisis services, is based on studies of mixed quality.

One mixed-methods multisite study of the voluntary sector contribution to crisis care was highly matched to the theory component, providing a rich description of context, mechanism and outcome. 10 This study was methodologically rigorous and highly relevant, and was included as a key document.

Included documents reporting randomised controlled trials (RCTs) were all focused on joint crisis plans (JCPs). 110,111,116,117 These studies provided context and outcome but provided less substantial theoretical relevance. One study of JCPs reported the economic data drawn from the RCT, but the small sample size and lack of quality-of-life measures limited the rigour of these findings. 110 All included RCTs reported methodological problems, including small sample sizes110 due to under-recruitment;111 a lack of reliability in the measures;116 and inadequate staff training, causing intervention implementation problems. 117

The included mixed-methods evaluations reported small samples, mostly limited to a single site; lacked comparators; and did not consistently synthesise across methods, but provided less substantial95,115 to moderate89,92,93,98,102 framing of context, mechanism and outcome.

The included qualitative studies were theoretically rich, despite problems of rigour related to small samples that lack variance. 78,88,104,112,113 Included qualitative studies with less substantial theoretical relevance were rigorous. 15,87,90,96,97,99,114 One qualitative study provided context, but lacked rigour. 108

A scoping review85 and a realist review91 provided less substantial relevance, and one systematic review provided moderate relevance, to the theory. 94 A case study design provided less substantial relevance and rigour, but was retained because of its specific focus on triage decisions. 101

Expert reports were not assessed for rigour. Appraisal of these documents for relevance found that documents moderately framed to the theory provided context and mechanism. 2,103 Documents with less substantial theoretical relevance provided context. 83,100,105 Three documents were retained because they provided specific contextual relevance from policy or services, but were less substantially relevant. 84,106,107

Service user outcomes

Front-line staff outcomes

System outcomes

Service users

Front-line staff

System

Service users

Services that people perceive as providing a guaranteed response91,104 (resource) (e.g. urgent care services) engender trust,10,15,90,91,102 a sense of safety2,10,90 and a belief that the service can help102 (response). Complex referral routes, delays in responses and failure to secure crisis responses reinforce a sense that crisis care is not guaranteed, driving people to contact urgent care services, despite the likelihood of having to wait. People in crisis and their families have a reduced sense of urgency and can tolerate the discomfort of waiting for an intervention or follow-up91,104 (response) when information is shared with them about the nature of the follow-up and the timing of interventions104 (resource). People kept waiting without accurate information may become increasingly desperate for support and contact other services or leave without treatment. People are reassured and have a reduced sense of urgency (response) when crisis services require little effort to navigate and are available 24/7 (resource). 2,10,15,91,92 Access to crisis services with low financial and social burden, by having no cost91,95 (e.g. freephone,108 not requiring travel or fee-paid transport10,92) and with minimum disruption to family life through agreed timing and venue of visits10,92,109 (resource), encourages people to make contact sooner and choose mental health crisis services rather than urgent care services (response).

The person’s (or their family’s) perception of the urgency of the crisis is acknowledged,2,90,96,112,113 and front-line staff acknowledge that the person may have already used their personal resources or that their own resources (such as family) are not available (such as during the night),89,112 or are perceived as unavailable because they do not wish to be a burden to family members10,112 (resource). When these acknowledgements are made by front-line staff, there is mutual understanding, open communication,87,88,91,98 a sense of collaboration104 and respect112 (response). Support from family and friends in a crisis may also generate continuity not available in some crisis services. Safe and welcoming community spaces with peer support2,10 provide a proxy for family and friends, generating a sense of safety and belonging. 10 Consistency, continuity and relational safety are critical factors for people in crisis10,102 (resource), giving an increased sense of safety, control and a reduced sense of urgency,110,112–114 and motivating people to make contact with the service again in future crises2,10,104 (response).

When a mutual agreement about the support required is agreed alongside negotiation about the timing of any intervention (shared decision-making)112 (resource), there is motivation to engage, relational safety10 and a reduced sense of urgency,2,10,88,115 and people are less likely to feel patronised (response). Shared decision-making requires front-line staff to have positive attitudes,2,10,100 mental health knowledge, and interpersonal and therapeutic skills87,96 (resource) so that the person in crisis and their family can have a sense of safety and an ability to communicate their need,10 and so that trust can be developed2 (response). When there is trust in front-line staff, a therapeutic alliance is developed whereby immediate supportive interventions to reduce distress can be provided,89 such as active listening and supportive counselling10,87 (resource), the crisis experience is validated. The person feels believed, that their concerns have been taken seriously,2,10,83,109 that they have been prioritised,104 respected and able to engage in their care117 (response).

Front-line staff

When front-line staff have mental health knowledge appropriate to their role and have knowledge about the role of other agencies83 through access to training appropriate to the context in which they encounter people in crisis,87,98,117 they provide immediate interventions to reduce distress104 (resource). Staff invest in shared decision-making10,92,112,113 and are confident in their role89 (response). Confident staff are open, flexible and able to take account of the views and experiences of a person in crisis113,117 and their family,10,90 thereby empowering service users during the decision process. 117 They understand the importance of providing explanations to the person and their family, including proactively supporting people as they wait for follow-up87,98 (response). Referral, liaison and transitions to follow-up care (resource) are achieved through mutual understanding and negotiation between front-line staff,85,92 the person experiencing the crisis112,113,117 and their family/carer,90 enabling a shared commitment to the decisions made109,110 (response).

Front-line staff have access to support, supervision10,98 and consultation about decisions and debriefing10,87 (i.e. immediately available to them, by telephone, face to face98 or video call109) from peers10 and leaders10,98 (resource). Staff are less stressed; have improved emotional well-being;10,98 and believe that seeking advice and support is an accepted norm10 that creates a working environment with a sense of safety, tolerance and acceptance between staff98 (response). When responsibility is shared in high-risk situations (resource), staff are confident with decision-making, less inhibited by fear of blame98,117 and less paternalistic in their attitudes to service users (response).

Systems

Service thresholds, staff resources needed to manage workload15,108 and systems to deliver care across complex commissioning boundaries have been planned for (resource), allowing front-line staff to confidently liaise and refer people in crisis in a supportive and timely way108 (response). Morale98 and motivation enable staff to believe in the service and trust that services are available to respond urgently to referrals10,89 (response).

There is investment in a shared approach to maintaining up-to-date information across agencies, which is easily accessible to members of the public83,91 and includes all types of crisis service10,83,89 (resource). People in crisis, their families and front-line staff can make informed decisions about sources of support appropriate to individual circumstances102 (response).

If–then programme theory: urgent and accessible crisis services

Retained records

Details of the 18 retained records for the compassionate and therapeutic care focused review are in Appendix 4.

Settings

The research settings reported in included documents were as follows: all health care,127,134–136 adult social care,137 the NHS in England,138 hospitals,131,132 primary care,139 community mental health services,118,125,126,129,133,140 the ambulance service128 and the voluntary sector. 10,130

Focus

The included documents investigated lived experiences of crisis,130,137 fostering compassionate care through Schwartz Rounds,125 culture of care,127 crisis planning,118 care quality,128,138,139 care planning and co-ordination,118,129 compassion,126,134–136 the impact of waiting,132 risk management,133 therapeutic relationships131 and team functioning. 140

Appraisal of relevance and rigour

All included documents were appraised for relevance and richness using a modified realist appraisal tool72 (J Jagosh, personal communication) (see Report Supplementary Material 3). Documents containing primary research data were also appraised for rigour using the MMAT73 (see Report Supplementary Material 3). The combined appraisal of relevance and rigour identified that the evidence supporting IPT 2, compassionate and therapeutic care, is based on studies of mixed quality.

Three records had high relevance, in that they were highly matched to the theory and provided a rich description of mechanisms and context. 10,126,133 A high-quality mixed-methods study of the voluntary sector contribution to crisis services was methodologically rigorous. 10 A qualitative study provided rich accounts of lived experience from a small sample that was limited by a lack of ethnic diversity. 133 The expert report was strongly linked to theory related to practice, but reported no method. 126

Most documents were moderately framed to the theory, reporting different, but related, phenomena, and containing multiple relevant areas of interest. 125,127–129,134,136,138–140 Of these, four were mixed-methods or multimethod studies127,129,136,138 reporting rigorous methods, but limited by low response rates to a survey,129 the narrow focus of the research setting,127,138 a stated limitation of the interpretative nature of the study risking biases from the research team,138 and a lack of detail in the stated method and approach to sampling. 136 A realist evaluation125 provided rich data related to mechanism, but had limitations in linking these and any reported outcomes specifically to Schwartz Rounds within a complex clinical setting. Two qualitative studies provided in-depth rich data limited by small samples that lacked variation139,140 and data that were more than 5 years old. 140

There were four relevant articles in which the pertinent context, mechanisms and outcomes were less substantial;118,130–132 a further two less substantially framed documents were retained on the basis of their contribution to understanding risk137 and leadership. 135 Of these, a meta-synthesis lacked a focus on community settings,131 and two qualitative studies provided in-depth rich data from small samples. 118,130 The remaining documents were expert reports with no stated methods. 132,135,137

Service user, staff and system outcomes

Where organisations prioritise reducing coercive practices, described as ‘measures such as compulsory admission, restraint and forced medication’, people in crisis are supported to keep themselves safe, and risk reduces. 133 Front-line workers feel valued, engaged, respected, less isolated and less exposed when they are fully supported by their organisation. 125,139 When front-line workers feel supported, staff groups are more stable and staff turnover is reduced. 126 Hence, compassionate treatment, of both people in crisis and front-line staff, produces better service user and organisational outcomes. Furthermore, in interviews, front-line staff provided insights into how favourable outcomes may depend on how individual staff respond:

Compassionate response is completely dependent on the individual officer. Looking back on my own behaviour, I used to be quite by the book. I wouldn’t consider and reflect and alter my behaviour at the time, but I don’t recognise myself even looking back 5 years. I’ve changed so much and that’s moving from a response officer to a community behaviour, it really brought about a sea change.

KB1, police officer

Furthermore, the well-being of the staff has an impact on how able they are to provide compassionate responses:

My personal mental health experience shapes how I deal with people. I notice how much patience I have depends on how tired I am, how hungry I am, whether I’ve had a break . . .

KB2, paramedic

At times, the behaviour of the person and the escalating nature of some crises make staying compassionate difficult for staff. The ability of staff to retain compassion relies on staff members’ ability to draw on their personal resources and resilience:

Some people will be deliberately antagonistic and aggressive – people shouting, screaming and swearing, trying to push me out of the room. They are going to be threatened with police involvement; they are going to get a very different response than those sitting quietly. Even if aggression is a result of distress, I don’t want to be hit!

KB2, paramedic

Service users

The crisis care experienced by some people in crisis appears, in many ways, to lack compassion, being experienced as humiliating, invasive,; traumatising and, in some instances, reminiscent of previous abuse. 133 Service users have reported being domineered by front-line workers. 118 They describe feeling dehumanised and losing trust. 130 Loss of trust in services and workers may affect self-confidence, abilities, activities and thoughts, and lead to isolation and anxiety. 130 Inadequate responses and rejection from services has been reported to lead to increasingly desperate people:10

If they ever offered me any support at all, I might have trust in them [crisis services].

JT2, service user

NHS services, such as CRTs, working to the national model may not be able to offer continuity. A service user interviewee (JT1) reported that the lack of continuity was distressing and prevented them from establishing rapport with workers; this point was further explained by a mental health nurse:

A criticism we get is consistency of staff: when people come to the CRHT [Crisis Resolution and Home Treatment Team], they see different nurses each time. It is really hard for them to connect with people. We try to allocate the same worker to the same service user if we can. But with holidays and breaks in shifts, it’s really hard to do. They are having to tell the same story. However much we document things, they feel we ask the same questions, and, in many ways, we are, we ask about risk and medication.

LS6, mental health nurse

A lack of investment in suitable safe spaces for people in a mental health crisis130 is a challenge for the delivery of compassionate and therapeutic care. Service users may have no choice other than to attend the emergency department, where care and assessment may not routinely be trauma-informed:132

I have to be in a very bad place to present at A&E. I find it triggers me there. It is not a helpful place to be.

JT2, service user

If someone is stressed, anxious and suffering, to put them in a stressed and anxious environment is not going to help.

LS6, mental health nurse

Front-line staff

Poor leadership at the level of middle management has been identified as detrimental to delivering compassionate care. 129 Compassionate and therapeutic care delivery can be challenged by continuous tension between meeting the day-to-day demands of working in health care, developing and supporting teams, and embedding structures and policies. 140 Organisational policies may not support the workforce to make decisions. 138 Lines of accountability can be unclear within multidisciplinary mental health teams, with the result that it may be less apparent who is in charge, or unhelpful power dynamics may undermine operational and clinical decisions,140 leading to poor communications and disengagement:

It is hard to know whose opinion is the top one . . . so that can cause tensions.

LS1, operational manager

Workload pressures on clinical team leaders may limit their availability to team members. 136 This may be particularly noticeable when things go wrong, as identified by one interviewee:

Everyone’s a bit unsure where to go when there have been incidents.

LS7, psychiatrist

Inherent conflicts within the leadership role may paralyse the decision-making of clinical team leaders. This can lead to a tendency to resist innovation and constructive criticism; to label staff who raise concerns as disruptive or complainers; and to resort to blaming any service problems on either external circumstances or front-line staff, thus absolving themselves of responsibility. 135 In response to these tensions, clinical leaders may also display power and superiority,140 and engage in bullying to drive change. 134 Ultimately, these processes may diminish the capacity of a clinical team to deliver compassionate care. 134,140

Management of risk

In some areas, systemic failings are driving skills deficits at the front line,138 especially if there are limited opportunities for professional development. 129 A lack of therapeutic skills and the inability to provide interventions may result in an excessive focus on the mitigation of risk129 and the prioritisation of medicines and short-term physical safety. 133 If front-line workers fear making the wrong decision and being blamed for the consequences of that decision,137 anxiety may drive behaviours in individuals and teams that are not conducive to providing high-quality compassionate care. 126 In the crisis context, a lack of confidence may paralyse decision-making. 128 The effect may be that decisions around risk assessments do not involve the people directly affected by the risk:137

We have adopted a more litigious approach to society . . . so for anything neglectful or for omissions, there is a high degree of fear in organisations. It makes the formal relationship between the organisation and the service user cautious and risk averse.

LS3, emergency medical technician

Front-line workers may justify taking coercive actions in challenging situations, on grounds of risk. 128,131 Concern about paramedics taking people to hospital by force following self-harm, irrespective of motives or severity, has been highlighted in the light of the known likelihood that such action can escalate risk, particularly following self-harm. 128 An interviewee reflected on the value of interpersonal skills for risk management:

Understanding, rapport and de-escalation are soft skills that are important, but aren’t taught or given the time to be appreciated or reflected on in busy working environments.

LS7, psychiatrist

Coercive practices undermine the development of trusting relationships. 133 The lack of a trusting relationship may generate fear in staff around risk and decision-making,137 thereby increasing levels of coercion. Front-line workers may also respond to risk with complacency, for example ‘He’s not going to do it because he’s never done it yet’ (NC3, carer). Alternatively, they may engage in ‘positive risk-taking’:

Positive risk-taking can be kind of a little bit of a buzzword that we use when we’re talking about, you know, justifying why we’re not responding to something.

LS2, mental health nurse

Systems

There is potential for tensions between financial and performance management agendas and the delivery of compassionate and psychologically safe crisis care,138 because the systems for regulation, inspection and commissioning across the health-care system may directly undermine compassion in organisations. 136 The issue appears to relate to health quality improvement strategies that involve comparative indicators and external audits. It has been observed that evaluation strategies are underpinned by an assumption that public sector organisations cannot be trusted to manage performance, leading to an imposition of standardised quality assurance approaches across organisations. 125 Furthermore, strategies to drive out bad practice based on what can be measured and regulated may actually restrict the ability of staff to meet service user need:125,134

Sometimes you feel like you are governed by performance reporting and the things the commissioners want. You would hope that is about service user outcomes, but sometimes it feels like it does restrict you as opposed to free you . . ..

LS1, operational manager

The current institutional and regulatory environment requires public sector health-care organisations to achieve targets, measures and standards; to seek financial incentives; and to report compliance to multiple overlapping authorities. The obligation to supply metrics in different formats to multiple external organisations is costly and distracting and may lead to fragmented knowledge, competing and ambiguous priorities, diffusion of responsibility and a lack of innovation. 138 Some service user priorities, such as compassion and psychological safety, are difficult to measure,125,136 and may not be reflected in outcome data that focus on quantifiable aspects of care. This suggests that there may be better ways to approach service evaluation:

What we are not doing is using the data intelligently to improve outcomes.

KB4, researcher

Voluntary sector systems have been described as relatively compassionate and humane, compared with the public sector,10 although similar organisational dynamics appear to be affecting their potential for delivery of compassionate care. Numerous and diverse funding arrangements and relationships between the public and voluntary sectors are contingent on complex factors such as organisational ethos, mission and strategic objectives. 10,130

A key informant study in this review10 investigated the contribution of voluntary sector organisations to mental health crisis care and highlighted the complexity of commissioning decisions concerning voluntary sector mental health crisis care, particularly in the context of austerity and cuts to services. The authors concluded that the quality of services in the voluntary sector can be inhibited, similarly to the public sector, when commissioners operate a business model;10 however, in other ways, business models can be beneficial to voluntary organisations:

There are large ones [voluntary sector organisations] that are now run like businesses and well-off, others that are local and grassroots and struggle.

KB4, researcher

Prioritising finance and performance management over people appears to lead to a lack of investment in the following: peer support services; physically and psychologically safe and calm alternatives to admission (e.g. crisis houses, safe spaces and walk-in services);10 and safe, short-term options for people who are intoxicated or distressed. 132,133

An NHS England report136 identified a risk that organisational leaders feed reports and metrics to system leaders without any genuine oversight and interpretation of the data, thereby reducing the potential for outcome data to inform better service delivery. For example, the interpersonal and relational aspects of care may be less visible (and hence of lower priority) to organisational leaders because they are not readily captured, measured and reported:139

I don’t think there’s much connection between, sort of, ground-level staff and the directorate any more.

LS2, mental health nurse

Therefore, arguably, if primacy is given to achieving financial objectives and to rationalisation, the main unintended consequence of prioritising measurable organisational ‘business’ outcomes is that health-care delivery may become unduly performance and task orientated. 125,127 Employees may be required to operate as units of production,126 with little regard for the relational or compassionate aspects of care. 125 The Royal College of Psychiatrists has argued that hierarchical organisational structures may be unhelpful, especially if the organisation is not clinically led. 126 In an interview, a psychiatrist conveyed the frustrations of having to work with the system:

Your MDT [multidisciplinary team] time, your team meetings, governance meetings, those are important. If you don’t get the support, it can’t happen. They just look at numbers and go ‘you reached that target’.

LS7, psychiatrist

Managers are equally likely to be frustrated by performance target cultures, as explained by one interviewee:

There are things you might want to do more work on, but . . . it doesn’t fit your current performance target.

LS1, operational manager

An extensive comparative mixed-methods study of mental health teams across six NHS sites in England and Wales129 found a discernible gap between national policy aspirations for recovery-focused mental health care, and practice. The authors raised concerns about a lack of vision among leadership teams and the impact on front-line services as a consequence of budget cuts, under-resourcing and bureaucracy. 129 They highlighted a need to ensure that organisational aims and operations were consistent with supporting staff and service users. 129 The ESG discussions identified that organisational priorities determined ‘how well staff are looked after’ and ‘how well staff know what is expected of them’, that is were focused on staff rather than service users. 64

Although the systems, processes, practices and disciplines within organisations are reported to determine the boundaries of action, when the organisational structure is fragmented, or a product of isolated decisions and workarounds, there is potential for toxic cultures and unhelpful norms to emerge. 136 Increasing demands create conditions for defensive and reactive responses from organisational leaders, leading to quality improvement efforts that may not use robust methods, or findings that are not subject to scrutiny. There may be ‘magical thinking’138 and indiscriminate use of particular approaches. One interviewee described investing time in a quality improvement project only for ideas to be ‘rubbished’ (LS2, mental health nurse) by organisational leaders, providing an example of how a perceived lack of support risks discouraging front-line workers from engagement in quality improvement initiatives.

Service users

Service users have the opportunity to form supportive trusting relationships130 and retain control118 and their safety is balanced with independence in a collaborative way. 133,137 A consistent and stable staff group enables early signs of deterioration to be noticed, and service users trust the opinion of front-line workers who raise concerns133 (resource). Trust in front-line workers provides a sense of belonging to service users in crisis;130 front-line workers provide consistency and service users feel that front-line staff are interested in the psychosocial context of their distress and prioritise emotional safety, healing and relationships;133 service users are not turned away if they identify that they need help;137 treatments are individually tailored;118 there is freedom, control and privacy; and service users can engage in meaningful social and occupational activities and maintain social roles and relationships that are important to their well-being during a mental health crisis (response).

Staff

Systems

If–then programme theory

Retained new records: inter-agency working

Details of the 26 new retained records for the focused review on inter-agency working are in Appendix 4.

Settings

The research settings reported in the retained new records (n = 26) were as follows: inter-agency crisis services,27,28,30,89,155 and specific inter-agency collaborations, including between mental health services and the police81,98,141,146,148,150–152 and between mental health services and paramedics. 142 Other settings included voluntary sector services,10,156 integrated mental health services,147,157 integrated health services,144,149,153,154 commissioning health services145,158 and crisis mental health services. 143

Focus

The retained records reported on models of integrated service design,81,98,144,151–153,156,157 policy guidance,27,28,30,145 good-practice guidelines,89,147,154,155 service user experiences,146 voluntary sector contribution,10 staff experiences,149,150 decision-making,148 commissioning,158 inter-agency self-harm interventions142 and NHS access to urgent crisis help. 143

Appraisal of relevance and rigour

All included documents were appraised for relevance and richness using a modified realist appraisal tool72 (J Jagosh, personal communication) (see Report Supplementary Material 3). For this review, records were scrutinised for relevance to inter-agency working in mental health crisis services. Documents containing primary research data were also appraised for rigour using the MMAT73 (see Report Supplementary Material 3). The combined appraisal of relevance and rigour identified that the evidence supporting IPT 3, inter-agency working, is based on studies of mixed quality.

Two documents were highly matched to the theory component: they provide rich descriptions of context, mechanism and outcome and were key documents focusing on voluntary sector crisis services10 and inter-agency crisis care between police and health services. 98 Both studies reported rigorous mixed-methods studies limited only by being at the level of description10 and reporting inconsistencies in analysis related to limitations in clinical data recording. 98

One mixed-methods study evaluated the impact of implementation of the Crisis Care Concordat and was rich in context and outcome, providing moderate relevance to the theory component, although limited by evaluating data from 12 months of implementation. 28 Two further mixed-methods studies were less substantially relevant to the theory component, providing mostly context. 148,153 An included survey provided less substantial relevance and was limited by a small sample size. 149

Included qualitative studies provided rich descriptions of context related to integrated police and health services, with some reference to outcomes, but very little mechanism, and were moderately framed to the theory. 150,152 Both studies were limited by a lack of lived-experience perspective,150,152 and one study describes a model of inter-agency crisis care not available in the UK. 152 A further qualitative study was less substantially relevant to the theory, providing mostly context, with limited mechanism or outcome. This study reported a very small sample, with resulting lack of variation. 146 Two literature reviews, a scoping review81 and a meta-synthesis,142 were less substantially relevant, providing only context.

Expert reports containing rich descriptions of context and mechanism were moderately framed to the theory. 89,144,147 Expert reports less substantially relevant were limited to providing inter-agency context. 141,151,154,156–158 Included policy documents were moderately framed when they contained specific context related to inter-agency crisis services27,30 and were less substantially relevant when they were limited to local context155 or had a context wider than mental health. 145 Expert reports and policy documents were not appraised for rigour as they reported limited or no information on method.

Inter-agency working

Inter-agency working

Service users

When people recognise crisis services as being relevant for them (response) by having been developed with involvement from their community106 (resource), people have a sense of connectedness to the service2,10 (response). Inter-agency working that is planned and delivered with involvement with those who access services (resource) generates mutual trust and shared ownership89,147 (response). Inter-agency services that demonstrate culturally sensitive values106 by being staffed by people who represent the diversity within the community they serve, have had training to improve cultural awareness and use language and communication that avoids stigma and racial stereotyping89,106 (resource) promote a sense of belonging and shared values, which, in turn, lead to reduced fear or shame about accessing the service118 (response). When people are less afraid (or stigmatised), they take an active role in their own recovery,2,10 engendering a sense of ownership and pride that enables them to take an active role in their community, for example through volunteering or providing peer support, and challenging negative attitudes10,64 (response).

When an inter-agency system of crisis services provides user-focused interventions including peer support, befriending, advocacy and less clinical services, such as those delivered in the voluntary sector (resource), then people in crisis recognise services as being ‘for them’, providing a sense of safety that fosters engagement10,89,145,155 (response). Within a complex inter-agency system when individuals are provided with a ‘single trusted point of liaison’154 (resource), they feel confident to navigate the system and are able to build trust that provides the sense of continuity and relational safety that people seek10,147 (response).

Staff

When staff have attended training suited to their specific role in providing support in a mental health crisis, staff have increased knowledge, improved skills and increased confidence in providing crisis care, and there is more interpersonal contact between the person and staff member (resource). This can lead to practice that is less risk averse and restrictive. 98,147 The person in crisis is more likely to feel supported and trust the service to keep them safe by providing care that people value, staff are happier and have more confidence in their crisis response (response).

Clarity about how the pieces of the system fit together help to reduce concerns about, for example, a perceived shift from a more psychosocial focus to a biomedical and risk-focused approach15 (resource). Role clarity, coupled with system knowledge, within inter-agency services (resource) supports staff to hold a more balanced perspective between legitimate concerns about the nature and timing of responses and an informed acceptance that the spectrum of care is available across the system. This enables front-line staff to understand where their role fits, allowing each part of the system to ‘play to its strengths’ (response). 10 Clarity about how their role fits within the wider service enhances front-line staff confidence that they are working within an accepted role and enhances their belief that their contribution is valued, reducing suspicion and cynicism about other front-line staff roles10 and enabling improved communication161 (response).

Inter-agency working needs stable internal team-working, facilitated by leaders who drive integration147 (resource) to enable different agencies to operate collaboratively (response). When team-working is stable within and across agencies (resource), front-line staff are able to create a positive team culture that is open to ‘being challenged’10 (response). Openness and challenge within and across teams provide opportunities for professional and user debate that facilitates scrutiny and advocacy, (response), especially where this involves voluntary sector teams10 (resource).

Increasing interpersonal contact and rapport between staff in different agencies by facilitating gatherings at which people share experiences and perspectives,28,150 including through joint training10,94,98,149 and job shadowing27 (resource), fosters mutual understanding and respect for differences in values and approaches to crisis care10,15,153 (response). Mutual understanding achieved through interpersonal contact (resource) can help to facilitate the development of shared goals and understanding of each agency’s role, and the overall ‘landscape of demand’28 on all parties149,150 (response).

Learning together (resource) creates new lines of communication and a sense of collaboration and a shared language that can overcome communication difficulties caused by jargon98,104,149 (response). This means that, in future contact, they might know the person they are liaising with; this goodwill on both sides can make the process go more smoothly, with agreements being able to be made and negotiation over who will provide what (response). Leaders and staff within each agency have new perspectives (resource) through which they develop improved understanding of how demands on their services differ (response); through sharing (resource), new ways of responding could be fostered28 (response), and warm handovers provide opportunities for improved information-sharing between agencies64,161 (resource).

The presence of colleagues with different expertise in the same office or car, or by rapid remote access over video calls or telephone (co-location and co-response), provides instant access to specialist advice (resource) and can help the spread of good practice30,98 (response). Co-location and co-response models (resource) promote collaborative decision-making between staff in different agencies (response) by drawing on a range of expertise, including being able to draw on each other’s established knowledge of, and links to, services. Being able to share expertise in these ways helps to support decision-making regarding the appropriateness of onward referral (resource). The social contact of sharing a car, or office, responding to calls together and getting to know each other promotes professional trust and respect for the other’s expertise (response).

Systems

Agreements about shared procedures between agencies, including assessment, prioritisation processes and information-sharing that is supported by integrated IT systems147,150 (resource), enable staff to ‘buy into’ the system10,94,147 (response). Shared agreements such as these provide clear boundaries for staff (resource), thereby allowing them to overcome anxieties regarding rules about data protection and confidentiality, improving confidence when communicating between agencies and improving motivation to do so from all involved96 (response). Shared assessment processes (resource), streamlined communication and improved decision-making98 between agencies ensure that the assessment gathers all the information needed by each agency, so that each agency can quickly access the information it requires and reduce the number of times people in crisis have to repeat themselves64 (response).

When commissioners and leaders across agencies work to co-produce and co-design crisis care informed by local population and geographic data,64,158 in collaboration with representation from all stakeholders89 (resource), different values and definitions of crises are accommodated, and problem-solving is conducted across the system (response). When evaluation is carried out using methods that incorporate inter-agency data (resource), commissioning is informed by a whole-system approach153,161 (response). By taking a whole-system approach to commissioning, organisations, particularly smaller voluntary sector organisations, can have a stable presence10 and there is system-wide understanding of the resources required. A system-wide understanding helps to avoid the detrimental impact of agencies operating in silos and being focused on protecting resources, thereby improving the ability of the whole system to maintain the flow of people through the complex pathway. 147 There are links between national policy expectations and co-ordination of inter-agency crisis service commissioning and local delivery27,30,89,152 (resource). There is legitimacy for inter-agency quality improvement, evaluation and research direction for system leaders10,28 (response).

If/then/leading to

Summary of interventions

Box 10 provides a summary of the interventions identified through our research using the TIDieR Lite framework. 67 A table showing the full detail of the TIDieR Lite interventions is in Appendix 2. Contextually important intervention strategies operating within the interventions identified in the TIDieR Lite were synthesised to three levels of intervention, namely system, organisational and service levels, and linked to each of the three programme theories in Table 6.

How does urgency and accessibility of services determine what happens during a mental health crisis?

Although urgency (linked to perceived need) and accessibility (linked to a response that meets that need) are conceptually different, they are very closely related in the minds of service users. Service users adopt strategies to negotiate access to mental health care when emotions become overwhelming and harmful behaviours become likely. 184 These strategies may include contact with police services when health services fall short of the response times that users themselves expect. Urgency may increase when there are delays in accessing services. Services that are seen as accessible are less likely to be accessed prematurely or on a ‘just-in-case’ basis. Dissatisfaction with access to crisis care in England has driven a national service improvement programme and the introduction of the Crisis Care Concordat27 in 2014. New models aimed at offering improved access are discussed elsewhere within this report and include ‘crisis cafes’ (‘safe havens’ or ‘recovery cafes’), walk-in services, stand-alone community crisis assessment teams and 24-hour crisis lines. Inconsistency of coverage remains an important issue, with some people having improved access to numerous alternative forms of community crisis care, whereas others have access to the same CRTs that have been standard in the UK for the past 20 years.

However, when services are designed to be accessible, they may be accessed for convenience at the expense of those who most need them at that specific point in time. Conversely, access to regular health services, for example to GPs, by mental health service users may be limited, out of respect for appointment requests by other patients. 162 Queuing theory suggests that services that are accessible will be increasingly accessed until they stop being accessible, after which usage will drop until the service is used little enough to once more offer a timely response. In reality, service users want a just-in-time, not a just-in-case, response. However, a further complication in planning the accessibility of services is the fact that needs that are not addressed early may become exacerbated and require a more extensive, prolonged or resource-intensive response. Managing these processes is challenging because it not only requires a knowledge of the needs of an individual service user, but it also requires an overview situation report of the current usage and availability of the service as a whole.

With new crisis care models continuing to emerge, with the potential to improve access and service user satisfaction, it remains important to establish the evidence base for these models and, in line with the complexity approach assumed by this review, to explore the implications of these new models within the broader local crisis care system. 185

What is compassionate care during a mental health crisis and why is it important?

This review has confirmed that compassionate care is an important feature of a crisis management response. Kindness and compassion are highly valued by people in crisis:

A kind, compassionate, caring, and effective response followed by the right support can transform lives for the better.

(Lynch and Persaud, Lived Experience Commentary) Dalton-Locke et al. 185

Yet these attributes are often lacking in mental health crisis care. 12 It is worth noting that the Francis Inquiry181 put much store on remedial documentation and communication when users reported a basic ‘lack of compassion’;186 crisis mental health services would do well to avoid a similar pitfall. In particular, police services may resort to coercive powers in transporting service users to a location where they can access mental health services. This initial contact may serve to ‘frame’ the service user’s subsequent perceptions of crisis mental health services.

According to the theoretical model associated with compassion, compassion includes offering a welcoming environment. This link between programme theories 1 (urgent and accessible services) and 2 (compassionate leadership) suggests that the identity of an organisation as compassionate can be perceived from early contact and contributes positively to impressions of accessibility. Similarly, recovery alliance theory, which requires professional and service user rapport, can be seen as a facilitator of access to staff, and requires a compassionate environment. Person-centred care approaches strongly resonate with our analysis (e.g. recovery alliance theory;167 Goffman’s165 sociological theory, forms of talk;165 and compassion in psychotherapy87).

Perhaps what is unique from these findings is the realisation that compassion is not simply an attribute of individual staff with whom service users come into contact. Organisation leaders hold the potential to influence the culture of their organisation: to make it more compassionate. Routes to achieving culture change lie within social innovation theory, whereby organisation leaders consciously strive to innovate: they do this within structures that facilitate change; they are open to cross-fertilisation of ideas, including from other agencies; and they drive forward the movement towards change.

How does inter-agency working influence the service response to people experiencing a mental health crisis?

Our findings suggest that a fragmented and disjointed cross-agency response may serve to exacerbate the challenges of the first response. 187 This can operate at a practical level in terms of delays in referral2 or in having to repeat one’s history multiple times to different agencies. When police officers contribute to a mental health response, service users experience agencies working in isolation and being hastily referred from one agency to the next. 2 Equally, a disjointed response may be perceived to symbolise that the concerns of the service user or their significant others are not important or are not being taken seriously. 2 The transfer of mental health care into emergency health services or police custody has seen increasing interest in how services work together to offer help-seeking pathways to support people in crisis with mental health disorders. 149

Two approaches to help-seeking pathways have been distinguished: the ‘contingency’ approach, which describes and correlates service use with clinical and sociodemographic profiles of patients, and the ‘process-oriented’ view, which focuses on social and interpersonal processes that affect help-seeking behaviour in the community and in the health system. 188 The same research has shown that client pathways to mental health services are divergent and can be a product of choice, coercion or simply ‘muddling through’; the ‘muddling through’ approaches conflict with the demand for urgent and accessible service responses. 189

Evidence from novel models of police collaboration, such as crisis intervention teams and street triage,84,190 suggests that collaborative models between police and health-care professionals may achieve some success in improving service user experience.

Deviations from the protocol

The original plan to interview up to 50 experts in crisis care from across the UK was modified: after postponement of interview recruitment, we planned to include 25 participants, and recruited a total of 19 participants for 20 interviews. In addition, issues related to gaining ethics approval during the COVID-19 pandemic restricted recruitment to three NHS health trusts and one NHS ambulance trust in England. The planned six pen portraits were reduced to three linked to each of the three prioritised programme theories. The original plans to support the ESG members to produce a short animation to share the findings with a wide audience has not been achieved owing to delays earlier in the project delivery and funding limitations.

Our plans to engage with UK-wide stakeholders was limited by the impact of the COVID-19 pandemic. Difficulties obtaining ethics approvals delayed recruitment to interviews and a pragmatic decision was taken to recruit participants via three NHS trusts and one ambulance trust in England. Attempts to recruit lived experience participants from across the UK were also hampered by the pandemic. We made a pragmatic decision to continue to work with the stakeholders already recruited to the ESG to provide a stable and familiar online environment for discussion.

Inter-agency working

This study identified several important causal links between inter-agency models of crisis care and optimal outcomes. These outcomes are at times theoretical and have been subjected to limited testing in primary research. There are a number of different models of inter-agency working operating in UK crisis services, providing an opportunity for mixed-methods case study approaches to evaluate these models to facilitate implementation of models that produce optimal outcomes in different contexts. An important, but neglected, area of focus for this research should be on efficacy of models for rural populations.

Inter-agency models involving police and other ‘blue-light’ responders are rapidly evolving. Crisis interventions involving police and mental health services have a growing body of evidence; however, there is a lack of evidence for co-response models involving ambulance paramedic staff or emergency control rooms.

Technology

Information-sharing and communication are important to facilitate inter-agency working, involvement and shared decision-making. A lack of effective and shared technology was identified as a barrier to implementation of these important intervention strategies.

Stakeholders talked about the use of mobile apps and web pages to support self-care and self-management during a crisis, but no evidence was located on their use or efficacy. Research evaluating implementation of technology such as this would be a useful addition, particularly to support people as they wait for call-backs, visits or referrals.

The use of telephone and video-call technology has increased during the COVID-19 pandemic for initial contact and triage. Although some evidence was included in the review to show that important mechanisms such as relational safety and trust can be established using telehealth approaches in crisis care, there is a lack of UK evaluations of crisis lines, SPA, telephone triage and video-call assessment.

Meaningful stakeholder involvement

Meaningful co-production with service users and other expert stakeholders enhances the relevance of research and should be incorporated routinely into research design and delivery.

Marginalised populations accessing crisis services

Initial searches did not identify data at a granular enough level to allow exploration of the experiences of black and minority ethnic people and their experiences of accessing crisis services within the scope of this study. Despite this, it has been possible to identify some important contexts and mechanisms that improve the outcomes of crisis care for these populations, including shared decision-making, open access and non-clinical safe spaces. A realist evaluation is needed to explore in more depth the factors influencing access to and transition through crisis care for black and minority ethnic populations.

Although psychiatric or any other diagnoses have not been used to contextualise data in this study, some of the included evidence pointed to some diagnoses, or health conditions, generating additional barriers to access, for example substance use or alcohol use problems, personality disorders, physical health conditions and autistic spectrum disorders. There is a need to design research to identify effective crisis interventions for people marginalised by these diagnostic complexities.

Evaluating individual-level outcomes

It was notable in the conduct of this review that outcomes from crisis care are largely expressed in service outcomes such as reduction in hospital admissions, use of sections of the Mental Health Act 1983118 and 2007,119 and attendances at A&E. There is a lack of focus on individual or recovery outcomes. This review highlights the importance of mechanisms such as psychological and relational safety, compassion and trust in producing optimal crisis outcomes. Research is needed to develop evaluation approaches to measure the presence and impact of these mechanisms in crisis care.

Mental health triage

Mental health triage appears to be a promising approach, but has a limited evidence base. Future research could explore and test the implementation and effect of mental health triage systems. There is a paucity of research exploring the link between decisions in triage and access to appropriate crisis responses across an inter-agency system and how this is experienced by people in crisis and their family or friends. Taking an inter-agency approach to such research may uncover important insights into differences in values driving decisions and support closing of gaps and reduction of delays between services.

Inter-agency working

Inter-agency working is very important to the delivery of optimal community mental health crisis care. Policy-makers might consider strengthening national leadership of inter-agency crisis care, such as previously provided by the Crisis Care Concordat,27 to provide a framework for the development of agreed standards; to drive improvement; and, importantly, to provide a platform where expectations for inter-agency working can be shared. Within a national policy framework, support for local decisions about commissioning of community crisis services ensures that the needs of local populations are served, cognisant of local demographic characteristics, population size and dispersion. Further research could help improve understanding of how the geographic location of crisis services affects service delivery. There is, for example, limited evidence focused on inter-agency models of crisis care in rural UK communities.

Within an inter-agency crisis care system, cross-agency agreements about how services operate across traditional service boundaries, including information-sharing agreements, trigger mechanisms such as affiliation and shared goals, which, the evidence suggests, improve experiences of receiving crisis care. Advances in digital and telehealth technology used in crisis care, fast-tracked as a result of the COVID-19 pandemic,90 provide additional flexibility needed to improve cross-boundary information-sharing. Evidence focused on the use of technology to support delivery of crisis care was, however, sparse, and often not based on UK service designs. Future research could focus on understanding if telehealth is an acceptable intervention strategy and how these technologies contribute to producing optimal crisis outcomes. Although current technology provides a means for intervention and information-sharing, our review found that information was not shared evenly across the crisis system and that limitations with the technology itself led to frustrations among front-line staff. We also located very little in the evidence about how information-sharing supports families (or friends) during a crisis, with or without technology, despite availability of guidance for carer support in England,188 and the knowledge that families take a great deal of responsibility for providing support during crises.

Policy-makers and researchers might consider how inter-agency working can trigger distal outcomes such as stigma reduction, increased likelihood of co-production and reduction in restrictive responses. Co-production triggers mechanisms such as community engagement. When communities are engaged in local crisis services, people recognise the crisis service as ‘designed for them’ and this may reduce stigma and fear. When there is less fear, especially in marginalised and minority communities, the evidence suggests that people access crisis care sooner. When people access crisis care sooner, they may be less distressed at first contact, and this could be a way for services to reduce the likelihood of using restrictive approaches to care.

Co-production provides opportunities to design services that can accommodate the diverse ways in which people define mental health crises, including the different values and expectations people have about crisis responses. The evidence suggests that, to accommodate this diversity within any inter-agency crisis system, service commissioners might consider designing inter-agency crisis services that include services perceived as ‘clinical’ (including, e.g., CRT and A&E liaison), services perceived as ‘non-clinical’ (e.g. safe havens, crisis cafés) and ‘non-health’ services (such as interventions delivered by the police).

These findings suggest that inter-agency services are more likely to produce optimal outcomes when they are designed in ways that facilitate co-operation across service boundaries, such as between statutory and voluntary sectors, primary and secondary care, urgent care and crisis care, health and social care, justice systems and local authorities. The evidence suggests that, when crisis services prioritise opportunities for joint endeavour across agencies (including shared meeting structures, shared service evaluation, joint training), mutual understanding between leaders and staff in different agencies may improve. At the front line of service delivery, parallel assessment provides an opportunity for joint working, but also improves the accuracy of assessment and reduces the likelihood of traumatic retelling, known to risk further harm to those in crisis. Our findings conclude that joint working in these ways provides a platform for delivery of inter-agency crisis care that may reduce the stress, distress, delays and service gaps caused by boundary and responsibility disputes.

Although this synthesis identified important mechanisms triggered by interpersonal contact between staff in different agencies, mid-range theory identified that contact alone may not be sufficient to generate co-operative inter-agency working. What is also needed is equal status across different providers, common goals and external support such as through national policy co-ordination. Without these, there is a risk of reinforcing difference and of siloed thinking, leading to lasting damage to inter-agency working. 82,169,170 This is not to suggest that inter-agency working requires services to operate with the same values, but rather the approach to inter-agency collaboration may be required to accommodate different values to optimise outcomes. This synthesis also identified the importance of retaining the unique identity of services (particularly the voluntary sector), thereby facilitating challenge and debate that may protect the crisis care system from unhelpful attitudinal barriers to service delivery.

Stability and continuity

We concluded from evidence that commissioners of inter-agency crisis care are faced with the challenge of, on one hand, enabling voluntary sector providers (for example) to retain their unique identity, and therefore their ability to challenge attitudes, and, on the other hand, enabling these services to have a stable presence in the crisis care landscape. Policy-makes and commissioners might therefore consider novel ways to sustainably fund voluntary sector services while also enabling them to retain their values and independence. The stable presence of different crisis services across an inter-agency system may also enable the system-level mechanism of continuity through stable thresholds for access to services, which enable people in crisis to navigate to timely help.

Continuity was identified in the evidence as an important mechanism for optimal crisis care and was often conceptualised in the synthesis in terms of individual front-line workers who are known, trusted and co-ordinate service responses. Although care co-ordination is an existing feature of UK mental health services, and provides a degree of continuity, further consideration of how care co-ordination operates to support people to navigate crisis services, especially for those new to mental health services, may help to provide the sense of continuity people in crisis seek.

Although we located limited evidence focused on the role of family and friends during a mental health crisis, it was clear from the synthesis that, when family and friends are present and available for support, people in crisis often seek their help first; as a result, continuity may be best provided by them. When family support is not available, our review identified that peer support in non-clinical services may be conceptualised as providing a proxy for family support, further emphasising the importance of non-clinical crisis services having a stable presence in the crisis care system.

Compassion

The focused review of compassionate and therapeutic crisis care identified the importance of compassion throughout the crisis care system. The mechanism activated by compassionate leadership was that leaders set the tone for service delivery, influencing the culture of services and, as a result, how staff operate at the front line. If compassionate leadership supports front-line staff to retain their compassion, it may logically lead to improved experience for those accessing crisis care. The causal link between compassionate leaders and improved service outcomes is somewhat theoretical, in part because compassion is not easily measured. Future evaluation of compassionate leadership might focus on the support needs of leaders, as well as those of front-line staff. Furthermore, evaluation of crisis services more generally may be most impactful when conducted across the inter-agency crisis system, as people in crisis often encounter multiple agencies, thus providing a more complete picture of crisis services.

Decision-making, referral and transition

The evidence focused on accessing crisis care showed that first impressions of crisis services have a lasting impact on how people perceive crisis services (stigma and fear), and seem to then affect how individuals and families choose their route to help in future crises. These attitudinal barriers may be maintained by services having complex referral systems; coupled with service boundaries and thresholds that lack transparency, these seem to lead to more people circling the system (multiple assessments and no intervention) or accessing help via the urgent care pathway. Designing services that guarantee that a person will access a service (such as A&E or some voluntary sector crisis services) increases the likelihood of people in crisis prioritising that service for first contact.

Continued emphasis on commissioning of community crisis services with streamlined access routes, transparent thresholds for access and cross-boundary agreements may facilitate earlier access to crisis support for anyone in crisis, but are very important to people from minority ethnic groups, for whom stigma and fear present the greatest barriers to access. Although this review located limited evidence focused on crisis care for people who are black or from minority ethnic backgrounds in the UK, the use of joint crisis planning to facilitate shared decision-making showed more promising results in trials for black and minority ethnic populations than for other ethnic groups, and may be an important consideration for future intervention development and evaluation.

Difficulties accessing crisis care appropriate to perceived need was linked to inconsistent application of decision processes between services, often resulting in disputes about responsibility. Decision tools used in crisis care lack robust evidence regarding their measurement efficacy; this is an area recommended for future research. Support from leaders who operate close to the front line of crisis service delivery was an important mechanism to improve decision-making and confidence of front-line staff, enabling them to feel more in control of workload and retain their compassion. Commissioning of crisis services might consider development of clinical leadership roles and design these roles in ways that assure their availability to provide support to front-line crisis staff.

Google Scholar

Summary Google Scholar search strings (full version in Report Supplementary Material 1)

1. ((“logic model” OR “theory of change” OR “theory of action” OR “outcomes chain” OR “program* theory” OR “program* logic” OR “logical framework*”) AND (“mental health crisis”) AND (communit*)

2. to 30. ((AND (“24 hour*”); AND (“crisis house*”); AND (“helpline*”); AND (“crisis teams”); AND (“day service” or “day treatment”); AND (“police”); AND (“street triage”); AND (“home”); AND (“cafe*”); AND (“drop in”); AND (crisis or crises or acute or emergenc*); AND (“crisis line”); AND (GP); AND (“Accident and Emergency”); AND (“NHS 111”); AND (“NHS direct”); AND “hospital”; AND “liaison psychiatry”; AND “on-call”; AND “social service*”; AND “community mental health*”; AND “crisis team*”; AND “day service*”; AND “street triage”; AND “crisis resolution team*”; AND “crisis team*”; AND triage; AND crisis AND triage; AND crisis AND decision*

Academic databases

• MEDLINE (Ovid, 1946 to 2020)

• EMBASE (Ovid, 1974 to 2020)

  • Web of Science Core Collection (Clarivate Analytics)

  • Science Citation Index Expanded (1900–2020)

  • Social Science Citation Index (1956–2020)

  • Conference Proceedings Citation Index – Science (1990–2020)

  • Conference Proceedings Citation Index – Social Science & Humanities (1990–2020)

  • Cochrane Library

  • Cochrane Database of Systematic Reviews (1996–2020)

  • Cochrane Central Register of Controlled Trials (1898–2020)

• Cumulative Index to Nursing and Allied Health Literature (EBSCO, 1974–2018)

• PsycINFO (Ovid, 1806–2018)

• ASSIA (ProQuest, 1987–2018)

• ProQuest Dissertations & Theses A&I (ProQuest, 1743 to 2020)

• HMIC (NICE Healthcare Databases Advanced Search, 1979–2020).

Database search limits

• A highly sensitive filter (Ayiku et al. 68) was applied to all database searches except ASSIA, ProQuest Dissertations & Thesis A&I and HMIC.

• All searches were restricted to English-language studies.

• 2000 to present.

Grey literature searches

• Anxiety UK (www.anxietyuk.org.uk/get-help/crisis-support/).

• Breathing Space (Scotland), 0800 83 85 87 (https://breathingspace.scot/)).

• C.A.L.L. Helpline (Wales), 0800 132 737 (www.callhelpline.org.uk/).

• Centre for Mental Health (www.centreformentalhealth.org.uk).

• College of Paramedics (www.collegeofparamedics.co.uk/).

• Cornerstone Counselling Service (www.ccscounselling.org.uk/).

• Exclusive Secure Care Services (www.securecareservices.co.uk/).

• Inspire Wellbeing (www.inspirewellbeing.org/).

• Lifeline Helpline (Northern Ireland), 0808 808 8000 (www.lifelinehelpline.info/).

• Live Well Leeds (https://livewellleeds.org.uk/crisis-support/).

• Mental Health At Work (www.mentalhealthatwork.org.uk).

• Mental Health Crisis Care Concordat (www.crisiscareconcordat.org.uk).

• Mental health crisis helplines in the UK (2018) (www.youthemployment.org.uk/list-of-mental-health-crisis-helplines-in-the-uk/).

• Mental Health Foundation(www.mentalhealth.org.uk/).

• Mental Health Resource (www.mentalhealthresource.org.uk/).

• Mental Health Today (www.mentalhealthtoday.co.uk/).

• Mind (www.mind.org.uk).

• New Hope (www.newhope.org.uk/mental-health).

• NHS England publications (www.england.nhs.uk/publication/).

• Navigo (www.navigocare.co.uk/).

• Papyrus Prevention of Young Suicide HOPEline UK, 0800 068 41 41 (https://papyrus-uk.org/hopelineuk/).

• Porthedon (https://portheden.org/).

• Protect Life 2 – Suicide Prevention Strategy (www.health-ni.gov.uk/protectlife2).

• Royal College of Psychiatrists (www.rcpsych.ac.uk/).

• Samaritans, 116 123 (www.samaritans.org/).

• SANE (www.sane.org.uk/).

• Shout (www.giveusashout.org/).

• Stress Project (www.stressproject.org.uk/).

• Sunflowers Suicide Support (www.sunflowerssuicidesupport.org.uk/).

• Touchstone (www.touchstonesupport.org.uk/).

• Turning Point Crisis Support (www.turning-point.co.uk/services/mental-health/crisis-support.html).

• Young Minds (https://youngminds.org.uk/).

Search terms: MEDLINE (full database searches and supplementary searches in Report Supplementary Material 1)