Interventions to manage use of the emergency and urgent care system by people from vulnerable groups: a mapping review

Article history

The research reported here is the product of an HS&DR Evidence Synthesis Centre, contracted to provide rapid evidence syntheses on issues of relevance to the health service, and to inform future HS&DR calls for new research around identified gaps in evidence. Other reviews by the Evidence Synthesis Centres are also available in the HS&DR journal.

The research reported in this issue of the journal was funded by the HS&DR programme or one of its preceding programmes as project number 16/47/17. The contractual start date was in January 2018. The final report began editorial review in January 2019 and was accepted for publication in June 2019. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HS&DR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Andrew Booth is a member of the National Institute for Health Research Complex Reviews Support Unit Funding Board.

Copyright statement

© Queen’s Printer and Controller of HMSO 2019. This work was produced by Booth et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2019 Queen’s Printer and Controller of HMSO

Objectives

The aim of this three-stage review was to identify and map, using predefined population groups, interventions that have been developed and delivered to individuals or groups to manage their use of EUC services. These interventions may either reduce demand or ensure that populations use the appropriate EUC as needed. An additional aim was to classify these interventions, where the evidence permitted, in terms of intervention characteristics and to report headline messages of the outcomes of these interventions. Further detail on the content and delivery of these interventions was to be extracted using a Template for Intervention Description and Replication (TIDieR7). It was anticipated that the review, which also aimed to look at current UK initiatives, would help us to understand what can be learnt about delivering interventions to vulnerable service users and also to identify potential gaps within the research and practice agendas, with a view to stimulating future research and evaluation.

Research questions

The research questions for this review were as follows:

• What interventions exist to manage use of EUC by people from vulnerable groups?

• What are the characteristics of these interventions?

• Is there evidence of service delivery outcomes (for patients and the health service) resulting from these interventions?

An initial model to illustrate the complexity of associations and outcomes in this research area was developed by the team from the review of reviews by Coster et al. 4 The original review of reviews had identified six themes that accounted for most factors related to ED attendance and urgent care usage: access to and confidence in primary care; perceived urgency and anxiety and the value of reassurance from emergency-based services; views of family, friends or health-care professionals; convenience (in terms of location, not having to make appointments and opening hours); individual patient factors; and perceived need for emergency medical services (EMSs) or hospital care, treatment or investigations. This initial model, which helped us to conceptualise the review, is illustrated in Figure 1. Vulnerable patients had been identified as one factor within individual patient factors, linked but not invariably associated with deprivation. The team took these factors at a disaggregated level, transforming them from the original linear framework into a conceptual model.

FIGURE 1.

Diagram illustrating the complex relationships underpinning use of emergency care.

The overall aims of the project were as follows:

• Identify and map, using predefined population groups, interventions that have been developed and delivered to individuals or groups to manage their use of EUC services. These interventions may either reduce demand or ensure that populations use the appropriate EUC as needed.

• When the evidence permits, classify these interventions in terms of intervention characteristics.

• When the evidence permits, report headline messages of the outcomes of these interventions.

• When interventions include outcomes data (evaluative), assess how the content and delivery of these interventions has been reported using the TIDieR framework. 7

• When interventions do not report outcomes data (descriptive), report the content and delivery of the intervention using an abbreviated version of the TIDieR framework. 7

• Understand what can be learnt about delivering interventions to vulnerable service users.

• Identify potential gaps within the research and practice agendas with a view to stimulating future research and evaluation.

Overview

The initial phase of the project entailed carrying out a systematic mapping review, conducted in accordance with published methods. 8 This phase was followed by an intervention analysis examining the content and delivery of each intervention using the TIDieR framework. 7 The intervention analysis aimed to offer a broad overview of intervention content to facilitate further knowledge synthesis or primary research while summarising the current state of the evidence base through analysing the interventions reported. During the third and final phase of the work, we undertook a review of current UK initiatives (whether or not they have been evaluated) that had the aim of managing demand. The three phases of the project are summarised in Figure 2.

FIGURE 2.

Overview of the phases of the review.

Table 1 summarises how the evidence was funnelled through the review, outlining the actions taken during each phase and the main outcomes of the work and including comments on issues and challenges encountered during each phase.

Phase 1 mapping review methods

Methods for phases 2 and 3

Phases 2 and 3 of the project were iterative and involved two value-added synthesis processes: first, an intervention analysis to examine the components of each intervention, and, second, an initiative analysis to identify interventions in practice within a UK NHS setting. As candidate interventions from the initiative analysis were identified, the team retraced its steps to the evidence base, as captured in the EndNote database for the project and through targeted searches of the MEDLINE database, to augment the list of candidate interventions.

The intervention analysis and initiative analysis sought to add both intervention detail and UK context-sensitive breadth to the mapping review of published literature. These analyses aimed to extend the field of inquiry to include the wider whole system [e.g. not just the ED and ambulance care but also including, but not limited to, walk-in centres, urgent general practitioner (GP) access, same-day community-based nursing and telephone helplines, e.g. NHS 111] of EUC beyond the limited lens offered by the inclusion criteria of the mapping review.

Involvement of stakeholders

A patient and public involvement group provides ongoing advice to projects that are being carried out at the Sheffield HSDR Evidence Synthesis Centre. The group has eight members, with national representation. The second meeting of the group, held in February 2018, included discussion of this project, with public members reporting that research on EUC services is of particular interest to them, particularly given the high-profile nature of these services in the media. Input from the public advisors highlighted the need to avoid conveying a negative perception of vulnerable people during the review, as vulnerability could result from circumstances over which people have no control. The advisors perceived that vulnerable groups were generally under-researched and reported how public understanding may be unclear regarding how vulnerability may lead to discrimination and poorer health outcomes.

The group cautioned that it would be important to think about the impact of austerity on these populations during the review, both in terms of a general decrease in wealth in society and the differential impact that this might have on these groups and also in terms of the cuts to services, particularly those upstream that might have an impact on increased use of the health-care system. Input from the group highlighted the need to look for examples of interventions that had failed, as well as those that had been successful. The advisors also mentioned that the criminal justice system, in particular the police, often took a role in health and social services and that it might be worth looking at interventions delivered by the police to these groups in emergency settings. Notes from this meeting are available (see Report Supplementary Material 1).

In addition to our discussions with our patient and public involvement panel, we also sought advice from clinical academics within the School of Health and Related Research (ScHARR), University of Sheffield, who are involved in the delivery of care within the ED to ensure that our understanding of vulnerability made sense and that we were conceptualising the scope of EUC in an understandable manner.

Overview

The results from the three phases of the work are outlined separately in sequence in the following sections. The main findings and recommendations are combined in Chapter 4.

Studies included in the mapping review

The database search identified 18,381 records (MEDLINE Epub Ahead of Print, MEDLINE In-Process & Other Non-Indexed Citations, MEDLINE Daily and MEDLINE – 6210; CINAHL – 3647; and Web of Science and Conference Proceedings Citation Indexes – 8524). Following deduplication, a total of 11,890 records remained for screening. Figure 3 provides a summary of the process of study selection. A total of 169 references were initially screened as ‘include’. On further scrutiny, 141 of these were excluded. The most common reason for exclusion related to the study outcomes reported as clinical outcomes for patients with no health service-demand-related outcomes reported. Any paper that included an outcome that could be related to the use of or demand for UEC was included. The full texts of 28 papers were screened, with 16 included in the mapping review. 13–28 Full details of these papers are included in Report Supplementary Material 1.

FIGURE 3.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram illustrating the process of mapping review study selection.

Characteristics of studies included in the mapping review

The primary study papers that met the inclusion criteria all used an evaluative design. There were four randomised controlled trials (RCTs),14,22,26,28 three non-randomised trials19,23,24 and three cohort studies. 15,17,21 The majority of the literature was from North America, with three studies from the UK,13,15,21 a single study from Australia28 and a single study from Switzerland. 14

Synthesis of mapping review findings

Table 4 summarises the data extraction for the mapping review. None of the study populations was described in terms of being ‘vulnerable’. Included studies referred to participants as being frequent users or frequent callers,14,15,17,20–27 young people with alcohol-related presentations,28 people with high opioid use,22 people at high risk of drug-seeking behaviour16 or people on a low income. 18

The interventions evaluated in the primary studies included provision of a link worker for young people with an alcohol-related presentation, combined with a brief intervention and referral,28 and introduction of a system for people on a low income or who were uninsured whereby hospital EDs were linked to local primary care providers and ED staff referred patients to these clinics18 (both subsequently classified under ‘outreach services and teams’). Interventions also included a care plan initiated by ED staff for patients at high risk of drug-seeking behaviour. 16 Case management, which could include offering additional social support to patients, was described in four studies. 14,15,24,26 The difference between interventions described as case management and those referred to as care plans was not always clearly distinguishable, as establishing a plan was often a main component of management.

A care plan intervention was reported to have resulted in a significant reduction in mean annual ED visits among the 53 patients in the study group, from 7.6 visits [95% confidence interval (CI) 6.3 to 9.1 visits] to 2.3 visits (95% CI 1.5 to 3.1 visits; p ≤ 0.0001) at 1 year of follow-up and 1.5 visits (95% CI 0.9 to 2.1 visits; p ≤ 0.0001) at 2 years’ follow-up. 16

The link worker intervention was evaluated in a RCT. 28 The authors reported no difference at 10 years’ follow-up in the cost of ED presentations between the intervention group and the control group (AU$4266 vs. AU$4150; p = 0.916). There was also no significant difference in the number of ED events between groups (p = 0.849). However, ED attendances and rates of presentation specifically with an alcohol or other dependency mental health diagnosis were significantly reduced in the intervention group compared with the control group (0.03 vs. 0.25, p = 0.010; 0.03 vs. 0.25, p = 0.010, respectively).

The linking system intervention18 similarly appeared to have resulted in no reduction in ED visits overall, although the authors reported that there was a reduction for those with chronic or behavioural conditions or more frequent users (however, data for this finding were not provided in the paper).

A RCT evaluating case management for frequent attenders who had mental health or addiction difficulties found a 14% reduction in frequency of ED visits during the post-randomisation process; however, in line with the studies described above, this reduction did not reach statistical significance. 26 A pilot study of frequent ambulance callers suggested that case management was promising as a method to reduce the median call volume, although the authors highlighted the complex range of needs of patients. 15

Of the 16 included studies, there was overall uncertainty regarding the effectiveness of interventions aiming to reduce UEC use, with five studies indicating a reduction in service use16,17,20,21,23 and six studies suggesting no significant reduction in service use. 18,22,24,26–28 There appeared to be no clear pattern regarding which types of interventions reported positive outcomes, although studies indicating effectiveness tended to be of a non-comparative design.

The mapping review identified the extremely limited evidence regarding interventions and outcomes for people who might be classified as being ‘vulnerable’. The paucity of empirical evidence suggested that a further full systematic review of interventions for vulnerable patients was not indicated. The review identified the lack of clarity regarding intervention types and components, with a potentially useful direction for further work being to unravel elements of the interventions evaluated. It was anticipated that further exploration of the detail of interventions aimed at reducing UEC usage in patients more generally might assist in understanding which interventions may be promising for vulnerable groups. Given the limited available evidence in academic journals, it was also anticipated that grey literature and other sources of reporting may be a worthwhile avenue of exploration.

Developing a typology: the intervention and initiative analyses

The phase 2 intervention search focused on further exploring the characteristics of existing interventions aiming to reduce the use of UEC services. Following the iterative process described in Chapter 2, the review team developed a typology of nine different intervention types (Table 5). In addition to the literature found during the mapping review, the phase 2 intervention analysis identified four additional articles: three were related to multicomponent packages, including case management, case finding, care planning and outreach for the homeless and those with substance use problems,29–31 and one reported a programme to prevent less serious, and therefore potentially inappropriate, cases from presenting at an ED. 32

To supplement the evidence from the initial mapping review and phase 2 intervention search, in phase 3 of the study we sought further evidence from the grey literature and other sources beyond academic journals. In phase 3, we identified 22 further documents33–54 relating to 15 UK initiatives. Data from these documents were extracted against an abbreviated version of the TIDieR framework and examined to further develop and refine our understanding of the intervention characteristics and components.

Primary sources of these initiatives were good-practice case books and rapid reviews, commissioning plans and news articles, either in the GP press (e.g. Pulse) or in local newspapers. It was not always possible to determine the extent to which initiatives were truly innovative and the extent to which they had been developed elsewhere and then implemented in a new location. The method by which each type of intervention was identified is indicated in Table 5.

The following sections provides a summary description of each intervention in our typology, drawing on information from the three phases of the review. Completed TIDieRs for the interventions are available in Report Supplementary Material 1. For illustrative purposes only, a sample abbreviated TIDieR for targeted case management is provided in Table 6.

Further details of the sources by which the initiatives were identified are available in Appendix 2. Table 7 summarises the key details of each intervention.

What interventions were identified?

As previously mentioned for the mapping review, it is extremely challenging to identify interventions that are targeted specifically at the vulnerable groups nominated for this review. Interventions may not specifically identify vulnerability as the feature to be addressed by their programme theory. Alternatively, interventions may target the population generally, with the implication that vulnerable populations may benefit equally or even may benefit more than the general population. In some cases, we had to work backwards from interventions employed by initiatives introduced by specific health communities to target vulnerable populations (e.g. urgent care clinics) to look at the associated evidence for that intervention.

The shortlist of nine intervention types identified from the multiple routes used for this review, namely the mapping review, the intervention analysis and the survey of initiatives, comprises:

1. care navigators

2. care planning

3. case finding

4. case management

5. front of A&E general practice/front-door streaming model

6. migrant support programme

7. outreach services and teams

8. rapid access doctor/paramedic/urgent visiting services

9. urgent care clinics.

A previous narrative review of reviews56 categorised interventions to reduce frequent attendance at EDs into six types:

1. cost sharing

2. strengthening primary care

3. pre-hospital diversion (including telephone triage)

4. co-ordination

5. education and self-management support

6. barriers to access EDs.

The intervention typology developed during our review adds to this previous work by providing discrete types of interventions outlined in the literature, rather than categorising general approaches. The diversity of intervention types that we identified indicates different conceptualisations of the exact nature of the problem of frequent attendance and, consequently, a differential ability for these interventions to benefit vulnerable groups. Interventions relating to cost sharing or co-ordination may result in more efficient and more appropriate utilisation of care but are unlikely to benefit vulnerable populations directly unless specifically targeted at these groups (e.g. case management of the homeless).

More promising, particularly for vulnerable populations for whom physical access is an issue, are those interventions that challenge difficulties relating to the proximity of services, such as strengthening primary care (e.g. urgent care clinics) or pre-hospital diversion (e.g. telephone services, including triage). It should be noted that stereotypical representations of vulnerable populations may impede identification of potential interventions, for example a systematic review of access to health services by homeless people using technology found high rates of mobile phone and technology ownership. 57 Education and self-management support can prove challenging to several of the identified vulnerable groups, for example those with low levels of health literacy, those with a first language other than English and those with an itinerant lifestyle. The final category of interventions that address barriers to access overlaps partially with those interventions that address physical access or navigation barriers, as mentioned above, but, more widely, may address systemic difficulties, such as the requirement to provide proof of residence or the unsympathetic or unwelcoming attitudes of ED staff. 58

Why are these groups vulnerable/at higher risk?

The heterogeneous grouping of vulnerable patients specified for this review recognises that vulnerability presents in many forms. Indeed, one common form of vulnerability – frailty among older and chronically unwell patients – was specifically excluded from this review, having previously been separately targeted by the evidence synthesis programme. 59 In trying to conceptualise vulnerability at a level superordinate to the list of included groups, we identify the following ‘vulnerabilities’:

• vulnerability in terms of a limited ability to navigate the health system (either through language challenges or through limited health literacy)

• vulnerability in terms of constrained physical access to current health provision (e.g. rural communities, those likely to encounter transportation difficulties)

• vulnerability in terms of difficulties in demonstrating eligibility for current health services (e.g. migrants and the homeless)

• vulnerability in terms of having a higher likelihood of complex psychological and social issues not adequately or appropriately addressed by health services and requiring multiagency involvement (e.g. long-term unemployed, those in a non-stable housing situation, those experiencing substance use disorders)

• vulnerability in terms of requiring a higher degree of social or emotional support (a proportion of those from among the long-term employed, those experiencing substance use disorders, those from ethnic minorities and migrants).

These different vulnerabilities are addressed by different mechanisms present within the different interventions; for example, issues around physical access may be addressed by the provision of ‘more-local’ urgent care centres or an urgent visiting service. Conversely, an intervention may activate multiple mechanisms; for example, the particular type of case finding activated by the paramedic from Blackpool42 offers elements of navigation, outreach and social and emotional support. Finally, how an initiative is set up may differentially address different vulnerabilities; for example, a rapid access doctor accessed through participating general practices may result in the persistence and extension of an existing barrier of limited eligibility (based on practice registration and having a permanent address), whereas the same service accessed as an extension of the ambulance service would be equally available to the homeless and to migrants.

In a twist on the perspective presented in this report, Couture et al. 60 argue that frequent use of health services is itself a marker of vulnerability. Such frequent users, they maintain, deserve attention because of high costs and negative outcomes, such as lower quality of life and higher mortality. For this reason, they contend that health-care systems should offer interventions tailored to the needs of frequent users and to their level of health literacy, including strategies to promote activation. Such an approach is best evidenced in the health diversity initiative for migrants,44 but, it could be argued, it has wider applications for populations in whom health literacy, not language proficiency, remains a key issue.

A further paradox relates to the fact that the provision of multiple channels for accessing services may maximise the chances of certain vulnerable populations being able to access an appropriate service while, at the same time, adding to the complexity facing those vulnerable populations for whom being able to navigate the health-care system is the more significant problem. For example, one of the experiences of the pilot front of A&E services initiative was poor uptake related to poor awareness of the role of the new service compared with the existing ED. 46–49 Other services and initiatives specifically exclude specific vulnerable groups; for example, the Guy’s and St Thomas’ and King’s College Hospitals rapid response service40 excludes patients for whom the primary diagnosis is a mental health or substance disorder problem.

Strengths and limitations of this review

This three-stage review has optimised the use of resources by, first, undertaking a mapping phase, before undertaking more intensive analytical phases, ensuring that the final review approach is meaningful and potentially useful. As anticipated by the review team and HSDR programme commissioners, there is a limited body of evidence on interventions for vulnerable people who have service delivery outcomes. Findings from papers included in the mapping review confirmed the existence of evidence for the interventions used having a positive effect on ED attendance, although, as noted, many of these interventions did include an ‘outside the ED’ element. This finding, in turn, justified our widened scope for the subsequent phases of the review.

The mapping review looked only at interventions for vulnerable groups that are delivered within the EUC system. We were aware that interventions that are being delivered outside the EUC system, within acute or primary care, offer the potential to reduce ED use by these groups. For our intervention analysis, we imposed a more forgiving requirement in recognition that (1) interventions may be triggered within the ED but followed up in primary care, (2) re-presentation and/or readmission of frequent users made such a distinction somewhat arbitrary and (3) the EUC system could conceivably include extensions offered in primary care that are conceived as operating seamlessly as part of an extended EUC system (e.g. rapid access doctors, urgent care clinics and front of A&E general practice).

Although, as reported in Chapter 1, it is acknowledged that patients in these vulnerable groups are frequent and heavy users of the EUC system, there is limited evidence to show that interventions specifically target use of services by these groups. The screening process for this review indicated that interventions are often reported in terms of patient outcomes, rather than health system or service delivery outcomes, thus limiting the number of papers included in this review.

The intervention analysis ensured that details of the interventions can be identified and explored and differences manifest at a local level can be identified and clearly flagged. A major contribution of this overall review (particularly given the dearth of specifically targeted interventions identified from the literature) is the review of current practice (like many researchers, we resist the label of ‘good practice’ for promising and yet superficially evaluated or unevaluated initiatives). This has expanded our pool of candidate interventions, although, once again, evidence of specific targeting is limited and the vulnerable populations from our target list benefit, if at all, only collaterally from initiatives and, within these initiatives, differentially according to the specific nature of their vulnerabilities. This additional searching (citation and reference list checking) highlighted additional interventions to include in the review and the full intervention analysis has highlighted germane aspects of the interventions of interest. By combining the publication and practice evidence bases, we have been able to offer a potentially valuable perspective on the current landscape, where work is being done and where future work is required.

Weaknesses relate to the diverse nature of the evidence, ranging from published studies to brief mentions in news stories or ‘good-practice’ casebooks. This poses particular challenges for synthesis, especially as this tends to polarise the twin considerations of rigour and relevance, which, ideally, should be complementary when characterising the evidence base. The trialled interventions typically originate from the USA, and the locally sensitive initiatives from the UK are likely to be vulnerable to risk of bias, particularly selective reporting bias. Furthermore, initiatives that are evaluated in the short term are vulnerable to several implementation biases compared with routine roll-out and adoption, for example the disproportionate influence of the ‘hero innovator’ and the enthusiasms of early adopters. This narrative may partly, although not entirely (see below), explain the differential success of the front of A&E streaming initiatives at Luton and Dunstable, and Frimley Park. 46–49

Initiatives of alleged good practice carry known limitations in frequently displaying the presence of reporting bias, in the use of relative measures of performance (e.g. percentages) with inadequate measures of baseline performance and in limited evaluation objectives and time frames. As a consequence, they do not necessarily depict an accurate picture of real-world implementation. In particular, they typically invoke cost savings without presenting the actual cost of designing, planning and introducing the intervention. Nevertheless, they do indicate potential candidate interventions that might benefit from more rigorous evaluation. An illustrative example in this review is the front of A&E general practice initiative, which yielded vastly different results in Luton and Dunstable, and Frimley Park, leading at least one commentator to highlight the likely context sensitivity of such interventions. 50 Indeed, Edwards50 reminds us that there are no simple solutions or policies for complex problems. Notwithstanding a letter in 2017 from the chief executives of NHS England and NHS Improvement, instructing trusts to ensure that every hospital in England should use hospital-based GPs to triage and redirect patients as they are first seen in A&E,47 commentators could point to a differential effect between well-suited hospitals serving populations that use A&E extensively for primary care concerns48 and areas where GPs see a high proportion of emergency cases on the same day. 46 This context sensitivity provides a strong indicator for further realist evaluation61,62 and realist synthesis approaches63 within an ED context.

Implications for health care

• The evidence highlights that the reasons for the increased patterns of use of the ED by vulnerable groups are complicated and encompass a wide variety of drivers, including burden of disease, access to primary care and patient preference.

• The evidence indicates that the specific needs of and barriers for each subpopulation among those categorised as vulnerable may differ, requiring a nuanced understanding of these diverse populations. Decision-makers need to gain an accurate picture of the distribution of specific vulnerable populations in order to target service responses appropriately.

• The review found a notable shortage of interventions designed specifically to reduce demand for ED services from vulnerable groups, with existing interventions being mostly delivered within the community setting. Front of A&E services are being strongly promoted but there is little evidence that they particularly address the needs of vulnerable populations. Indeed, they may share some of the physical and geographical barriers to access typically encountered in connection with EDs.

• The review found that the majority of interventions aim to tackle the problem of increased patterns of ED use by vulnerable groups at a general population level (such as front of A&E general practice or urgent care clinic approaches) or target frequent attenders as a discrete group. Health-care decision-makers could seek to monitor the prevalence of identified vulnerable groups presenting at EDs and primary care alternatives and seek to avoid local barriers to appropriate use.

• The review identifies common components of interventions targeting frequent users: (1) analysis of frequent attenders, (2) multisectoral teams, (3) case management approaches and (4) shared documentation. Health-care providers can select from these core components to construct a locally sensitive package of responses and submit these for ongoing evaluation.

Recommendations for research

• The review found that interventions specifically targeting the vulnerable groups identified in this report need to be co-designed,64 developed, trialled and rigorously evaluated. A particular need was identified for further research on the clinical effectiveness and cost-effectiveness of promising UK initiatives.

• The evidence indicates that evaluations of these types of interventions will probably require mixed-methods approaches, with an important contribution to be made by qualitative inquiry.

• The review suggests that interventions may also require an explicit and conceptually sound theoretical basis, particularly in understanding vulnerability and how it affects ED use.

• The role of technology and health informatics, briefly touched on by this report, requires further examination, particularly relating to the identification and co-ordination of frequent users. 64

• The findings of the review further indicate the potential for realist evaluation approaches,63,65,66 especially as several interventions identified seem to be heavily context dependent.

Contributions of authors

Andrew Booth (Reader) co-led the study, including contributing to data analysis and report writing.

Louise Preston (Research Fellow) co-led the study, including contributing to data analysis and report writing.

Susan Baxter (Senior Research Fellow) contributed to the mapping review and report writing.

Ruth Wong (Information Specialist) contributed to the electronic database searching.

Duncan Chambers (Research Fellow) contributed to the review processes, including sifting and data extraction.

Janette Turner (Reader in Emergency and Urgent Care Research) provided topic expertise.

Data-sharing statement

All available data can be obtained from the corresponding author.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HS&DR programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HS&DR programme or the Department of Health and Social Care.

Search strategy

1. *Emergency Service, Hospital/

2. *Emergency Medical Services/

3. *Emergency Medicine/

4. (emergenc* adj2 service*).ti,ab.

5. emergenc* care.ti,ab.

6. ((urgent or unscheduled) adj care).ti,ab.

7. emergenc* department*.ti,ab.

8. *Ambulances/

9. ambulance*.ti,ab.

10. or/1-9

11. *Vulnerable populations/

12. *Poverty/

13. *Socioeconomic Factors/

14. (vulnerable or socioeconomic* or disadvantaged or depriv* or poverty or poor or low-income* or low income* or low pay or low* paid).ti.

15. ((vulnerable or socioeconomic* or disadvantaged or depriv* or poverty or poor or low-income* or low income* or low pay or low* paid) adj3 (individual* or people or person* or famil* or population* or communit* or neighbourhood* or group*)).ab.

16. ((financ* or economic* or money) adj2 (hardship* or difficult* or problem* or worries or worry)).ti,ab.

17. *Social Isolation/

18. (social* adj1 (exclu* or inequalit* or isolat*)).ti,ab.

19. or/11-18

20. Rural Population/

21. (rural or remote or coast* or geographical* isolat*).ti.

22. ((rural or remote or coast* or geographical* isolat*) adj3 (people or patient* or population* or communit* or neighbourhood* or group* or area*)).ab.

23. ((hard* or difficult) adj2 (reach or locate or find)).ti,ab.

24. or/20-23

25. refugee*.ti,ab.

26. asylum seeker*.ti,ab.

27. (migrant* or immigrant* or emigrant*).ti,ab.

28. Refugees/

29. “Emigrants and Immigrants”/

30. “transients and migrants”/

31. ((human or child or people or person) adj traffick*).ti,ab.

32. (“first generation” or “second generation” or “third generation”).ti,ab.

33. (“new arrival*” or settler* or newcomer*).ti,ab.

34. ((multi or trans or cross) adj cultural*).ti,ab.

35. (multi adj (ethnic or racial or lingual)).ti,ab.

36. diaspora.ti,ab.

37. ethnic groups/

38. (traveller* or gypsies or gypsy or gipsy or gipsies or romany or romanies or romani or romanis or rromani or rromanis or roma).ti,ab.

39. (african american or african americans or asian or asians or black or blacks or hispanic or hispanics or indian or indians or latino or latina or latinos or latinas or native american or native americans).ti.

40. (bme or black ethnic minorit* or black minorit* ethnic* or south asian* or bangladeshi* or pakistani* or indian* or sri lankan* or asian* or east asian* or chinese or taiwanese or vietnamese or korean* or japanese or afro-caribbean* or african-caribbean* or caribbean or african* or black* or afro* or islam* or hindu* or sikh* or buddhis* or muslim* or moslem* or christian* or catholic* or jew*).ti.

41. or/25-40

42. *Unemployment/

43. (unemploy* or jobless or workless).mp.

44. ((job* or work or employment) adj3 (redundan* or insecur* or loss* or lose or lost

45. or search* or seek* or find*)).ti,ab.

46. (economic* adj3 inactive).ti,ab.

47. or/42-45

48. exp *Homeless Persons/

49. (homeless* or rough sleep*).mp.

50. ((unstabl* or emergency or temporary or inadequate or poor or overcrowd*

51. or over crowd*) adj3 (hous* or accommodation or shelter* or hostel* or dwelling*)).ti,ab.

52. runaway*.mp.

53. (street adj3 (individual* or person* or people or group* or population*)).ti,ab.

54. or/47-51

55. (addict* or ((substance* or alcohol* or drug* or cocaine or heroin or amphetamine*

56. or marijuana or cannabis) adj2 (misus* or abus* or depend* or use* or using))).ti,ab.

57. (legal high* or (psychoactive adj (substance* or product*))).ti,ab.

58. *Substance-Related Disorders/

59. *Drug Users/

60. *Alcoholics/

61. or/53-57

62. 10 and (19 or 24 or 41 or 46 or 52 or 58)

63. Case report.tw.

64. Letter/

65. Historical article/

66. 60 or 61 or 62

67. exp Animals/

68. Humans/

69. 64 not (64 and 65)

70. 63 or 66

71. 59 not 67

72. limit 68 to (english language and yr = “2008 -Current”)