Youth violence intervention programme for vulnerable young people attending emergency departments in London: a rapid evaluation

Evidence review

The evidence review was conducted in two parts and focused on youth interventions delivered in hospital settings to reduce or prevent violent crime and harm to young people (e.g. from criminal and gang exploitation) and involving professionals such as social workers, trauma experts and youth specialists who work alongside clinicians. We followed recommendations on conducting systematic scoping reviews22 (e.g. predefined eligibility criteria and research questions) to map out the topic and to identify recent evidence available on this topic and any theories or conceptual frameworks that have been applied. We used a two-phased search process – one exploratory, one targeted – focusing on the medical literature to understand what was currently known about youth-orientated violence reduction services delivered in hospital settings, specifically their impact and outcomes monitored, and to determine whether there are gaps in knowledge such as, for example, cost-effectiveness. We also looked for evidence of factors that either support or hinder the implementation of such services and identified any conceptual or theoretical lenses applied in this area, such as behavioural concepts applied to evaluate ‘teachable moments’ and drawn from social science subdisciplines, such as cognitive psychology. Our searches looked for evidence about these interventions from both within the UK and internationally.

Further details on the methods used in the review including the search methodology are described in Chapter 3.

Qualitative scoping interviews and documentary analysis

Qualitative data collection was conducted in two phases, and involved semistructured interviews, observations of staff meetings and review of Redthread documents and materials. Phase 1 was an exploratory stage which aimed to understand the Redthread programme theory and the background to the introduction of the charity at UCLH. Phase 2 (described below) consisted of a single-site, process case study to understand implementation of the Redthread intervention at UCLH as well as staff perceptions of Redthread’s impact and progress.

All interviews were conducted from April 2021 once Redthread youth workers were back on site at UCLH. Recruitment used a mixture of purposive and snowball sampling to capture the views of a range of respondents, both those close to the Redthread intervention (e.g. youth workers), subject experts and those who might be less familiar with Redthread (e.g. junior doctors/nurses working in emergency care). The main criteria for UCLH staff respondents to take part was being directly involved in the care of young people at risk of harm and in a position to refer young people to the Redthread service. Staff were identified with the support of UCLH clinical collaborators and all respondents were emailed an information sheet prior to taking part and given the opportunity to ask questions about the evaluation and interview process.

During phase 1, available Redthread documents were supplied to the evaluation team to map out referral pathways into the service and analyse the programme theory; that is, what the intervention aimed to do and how, and its main component parts. Programme theory can be defined as describing ‘how an intervention is expected to lead to its effects and under what conditions’. 23 We sought to understand Redthread’s programme theory (what they call their ‘theory of change’) to explore how the service was being implemented and was understood by staff at UCLH and to identify any contextual adaptations. We did not attempt to further develop or revise the Redthread logic model but were aware that the charity regularly reviews and updates its own materials.

This work was supplemented by exploratory discussions with key stakeholders at UCLH and members of the evaluation advisory group. We also examined the findings of previous Redthread evaluations undertaken at other trusts to see how they interpreted the Redthread programme.

During phase 1, we conducted nine scoping qualitative interviews with key stakeholders. Respondents were: two hospital consultants working in children and young people’s services at UCLH, six Redthread youth workers and other staff (e.g. managers, programme co-ordinators) and one senior NHS director involved with youth violence reduction interventions with knowledge of similar programmes at other NHS trusts. The aim of these interviews, which were semistructured, was to capture insights about the early introduction of the Redthread service at UCLH and the wider context. In particular, the early interviews aimed to understand what meaningful success looked like to those involved in delivering the intervention at UCLH (e.g. reduction in admissions, onward referrals to other services, positive case work with an individual) and to explore any skills and training required to deliver the intervention. Finally, we noted any novel service components that were new to the UCLH setting or arising because of COVID-19 (e.g. virtual delivery). Further details of our methods are described in Chapter 4.

Feasibility study of a quantitative evaluation

The aim of the quantitative component of phase 1 was to explore the feasibility of evaluating the impact of the Redthread service given the available data. The results of this investigation would then inform the nature of any quantitative analysis that would be undertaken in phase 2.

For a measure of impact, we focused on the use of hospital services, specifically, future hospital attendance either at ED or as an inpatient relating to assault, mental health, substance abuse as well as those who were perceived to be at risk of harm. We developed a set of options for evaluating impact that covered different perspectives:

1. the perspective of someone using the service

2. the perspective of the acute trust where the service is based (UCLH)

3. the perspective of the local communities where people using Redthread’s service live.

These perspectives were assessed by investigating the data requirements of each option, including any access to individual person-level data, the linking of their records, the identification of comparators and necessary sample sizes. We also identified possible barriers to accessing the necessary data such as patient consent, information governance approvals and time to obtain these approvals in relation to the duration of the project. Our analysis was informed by an investigation of HES admitted patient care and ED datasets and the Emergency Care Dataset (ECDS) alongside discussions with Redthread, UCLH, NHS Digital and our expert advisory group. Further details on the methods used in this feasibility analysis are described in Chapter 6 and Appendices 3–5.

Preliminary economic assessment

During phase 1 we conducted a documentary analysis of the Redthread evaluations that had already taken place since 2021. We collected information on the evaluation aims, their main components and their results, to identify the existing gaps in the evidence. This informed the assumption and parameters we adopted in our economic analysis.

Set-up of an evaluation advisory group

During phase 1 we also set up an evaluation advisory group to meet up to three times during the course of the evaluation (virtually or in person) and involving representatives from the NHS, health-care and relevant public agencies. Terms of reference were drafted with the aim of each meeting to provide helpful challenge and advice to the evaluation team from stakeholders more external to the programme.

Process evaluation – qualitative case study

For phase 2 of the qualitative data collection, we completed a process evaluation, where the unit of analysis was emergency and specialist children’s and adolescent services at UCLH. A process evaluation was considered suitable because the Redthread service is a complex intervention and randomisation was not feasible in this study. What was required were insights about delivery and overall impact to inform future implementation. 24 Process evaluations aim to understand how a programme or intervention is implemented, including any important decisions that influence how it operates in practice, any important adaptations, and the contextual factors that influence the intervention and its implementation. 25 We were therefore particularly interested in understanding:

1. the different processes and mechanisms at work locally, such as different referral pathways to access the Redthread service

2. any adaptations made to the service over time (e.g. due to COVID-19)

3. the reach of the service (e.g. the extent to which it had spread across different hospital departments)

1. critical implementation factors (e.g. what was reported to help youth workers to deliver the programme, or hospital staff to refer young people to it).

This part of the study involved additional qualitative data collection (13 further semistructured interviews and three observations of staff meetings) and focused on the mechanisms and emergent themes identified in phase 1, including any linkages between them, and any features of the hospital setting and its environment that were shaping delivery of the Redthread programme. Examples of the factors that we explored in this phase included:

1. Internal context: departmental leadership and cross-departmental working; professional buy-in (especially by emergency, trauma and paediatric staff); hospital data sharing and governance policies; senior/executive team support for the intervention; staff training; perceptions of need; communication of information about the intervention.

2. External context: demands on hospital services (e.g. young people presenting at UCLH and their needs); any trust collaboration with external public agencies; lines of accountability within the area (e.g. responsibility for youth crime prevention and safeguarding for children and young adults).

The additional semistructured interviews were with clinical and non-clinical UCLH employees and Redthread staff. They included hospital social workers, Redthread youth workers and managers, a paediatric nurse, consultants (children and young people’s services, child and adolescent psychiatry) and a junior doctor. Of the interview respondents, a small number of individuals were interviewed twice (in phases 1 and 2) because of their close involvement with implementing the Redthread service. All interviews and observations were conducted remotely via Microsoft Teams^® (due to COVID-19) following consent to participate. Interviews were digitally recorded and transcribed verbatim. Three observations of staff meetings were completed [e.g. an adolescent ward psychosocial multidisciplinary team (MDT) meeting] alongside review of essential Redthread documents. Anonymised field notes taken during staff meetings.

To support analysis and interpretation of the qualitative findings, the qualitative researcher and wider evaluation team held discussions with the evaluation advisory group, and held meetings with other researchers involved in Redthread evaluations at different NHS trusts, plus meetings with Redthread staff and UCLH clinical collaborators. For the final analysis, all interview transcripts (n = 22) and observational field notes were read by the lead researcher for data familiarisation, alongside key materials such as the Redthread planning and implementation guidance, and other relevant documents (e.g. the Redthread youth worker manual). Findings were analysed thematically with a specific focus on answering the evaluation questions (see Chapter 4 for further details of the themes).

Unfortunately, it was not possible to conduct interviews with young people who had experienced the Redthread service at UCLH or at other NHS trusts for ethical and practical reasons (e.g. identifying young people would have data sharing and confidentiality implications), although this possibility was explored with Redthread. Further details on the methods employed are described in Chapter 4.

Analysis of local Redthread data

Over the course of the project, we obtained data from Redthread that included the characteristics and reasons for hospital presentation of individuals who engaged with the service and of those who declined to take part. Using univariate analyses, we compared these characteristics to identify any differences between those who chose to engage and those who did not. All the data were provided at an aggregated level for each characteristic separately, which precluded more complex multivariate approaches. Further details of the methods employed are described in Chapter 5.

Economic evaluation

Based on the outcomes of phase 1 of the evaluation and documentary analysis, we conducted a CCA of the Redthread service at UCLH. Consequences were derived from Redthread’s risk assessment tool, which measures changes in perceived risks faced by young people before and after intervention. Costs of delivering the service were obtained from Redthread, and these were compared with the costs of hospital ED attendances and admissions for reasons that might suggest eligibility for Redthread, over the three-year period 2018–2021. Hospital costs were obtained from UCLH. Further details of the methods are described in Chapter 6.

Phase 1: exploratory search

Phase 1 consisted of an initial scoping review focused on youth-focused interventions delivered in emergency and hospital settings to reduce violent crime and address safeguarding risks (e.g. from knife crime, assault). It was led by one researcher (JF) with input from another researcher (JL). The review was intended to be broad in scope, with the aim of better understanding the topic and confirming the types of key terms that could be employed in a more targeted search to be conducted later. It was conducted in the earliest stages of the project, prior to empirical data collection.

A population, intervention, control/comparison, outcome framework was used to define the search terms:

1. population: youth patients 10–24 years of age, victims of interpersonal violence (excluding victims of self-harm, sexual violence, and child abuse)

2. intervention: youth or social worker hospital-based interventions

3. comparison: standard of care, no treatment, or differential treatment of a control group

4. outcome: recidivism, readmission, social services use, feasibility or patient self-reported outcomes.

The exploratory search was conducted in January 2020 and limited to English text sources but with no date restriction as the review was intended to be exploratory, as part of an initial search for available papers. The databases searched were PubMed, MEDLINE, the Cochrane Library and Embase databases. The a priori decision was made to limit screening to the first 200 hits (as sorted by best match) to provide a manageable number of initial sources of evidence; thereafter, further inclusions were determined based on reviewing cited literature for its relevance. Conference abstracts, editorials and commentaries were excluded. Fourteen studies were identified for inclusion due to their relevance to the evaluation, namely, studies that could help the researchers to understand the wider context of youth violence hospital interventions and programmes, and the types of prevention strategies commonly found in health care settings. These findings are presented in this chapter.

Phase 2: structured scoping review

A rapid scoping review was deemed appropriate due to the complexity of the topic, the emerging nature of the knowledge in this field, a need to understand the nature of the evidence base quickly and any knowledge gaps. In short, we needed a quick overview of the latest evidence related to hospital-based, youth-focused emergency care interventions such as Redthread. The findings of the phase 1 exploratory search informed drafting a review plan (not published) to guide a more systematic search of the medical and health care literature databases, which received team input (e.g. listing key words). The phase 2 scoping search also took into account early empirical findings from the qualitative interviews to understand the Redthread programme at UCLH (see Chapter 4).

This review was open to capturing a range of potential benefits of youth violence prevention services and youth worker programmes in hospital settings and followed a modified version of the PICO structure used in phase 1. As per guidelines for scoping reviews, we were chiefly focused on the population, concept (e.g. ‘teachable moment’, youth intervention following admission for trauma) and context (e.g. EDs) due to the wide range of possible outcomes of YVIPs. In terms of types of evidence, we were interested in identifying peer-reviewed published studies using a variety of study designs and any other evidence reviews (e.g. systematic and scoping).

The guiding questions and aims of the review remained the same, as outlined above, and the review inclusion criteria was as follows:

1. Population: young people, adolescents and children (aged up to age 30 years, to capture a wider literature), specifically groups at risk from gang-related exploitation, physical assault and injury (e.g. from knife attacks, shootings), sexual exploitation, human trafficking and other forms of violence (e.g. from peers and fighting).

2. Intervention: youth or social worker (or equivalent roles) hospital-based interventions aimed at risk management and prevention, and initiated within EDs/major trauma centres (MTCs).

3. Context: hospital-based interventions.

Inclusion parameters:

1. Study type: any [e.g. feasibility study/pilot, RCT, qualitative, evaluation, mixed-methods, cost–benefit analysis (CBA)] and literature reviews (narrative, scoping, systematic).

2. English language.

3. Peer-reviewed (i.e. no conference proceedings or abstracts).

4. Publication period: 2012–2022.

Searches were carried out on two medical databases (Medline and Embase) in February 2022 to focus on the health literature using key words and medical subject headings terms (see Appendix 1).

Results from Medline and Embase were imported into EndNote for deduplication and then exported to Rayyan.ai software for screening. Two researchers (JL and JF) independently assessed the retrieved sources using the inclusion criteria and considering relevance for answering the aims of the review. Papers were discussed and selected against the inclusion criteria and in light of their quality (e.g. an explication of the study design, nature of the intervention, intervention context and any limitations). Excluded papers included crisis interventions that did not feature a hospital-based youth or social worker’s input or equivalent role (e.g. sexual health crisis teams), conference and meeting abstracts and clinical case reports. A large number of studies were found to focus on community-based youth interventions, where young people are recruited to external programmes via EDs and seen by community youth workers or mentors, and this required further discussion and accessing full papers. It was decided that these studies should also be excluded. However, the boundary between hospital and community interventions was sometimes difficult to discern. Finally, mental health crisis interventions and any programmes designed to educate health professionals about youth violence (e.g. e-learning modules) were also excluded. The results for phase 2 are provided in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis diagram below (Figure 3).

FIGURE 3.

Phase 2 search results.

As we were following guidance for scoping reviews and conducting the search under resource and time restrictions, the focus was on mapping the evidence base in the literature rapidly, with the aim of describing the types of studies available according to our predefined eligibility criteria and analysing any knowledge or research gaps in light of this rapid evaluation. The evidence search was not intended to be a systematic review, and no meta-analysis of the results was performed. Results were limited to two databases and we summarised the findings in a table detailing the study type and main findings. Extracted data focused on study location, intervention characteristics/components, population; setting/context, outcomes and key findings.

In the final phase, we performed Google search for grey literature using key terms (e.g. ‘youth hospital violence programme’, ‘Redthread evaluation’, ‘violence prevention and hospital’) to see if there was any additional evidence arising from the UK. We also engaged with the Redthread charity about past or current service evaluations with which they were involved. In this way, we identified some evidence scans and evaluation reports that had not been picked up in our search of health-care databases, as well as published evaluations of youth violence prevention programmes. We discuss these at the end of this chapter. All sources were assessed for their relevance to the evaluation, namely, a focus on youth violence prevention and hospital-based interventions (i.e. programmes such as Redthread or similar).

Phase 1: exploratory review

Summary: phase 1

Our exploratory review found evidence regarding different types of interventions to reduce youth violence recidivism, dating from the early 1990s onwards. This suggests that: (1) successful programmes are generally long term, restorative in nature, and based on CBT techniques designed to help victims identify cognitive distortions and develop adaptive social skills, (2) successful interventions often include active participation of family members and coordinated community services, and (3) hospital-based youth worker programmes may increase social services use, decrease self-reported peer violence, and lower hospital recidivism rates. However, most of the available literature is limited by reporting outcomes over a period of less than one year, and the long-term durability of self-reported effects and recidivism rates is unknown. Moreover, recidivism rates are often described at single institutions, potentially limiting their validity as an outcome measure. Long-term, multicentre longitudinal studies are necessary to better understand the effect of hospital-based youth violence interventions.

Phase 2

The phase 2 structured search identified 20 academic articles and more recent studies. The search focused on papers published post-2012 to bring the first review up-to-date. The findings below are organised thematically around our evaluation questions and lines of enquiry. Table 2 outlines the 20 studies, 15 of which involved young people directly. Studies were predominately from the USA, plus one study from Canada and one from the UK. Programmes covered a wide age range (6–30 year olds), included males and females and different ethnic groups (Black, White, Asian, Turkish, Mixed, Hispanic).

Summary: phase 2

Our updated phase 2 search found new studies published after 2012, which had not been discussed in previous reviews due to our search terms and more recent date range. In particular, we found two new economic studies, new RCT and feasibility studies and a UK service evaluation. We observed that factors described in the literature that support or hinder the implementation of youth violence interventions in hospital settings include:

1. staff training and awareness to make referrals

2. characteristics of young people who may decline to engage with programmes (e.g. older adolescents)

3. staff turnover

4. co-location of specially trained social and youth workers in EDs.

Lessons are primarily drawn from North America, however, where case management approaches and brief interventions to youth violence prevention typically apply motivational interviewing techniques and cognitive behavioural approaches. Notably, some brief interventions now employ technology to standardise delivery, such as automated messaging to young people.

The evidence for efficacy and effectiveness across different interventions is mixed, owing to variations in study design, the professionals involved in delivery (youth workers, trauma nurses, social workers, researchers, mentors), relatively small sample sizes (given the numbers of young people presenting to EDs clinically with violent injuries and from exploitation), the nature of the interventions themselves, variable follow-up time periods, and differences in the primary and secondary outcomes reported.

Many studies rely on self-reported measures using questionnaires and only one study had attempted to examine reinjury and admission rates across more than one hospital setting. The theoretical lens we found applied was a social ecological model while the concept of the ‘teachable moment’ was discussed widely in the literature.

Our review found very limited qualitative evidence exploring the perceptions of young people and staff who had implemented the interventions, and reasons for engagement (or lack thereof). Any evidence that was presented tended to be positive and represented the views of those that had chosen to take part in these programmes.

Evaluations in progress

1. Redthread in the Midlands: a service evaluation funded by NHS England Midlands (2020–22) was being undertaken by Liverpool John Moores University and nearing completion.

2. Redthread: a quasi-experimental design evaluation funded by the Youth Endowment Fund (2021–23) involving the University of Birmingham. This is an initial one-year pilot from April 2022 to March 2023 focused on 10- to 17-year-olds as part of the Another Chance – Diversion from the Criminal Justice System round. 82

Finally, a six-year, mixed-methods evaluation of the Oasis hospital-based violence prevention programme at St. Thomas’ Hospital, London, by researchers at Middlesex University was published in 2016. This evaluation found positive findings in terms of efficiency, implementation and impact on young people who had engaged with the service, some of whom were followed up after at least one year. 6 Benefits also included high staff satisfaction with the services and fewer reattendances among those had received the intervention.

Interview data collection and analysis

A mixture of purposive and snowball sampling was used to identify UCLH staff that were both central and more peripheral to the Redthread intervention to take part in the evaluation. Interviews took place with a range of clinical health-care staff (e.g. nurses, junior doctors, senior psychiatrists and paediatricians) and non-clinical staff (safeguarding teams, Redthread youth workers and hospital social workers). The primary objective was to speak with staff involved in the care of young people at risk of harm at UCLH who either knew about the Redthread service and/or might refer young people to it. Phase 1 interviewees were identified through scoping work, while phase 2 interviewees were largely recommended by UCLH and Redthread staff involved with the service. The team were able to capture the views of both service ‘champions’ and staff with lower levels of awareness of Redthread service (e.g. junior doctors who worked across different EDs and wards). All respondents were individually emailed a study information sheet, consent form and given the opportunity to ask questions about the evaluation to a qualitative researcher before taking part. Some clinicians emailed their colleagues on behalf of the researcher to inform them about the study and encourage their participation.

Due to the impact of COVID-19, recruitment of front-line staff from the adult ED was particularly challenging, with some staff declining to take part because they felt they did not know enough about the Redthread service. This confirmed a finding from the case study: that the intervention had not fully embedded outside of children’s and adolescent services at the time of this evaluation and that Redthread’s awareness-raising activities had been severely disrupted by the pandemic.

One lead qualitative researcher (JL) conducted all the interviews, with a NIHR RSET colleague present in some instances where quantitative aspects (e.g. hospital and Redthread data flows) were being explored. Interviews were conducted remotely and on a confidential basis by telephone/MS Teams^®. Interviews were digitally recorded with explicit written and/or verbal consent and were transcribed verbatim. A total of 22 interviews were completed across phases 1 and 2, with a small number of individuals interviewed twice because of their close involvement with the service. Different topic guides were used for interviews with hospital and Redthread staff but explored similar themes. Each topic was piloted to ensure that the questions resonated and were clear to respondents and then finalised (for topic guides, see Appendix 2).

Compared with phase 1 interviews, which were more exploratory, phase 2 interviews had a focus on the perceived impact and effectiveness of the service. Staff were asked to provide anonymised clinical case examples to explain how a young person had been referred to Redthread, why the young person had presented at UCLH, and how they had been supported by Redthread youth workers. This allowed two anonymised clinical case vignettes to be included in the findings, helping to elucidate how the Redthread service operates in practice and connections between the Redthread service and other organisations, especially those within the community. Finally, the phase 2 interviews explored factors that appeared especially important for the successful delivery of the Redthread programme in the hospital (enablers and barriers).

It was not possible to conduct interviews with young people who had experienced the Redthread service at UCLH or at other NHS trusts for ethical and practical reasons (e.g. identifying young people would have data sharing implications), although this possibility was explored with Redthread.

Observations

A qualitative researcher conducted a total of three non-participant observations of staff meetings to understand the implementation of the service, and to explain the purpose of the NIHR RSET evaluation to hospital staff at UCLH. This proved to be particularly helpful for phase 2 interview recruitment, especially given that hospitals were dealing with the ongoing effects of COVID-19. Meeting observations were undertaken between October and December 2021 with explicit permission from UCLH clinical collaborators and attending staff. Two virtual MDT meetings were observed during which Redthread youth workers worked alongside clinicians, helping to identify young people potentially eligible for their support. A staff safeguarding training session was also observed in which Redthread discussed their service. All meetings were held via MS Teams on account of the COVID-19 pandemic. The researcher took carefully anonymised field notes which were typed up and then securely saved electronically in UCL folders only accessible to the evaluation team.

Analysis

Interview transcripts and observational field notes were read by the lead researcher for data familiarisation and triangulated with the Redthread planning and implementation guidance gathered in phase 1 and other relevant documents (e.g. the Redthread youth worker manual). Emergent findings were discussed in team meetings and with clinical collaborators to confirm initial interpretations. The final qualitative data set (integrated phases 1 and 2 findings) were analysed with a specific focus on answering the evaluation questions. Interview transcripts (transcribed verbatim), Redthread documents and field notes were reviewed by the qualitative researcher and data extracted and organised according to following themes:

1. Redthread history and background; theory of change (e.g. ‘teachable moment’); key components of the intervention and associated processes (e.g. eligibility, consent model); data capture and monitoring; implementation context at UCLH (services and population); adaptations due to COVID-19; youth worker perceptions (e.g. implementation journey, impact, successes and challenges); staff perceptions (awareness of Redthread, impact, fit with existing services; successes and challenges); staff training; enablers and barriers to implementation; timeline and key events (e.g. Redthread introduction to UCLH).

Line-by-line coding of transcripts was not performed, but qualitative findings were compared in an iterative process with those arising from the economic and quantitative aspects of the evaluation. In addition, validation of findings was supported by providing draft chapters to Redthread for review; this was important to confirm a correct interpretation of the history of the charity, the local adaptations of the service offer at UCLH during COVID-19 and any sensitive issues regarding the presentation of qualitative findings (e.g. clinical vignettes to ensure patient anonymity).

Background to the Redthread intervention at UCLH

At the time of the evaluation, Redthread staff confirmed that the organisation (a registered charity) was established in 1995, in south-east London, initially as a group set up by parishioners from a number of local churches in Herne Hill, Dulwich and the surrounding area. Researchers were informed that, in the early days of the charity, it was felt that local children and young people could be more involved in the community if provided with the facilities and activities to do so. This led to a Redthread youth club being set up, providing a range of regular activities for young people.

Later in 2002, the senior manager of the charity started to move its youth work into health-care settings, realising that there was an opportunity to engage with young people (in particular, higher-risk groups) at the point when they accessed health care services. One Redthread interviewee described how the charity had been attentive to ‘high-risk’ young people who were falling outside of public services, such as social care, and often had a deep mistrust of authorities. Redthread started a young people’s clinic in partnership with a local general practice, which was run after school for 13- to 19-year-olds, regardless of whether or not they were registered. In primary care, it had been observed that young people could be missed and unknown to general practices while being frequent attenders at local EDs or other health services, such as child and adolescent mental health services (CAMHS) or sexual health clinics. Young people also risked being missed on account of silos existing between statutory services, and a lack of a joined-up preventative health approach for young people at risk of violence and exploitation at the pan-London level.

Against this backdrop, the first Redthread site was founded in 2006 at King’s College Hospital in south-east London. Later, the charity scaled up its activities with additional MTCs coming on board (at St. Mary’s Hospital in north-west London in 2014, and St. George’s Hospital in south-west London in 2015). This was enabled by funding from the Mayor’s Office for Policing and Crime and other funders. A senior Redthread manager explained how youth workers and clinicians were noticing patterns of injury affecting young people presenting in the EDs of MTCs. Injuries ranged from knife and gunshot wounds to attacks from broken bottles that a young person would present as an ‘accidental’ injury, even if the wound was likely to have been caused by a violent assault. By running audits on hospital data at the adopting MTCs for the cohort aged 11–24 years, evidence about who attended for violent injuries and later reattended in EDs was identified: ‘there was this sort of propensity for reinjury and this escalation of injury’ (Redthread manager, RED004).

Redthread gradually moved into local EDs across London, such as Homerton University Hospital and later UCLH, as well as to other hospitals based in the Midlands. This was viewed as a way of ‘joining the dots’ and not only focusing the charity’s work in MTCs.

The introduction of Redthread youth workers to hospital EDs coincided with growing policy attention on a holistic ‘public health’ approach to preventing youth violence in England,83 and the introduction of violence reduction services within the NHS, both in and outside London. 18 Redthread is one of a number of charities providing hospital-based interventions in London at the present time. For example, the St. Giles Trust, which is located at the Royal London Hospital (an MTC, part of Barts Health NHS Trust),84 Newham University Hospital, the Whittington Hospital and Northwick Park’s ED (part of London North West University Healthcare NHS Trust). 85 Oasis, another charity, has youth workers based in the ED of St. Thomas’ Hospital (part of Guy’s and St Thomas’ NHS Foundation Trust) and North Middlesex University Hospital.

Introduction of the Redthread service to UCLH

In the London Borough of Camden, where UCLH is located, the Camden Youth Safety Taskforce had been established in 2017 and published a report making recommendations to treat ‘youth violence as a broad public health issue rather than simply a law and order issue’, with better collaboration between public services. 83 The report encouraged the early identification of young people at risk of youth violence as a preventative measure, stressing the need for ‘appropriate referrals’ from public agencies and professionals, such as youth workers. 83 In 2018, two young men from Camden estates died from stab wounds drawing attention to a high number of violent incidences occurring within the borough, and highlighting the ongoing risks of violence facing young people in the local community. 83 It was in this context that two hospital consultants who were involved in safeguarding initiatives in the borough and aware of the work of the Camden Youth Safety Taskforce, began to consider the introduction of a violence reduction service at UCLH, around the same time that Redthread was starting to move into local hospital EDs. As one consultant closely involved in the service noted: ‘Redthread originally started in the community and it almost feels like it started out, it’s come into hospitals and we are almost going out again a bit, and so that’s our motivation’.

Clinicians involved in the introduction of Redthread’s service to UCLH spoke of their interest in providing ‘age-appropriate’ services for young people most at risk of harm in the community. Their review of hospital data at UCLH suggested that there was a cohort of vulnerable young people presenting with suspicious injuries and being discharged into the community, meaning this was a missed opportunity to engage with and protect young people at risk of harm:

we did an audit … although this was quite a while ago, that demonstrated that, although we don’t have major trauma here, a good twenty-five per cent of the young people coming through – say 13- to 18-year-olds – with soft-tissue injuries, or minor breaks, they would get patched up and sent out without anybody asking anything other than their tetanus status, actually had a worrying back-story.

Therefore, consultants working in paediatric and adolescent services at the trust played a key role in supporting Redthread service adoption at UCLH and brokering initial discussions between hospital managers and the UCLH Charity in the period 2017–18. Paediatric and adolescent clinical staff were interested in linking up with the Redthread charity to learn more about young people’s reattendances for violent injuries and were motivated by the opportunity for skilled youth workers to interact with young people within the hospital setting in a holistic way. They observed that, while a multidisciplinary approach to adolescent and child health care was already employed at UCLH (e.g. with a hospital-based social worker and play worker already part of the team), the youth worker element was still missing:

Because older teenagers when they come in tend to not be that medically unwell, nobody really asks them who they are having sex with, or how they got a broken bone in their hand. They just go, ‘Okay, you’re fine. You can go’. It was increasingly becoming clear to me that that wasn’t good enough and that we needed to do things differently.

Following discussions between Redthread and UCLH, it was agreed that the service would be introduced, supported by a three-year grant agreement from UCLH (with the funding being provided by UCLH Charity), with additional contributions from the London boroughs of Camden and Islington’s public health directorate in year 1 to make the service fully funded during its mobilisation period and first year of operations. Clinicians favoured such a funding model because it was distinctive from ‘policing money’ (associated with prosecution) and meant that a hospital-based violence reduction service would predominantly be funded by health resources and viewed as a health community initiative. Redthread relies on different sources of funding at each hospital site. This can include money from hospital charities, Redthread’s own fundraising activities, external government sources (e.g. the Mayor of London’s Office) as well as local authorities. Therefore, each hospital adopting the Redthread service will have its own funding arrangement to cover Redthread’s operating costs.

In early 2020, the Redthread service formally launched at UCLH, a few weeks before the COVID-19 pandemic struck.

UCLH: the hospital context

At UCLH, there is a dedicated children and young person’s ED (referred to by staff as ‘paeds ED’) which sees patients up to the age of 17 years. There are also two inpatient wards: an adolescent inpatient ward which treats patients from 13 years upwards and a children’s ward treating children aged under 13 years. A multidisciplinary approach is taken to providing care across these services, in particular the adolescent unit, which encourages ‘contextual safeguarding’ and ‘psychosocial’ support.

In addition, there is a separate adult ED at UCLH which sees patients aged 18 years and over. This is significant, given that Redthread covers patients up to their 25th birthday (11–24 years); thus, the Redthread service at UCLH had presence across two EDs to cater to patients aged 11–24 years.

UCLH is located near Euston and King’s Cross rail stations, both major transport hubs in central London. Staff at UCLH pointed out the significance of the hospital’s location for the population it served: vulnerable young people often present from out of area and it was felt that other local EDs in London might serve a more borough-based patient population (e.g. at Homerton University Hospital in Hackney).

displaced children that would just come from Euston, or from King’s Cross and either be found by the police or just be found, vulnerably, in the street. It just made me think about how can we serve this population better.

Vulnerable young people are sometimes brought to UCLH by emergency services and the police if they are experiencing a severe mental health crisis requiring urgent treatment. The hospital also serves a university population, refugees and young people caught up in drug trafficking and other forms of criminal exploitation. While staff had less experience of serious knife-crime injuries presenting, since these are typically directed to MTCs, soft-tissue injuries and wounds that raised safeguarding concerns about a young person at risk were reported to be common, along with severe mental health issues (such as suicidal ideation and suicide attempts), sexual assault, domestic violence, attacks/fights with peers, and problems related to substance misuse. Some clinical staff saw clear links between urgent clinical presentations among young people and potential gang involvement or sexual exploitation. Other staff felt that gang violence was not the main driver for young people presenting at UCLH:

We see quite a lot of young people that are either on the cusp or involved in gang culture, gang crime.

(Nurse, RED0012)

We have lots of young people coming in with self-harm. Sometimes that will link to violence that’s happening to them or around them. We have a small but clear group of young people who will come in having been assaulted or at risk of gun violence or involved with the gangs, particularly where we’re located.

(Consultant, RED0014)

I think what you’d find is that it’s mostly alcohol, drugs, child mental health.

(Non-clinical staff, RED0010)

When asked whether case presentations had shifted as a result of the COVID-19 pandemic, staff were unanimous in their perception that they were seeing a greater number of young people suffering from severe mental health crises associated with problems within the community or family home (e.g. isolation, unemployment, domestic abuse, housing instability).

Guiding principles and theory of change

As described in Chapter 1, the Redthread intervention follows on from a young person experiencing a traumatic event that has led to a hospital presentation. The ‘classic’ Redthread intervention consists of youth workers engaging with young people aged between 11 and 24 years attending EDs as a victim of violence, assault or exploitation, with youth workers placed directly within hospitals to work ‘shoulder to shoulder’ with clinicians. This model is most familiar with the London MTCs, which receive higher numbers of young people presenting due to violent assault from knife crime. However, as is explained in more detail below, the Redthread service at UCLH was found to employ a number of referral pathways to the service, covering direct referrals from staff and active searching by youth workers for eligible patients.

The Redthread model draws on a range of research and academic influences: behaviour change theory (e.g. smoking cessation research), concepts such as the ‘teachable moment’ and ‘contextual safeguarding’, and frameworks such as the Anna Freud Centre’s AMBIT (adaptive mentalization based integrative treatment) model. 86 According to the Anna Freud website, AMBIT is:

a collaborative project that involves feedback from practitioners, clients and evidence-based treatment designers. It has been designed by and for community teams from mental health, social care, youth work, or that may be purposefully multidisciplinary/multiagency. It emphasises the need to strengthen integration in the complex networks that tend to gather around such clients, minimising the likelihood of an experience of care that is aversive.

Redthread’s youth worker manual describes the ‘teachable moment’ as comprising three essential parts:87

1. dialogue that links a young person’s concerns to the health risk

2. youth worker dialogue that fosters the young person’s motivation to change

3. the young person’s response that shows a willingness to engage and commit to change the specific behaviour(s).

One Redthread interviewee described this concept as going beyond a ‘reachable moment’ to a ‘teachable moment’. This means that youth workers interact with a young person within the hospital setting and involve them in decision-making and learning to empower them to identify where support is available, a task that clinical staff do not have time to do due to other priorities:

The job of the Redthread worker is to help the young person identify who their network of professionals is. And then look to scaffold that teachable moment out to the appropriate key worker, the one that the young person identifies. Don’t assume that you are going to be the key worker for the long term.

(Redthread manager, RED004)

Redthread shared their ‘theory of change and assessment process’ document, which is outlined in Figure 4 (see Appendix 6 for the most recent version). 88 Youth workers discussed Redthread’s role at UCLH as having a rather broad remit, yet ultimately being focused on taking a sensitive and holistic approach to caring for traumatised young people who had come to harm through violence, criminal or sexual exploitation, or who were affected by other safeguarding issues:

FIGURE 4.

Redthread theory of change (2018).

the main aims overarchingly are to contribute to the health, safety and happiness of young people. Obviously, we talk a lot about violence and harm but it actually exists more broadly just to make sure young people are well from a broader perspective as part of this whole public health approach to youth issues and to violence specifically.

(Redthread staff, RED009)

young people who come in really scared, anxious, unsure about what’s happening, and we can be that really comforting and compassionate presence for them in a time of kind of trauma … providing a very compassionate and trauma-informed approach to working with them, advocating for them, and ensuring that they are at the centre of their care, where sometimes things are done to young people rather than with young people.

(Redthread staff, RED003)

Intervention components

Redthread-employed youth workers become embedded within local hospital EDs and work alongside clinicians as part of the Redthread service offer. Youth worker engagement with a young person – the ‘teachable moment’ – can only occur if consent is received from the patient (see below for further details) following their presentation at hospital. The service provides a referral pathway for clinicians in EDs who can identify young people at risk, obtain consent, and refer them to youth workers for engagement and support.

Youth workers work with young people on a case-by-case basis providing tailored support (see the two clinical vignettes below for illustrative examples). The charity also delivers a wider set of services to the hospital once a Redthread team is on site, therefore the intervention has a number of interacting components. Table 4 outlines these components drawing on the qualitative data collected at UCLH. Interviews with Redthread staff, both managers and youth workers, indicated that the service was underpinned by a strong commitment to a voluntary consent model. This was deemed to be significant because Redthread is not a statutory service and young people have a choice to use the service.

Individual consent is therefore sought from young people because the charity’s focus is on empowering young people and putting them at the centre of decision-making about the support they receive and their care in hospital. Redthread’s distance from entities such as social care and the police was believed to be critical because young people often distrust ‘authoritative’ bodies and situations where they sense they ‘have no choice about whether they can engage or not’ (RED009). By being a voluntary service and charity, it was felt that young people could be more open about sharing personal (and relevant clinical or social) information with Redthread youth workers:

we’re trying to make other organisations realise that young people will simply disengage, will be dishonest, if we come in authoritatively and we have to recognise their agency and have to empower them as opposed to force them through procedures that are now archaic.

(RED009)

Indeed, Redthread youth workers attributed high value to ensuring that young people consented to use their service, and that consent was obtained in a language that they would understand. ‘Staged consent’ was a term used to describe the process of obtaining consent for different activities that youth workers undertook as part of their ongoing engagement and support, with one stressing the need for ‘continual negotiation of a young person’s autonomy’ to secure willing participation and, at the same time, recognising that young people could at any time withdraw their consent or disengage.

Redthread staff confirmed that interventions are based on assessment of the needs of an individual, so the duration of support provided to a young person could vary from a few days to a number of months. The case management aspect of the Redthread service was typically an intervention of 6–12 weeks, although longer support was sometimes needed to have ‘meaningful impact’ depending on the specific circumstances and needs of a young person. Youth workers confirmed that trying to get a young person to engage could also take a lot of ‘persistence’ before they could begin to work meaningfully with an individual. Young people might initially decline support then later accept it following a subsequent traumatic incident and reattendance at hospital.

If a young person consents to receiving help from Redthread, youth workers begin by developing a ‘safety plan’ for the individual and understanding their support needs. Youth workers undertake different types of activities within the hospital setting, such as explaining medical terms, liaising between the young person and health professionals, getting a young person food, and generally trying to improve their overall experience of the hospital setting, and allaying any fears and concerns they might have. Support activities are therefore far ranging; for example, safety planning, completing safeguarding referrals, providing emotional support/containment, signposting young people –(or their families) to statutory and non-statutory services (e.g. Victim Support, drug and alcohol services), organising safe transport home, providing food vouchers, helping with clothing, and assisting a young person with accessing follow-up medical treatment/prescriptions.

Two main modes of youth worker engagement were therefore apparent: short-term crisis support and engaged, longer-term support. In the former category, a young person consents to talk with Redthread, such as when they are an inpatient on an adolescent ward or in the ED, yet no further support work takes place following discharge from the hospital. For example, because the young person did not wish to work further with youth workers, was not eligible for the service, or was unable to make an informed decision about the service while in hospital (i.e. due to intoxication). Even if a young person declines initially, if they are eligible for additional support, they will be informed that they can contact Redthread at a later date, should they wish to self-refer to the service. There is therefore a cohort of patients typically provided with short-term assistance in the hospital (e.g. access to food, safety planning, information about other services, help with onward travel) and recorded in the Redthread referral log as having received this type of intervention. Discussions with Redthread staff confirmed their observations that even short-term interactions with young people and short-term crisis support modes of engagement could have a positive impact (e.g. by helping a young person to feel comfortable and assured in the unfamiliar hospital setting and better engage in their care).

The second more ‘engaged’ group become part of Redthread’s formal case management work. This group of young people provide consent to work with a youth worker and a risk assessment is completed, along with a plan for tailored support. Youth workers engage with this cohort over a number of weeks, or even months, and enter information into the Redthread Lamplight system which records youth worker case notes and information regarding which other professionals are working with the young person (e.g. probation, social care). Indeed, Redthread staff emphasised the commitment of the charity to working closely with relevant hospital-based services, such as hospital social workers and safeguarding teams (adult and child), and external services, both statutory and non-statutory, to ensure that continuing support was provided to a young person following hospital discharge. Indeed, part of the set-up phase for the service at each hospital involved scoping work to establish ‘key partners’ and to ensure that Redthread avoided overlapping in ways unhelpful with other ‘jurisdictions’ and voluntary organisations. This meant that Redthread could easily refer young people to other local services where appropriate.

Identifying and referring young people eligible for support from the Redthread service

A relevant finding of this evaluation was that the context for Redthread service adoption at UCLH went beyond a single ED to a cluster of specialised services for children, adolescents and young adults at UCLH, including outpatient departments, fracture clinic and adolescent wards. This has implications for how young people are identified and referred to the Redthread service.

Staff and youth workers discussed how young people present to the hospital with a range of clinical and psychological problems that trigger safeguarding concerns and can indicate that a young person is vulnerable within their community. Examples included young people assaulted with an implement or sharp object (e.g. glass or knife), which they might report as happening ‘accidentally’, substance misuse, suicidal ideation and suicide attempts, sexual assault, and severe mental health crises requiring urgent treatment. A group of vulnerable young people are also reported to experience problems with managing a long-term condition such as diabetes, asthma, or a rare condition sometimes requiring urgent care. Redthread youth workers were therefore actively engaged in multiple processes to identify eligible young people for the service at UCLH, and this went beyond victims of violent injuries from gun or knife attacks. Indeed, hospital staff noted that because the hospital was not a London MTC, knife injuries were less frequent:

our major trauma centres, they get predominantly very violent assaults, whereas I think at local hospitals Redthread are known to receive a lot more varied referrals.

(RED005)

young people who come in with mundane or seemingly mundane presentations around asthma during the course of their care will divulge that the trigger might have been them being chased for example by a gang, which then makes them eligible for us or young people having seizures and then they divulge that it’s stress-related because they’re being targeted by certain groups.

(RED009)

the demographic of patients that present to A&E at UCH [sic], there’s very few knife crime directly related, but there tends to be more kind of either self-harm and that type of presentation, particularly in younger people 17 to 21, 23, 24.

(Doctor, RED0022)

At UCLH, staff suggested there was a cohort of young people, particularly those aged 18–24 years, who present ‘with less physically traumatic injuries’ (Redthread youth worker, RED009) and from ‘assaults without a weapon’ (Redthread staff, RED0020). These young people are less likely to be admitted to hospital for longer periods of time, are often seen outside of daytime hours, and can be especially difficult to follow up due to different processes of consent for those aged under 18 years and those aged over 18 years. For example, young adults over 18 years might sometimes give a false name or contact number or leave without providing any contact details, in which case, Redthread engagement and consent cannot happen (‘the pre-engagement stage’) within the hospital setting. Other young people were reported to engage more easily on site, particularly in the paediatric ED, emphasising the importance of Redthread youth workers having high visibility across EDs:

If the young person is in the A&E and we have a chance to see them face-to-face we don’t have to have consent per se to speak to them, but what we will ask the clinician to say is, ‘I’ve got a youth worker here do you mind if they come and talk to you?’, just so that they’re not a professional walking in and they have no idea who we are.

(RED003)

Clinicians close to the service with direct experience of engaging with Redthread youth workers, or making a referral to the service, also recognised the importance of timely identification of young people eligible for support by Redthread, especially once aged 18 years and over and beyond child safeguarding protections:

once they cross that magic line of age, if they are not consented in real time, it’s very hard to reach them if they don’t want to be reached, or if they are difficult to reach, which they often are.

(UCLH Clinician 1)

There were different referral processes that Redthread instigated at UCLH to ensure that no young person eligible for their service was missed (see Chapter 5). One important activity in this regard for Redthread was screening the electronic hospital electronic record (Epic) for admissions among 18- to 24-year-olds in the last seven days (e.g. from violent or sexual assault) since this group was most likely to be missed. In addition, youth workers attended MDT meetings (e.g. safeguarding) to identify young people and received direct referrals from staff. Youth workers also monitored patient admissions in real time (e.g. ED ‘track boards’) to identify young people that they could approach before they were discharged and who might respond to a ‘teachable moment’.

Redthread staff observed that while most referrals they received were from the paediatrics ED, their proactive screening of Epic suggested there was an equal proportion of patients eligible for the service across paediatrics and adult EDs, and that the ‘missed referrals are almost entirely concentrated in people aged 18 and over’ (Redthread staff, RED0020).

Interviews confirmed that the weekly paediatric ED safeguarding meetings at UCLH had become a central point of referral to the service with youth workers now fully embedded within them. This had been especially useful for the identification of young people prior to the integration of a Redthread referral form into Epic, enabling clinicians to make direct referrals via the hospital system. Redthread staff spoke of the need to ‘diversify referral pathways’ during early implementation of the service at UCLH due to impact of COVID-19 and the redirection of UCLH patients to other hospital sites (i.e. the Whittington and Great Ormond Street Children’s Hospital). This had resulted in both the closure of the paediatric ED, and led to discussions with outpatient and other specialist departments (e.g. rheumatology, diabetes) about staff referring young people to Redthread as an additional route. As one youth worker observed, ‘that left a legacy for our service’ and ‘represented a bit of a pivot’ in the type of patients they were now engaging with.

Use of data and service monitoring

To work optimally in the hospital setting, youth workers required an honorary contract to access Epic. Access to this system was introduced early into implementation of the service and enabled Redthread staff to proactively search for young people eligible for youth worker support who may been missed, and to log youth worker interactions with young people (e.g. at bedside) in the electronic health record. Epic further enabled direct staff referrals to the service through an integrated form. Youth workers would then liaise with clinical staff responsible for the young person’s care, and the clinician who had made the referral, and confirm whether it was appropriate to engage and offer this patient the Redthread service:

We had a young person who was stabbed with a bottle last week and they absconded before they were seen. They attended with police, I don’t know how they managed to slip away … So that’s why having access to the [Epic hospital electronic health record] system is good because you can see if there are any follow-up appointments and we’ll put it in our calendar and visit the hospital on that day because we know that if we see them in person it increases the likelihood of engagement.

(RED009)

Staff involved with safeguarding and monitoring patient records appeared most likely to make referrals via the Redthread referral form integrated with Epic. This was also the case for clinicians providing overnight care when youth workers were not available on site.

Redthread staff use a referral log template to capture information about young people referred to the service at UCLH, including who is not eligible for or declines support. This log (which is only available to view by Redthread staff) contains details such as the name of patient, their hospital number, sociodemographic details (e.g. age, gender and ethnicity), the arrival date and time, referral type, reason for hospital presentation and, where applicable, notes about the incident and any previous attendances for assault. The log is used to record who has made the referral (e.g. clinical staff) and whether there is consent for Redthread to contact the young person. Finally, the log details the type of support engaged and consented young people receive.

Data are only transferred to Redthread’s Lamplight system, which researchers/evaluators do not have permission to access or use, if a person is engaged for longer-term support and risk assessed. This system allows for the recording of more person-level sensitive information, such as case notes and the names of key statutory professionals in the young person’s network. The case management tool is used to record and monitor the allocation of young people to youth workers, ongoing contact, the duration of the support provided and date of case closure.

Redthread staff described a number of issues regarding data capture, flows and processes in interviews:

1. Missing useful information in Epic, the hospital record system, to flag when a young person is eligible for support, meaning young people can be missed (e.g. clinicians too busy to provide detail in the electronic health record, or unsure about the type of information that would be most useful to record, e.g. risk indicators).

2. The Redthread referral form within the Epic form being under-used (as mentioned, staff tended to make referral to Redthread directly via in-person contact, phone or email).

3. On the other hand, youth workers entering too much detail in the hospital record that could potentially lead to changes in how the individual is perceived and/or treated by clinicians (e.g. if the young person is involved in gang violence or has been a victim of sexual exploitation). Youth workers were especially concerned about notes by social care services and the language used to describe a young person and their needs.

4. A lack of information sharing between hospitals to identify where young people have been admitted before for violent injury or assault.

As one member of Redthread staff observed: ‘it would also just be incredibly helpful if there was essentially a button when anybody was logging a patient if they could just say this person is eligible for Redthread’ (Redthread staff, RED0020).

The Health Information Exchange portal was deemed to be useful because youth workers could review discharge summaries from other participating NHS trusts, and UCLH’s own Epic system was therefore found to be useful by staff for flagging the number of times a person had attended the hospital within a given period. Overall, however, issues remained during the evaluation regarding the lack of formal and systematic recording of Redthread referrals and markers in Epic to flag young people potentially eligible for the service. For more information about data processes and flows, please see Chapter 5.

Impact of COVID-19 and redirection of services COVID-19 service-level impacts early implementation (February 2020 to March 2021)

The Redthread service was introduced to UCLH in January 2020, shortly before the COVID-19 pandemic started and severely affected health systems and societies internationally. This led to the four main impacts in terms of implementing the Redthread service at UCLH, as outlined below:

1. Paediatric patients of UCLH were temporarily redirected to the Whittington Hospital (a provisional arrangement until March 2021) and referral processes were regularly reviewed by UCLH and neighbouring NHS trusts.

2. Redthread youth workers adapted to provide a virtual youth work service offer during the height of the pandemic and were not physically back ‘on site’ at UCLH until August 2020, and worked remotely for an interim period. Youth workers returned to fully working on site in April 2021. The service therefore continued to operate as usual, but for certain periods in a remote capacity. Youth workers received referrals from UCLH staff, including live referrals presenting within ED. Redthread youth workers also continued to work with young people once they were discharged from hospital.

3. Clinical presentations: UCLH staff observed that there were fewer cases of violent injury during lockdowns, yet more severe mental health and other types of safeguarding concerns presenting among young people during the pandemic (e.g. domestic violence and sexual assault).

4. There was significant disruption of Redthread’s launch and staff-awareness activities because youth workers and paediatric services were not on site and many staff at UCLH were focused on treating COVID-19 patients.

Interview respondents, both Redthread and UCLH staff, highlighted that the pandemic had not only influenced the types of clinical cases presenting at UCLH, it had also severely disrupted the service offer. Nevertheless, youth workers perceived that a remote support offer had been well received by many young people who were comfortable interacting via mobile phones: ‘we’ve probably engaged some young people that wouldn’t have engaged at bedside because remote conversation is better for them’ (Redthread staff, RED003).

Staff awareness and perceptions of the Redthread intervention

Several clinical staff interviewed had prior awareness of the Redthread service from its implementation at other London hospitals and MTCs. Staff viewed the charity as a youth violence service targeting young people caught up in gangs.

Staff spoke of Redthread youth workers as having a complementary and beneficial presence for both clinicians and young people, especially when young people were suddenly confronted with an unfamiliar and daunting hospital environment. The service was seen to align well with UCLH’s specialist services (e.g. adolescent wards and paediatric ED), with the result that youth workers had been enthusiastically welcomed and staff reported being ‘excited’ and ‘passionate’ about Redthread. The quotes below illustrate the value respondents placed on the service locally. In particular, staff praised the comforting and neutral presence of youth workers when working with young people and their ability to provide support to young people following their discharge from hospital:

I think it’s reassuring for the nursing staff if you’ve got a young person that’s obviously really unhappy and they can have somebody that’s there that can chat with them on their level.

(RED0012)

In terms of the intervention that they’ve been able to provide through our young people, it’s been absolutely invaluable. I think they essentially deliver a very assertive youth working intervention. They will meet young people on the ward, they will follow them up into the community… They tend to be very good at engaging young people who, which traditional services such as CAMHS or social care, often find it quite hard to reach.

(Consultant, RED0021)

that one kind of familiar face is very worthwhile.

(Doctor, RED0022)

it’s massively strengthened our offer of holistic care to young people … I think it has been a really instrumental part of facilitating that care. Particularly in those presenting in the greatest distress.

(Consultant, RED0015)

it’s helpful to share the burden around ensuring the safe plan for when young people are discharged. I think the team, they’re really positive to have around.

(RED0011)

Hospital consultant specialists (e.g. paediatric, adolescent, psychiatry) were particularly clear about Redthread youth workers filling a ‘gap’ as a non-statutory organisation able to connect with internal and external services to assist a young person, as well as hospital safeguarding processes:

the potential safeguarding connections and networking and linking in around trying to prevent someone at risk of violence, let alone someone who’s actually already experienced violence, I think puts Redthread in this meta position.

(RED0014, Consultant)

I would quite happily have a youth worker for every young person we saw in the rust, quite frankly.

(UCLH Clinician 2)

Staff were eager to refer young people to the Redthread service and would usually do so during the day via telephone, email or by approaching youth workers directly. Within paediatrics and adolescent services, attempts had been made to embed Redthread youth workers as part of different MDTs, rather than simply have the charity ‘co-located’ without proper involvement with standard care processes. Respondents were, however, aware that the UCLH’s dedicated adolescent services were unique and that staff are very ‘adolescent focused’ and receptive to youth workers, meaning the paediatrics ED was a highly conducive context for the Redthread service as a whole. By contrast, it was observed by both clinical staff and Redthread youth workers that embedding the service in adult emergency services covering the cohort aged 18–24 years had proven far more challenging:

the adult A&E is really very separate. That’s not to say that there aren’t fantastic medics and nurses around there, but I would say that as a paediatrician, I’ve got quite limited influence over them. I’m not really sure how much they know about it, how much they engage with it.

(Consultant, RED0015)

I think if you were predominantly interested in adult A&E, I think it’s probably a service that wouldn’t come to mind.

(Doctor, RED0022)

those of us who work in children and adolescents we got it much earlier on that it was a really good preventative way of working with young people.

(RED0010)

Staff also struggled to say whether Redthread had resulted in specific clinical or quantifiable outcomes and questioned whether hospital reattendances were an appropriate measure of the impact of the Redthread service. Staff discussed general positive benefits and impacts based on their observations of youth worker interactions with individuals and youth worker involvement within a multidisciplinary approach towards caring for adolescents and young people at risk of exploitation, gang crime and the most vulnerable within their local communities (see anonymised clinical case vignettes (Box 2 and Box 3) for examples).

Some UCLH staff wanted to receive more feedback from Redthread about what happened to supported young people given that reattendance for knife injury, for example, was not necessarily a useful indicator about their impact:

I want to hear some narratives about how the work had shifted something for a young person. To hear their story and what they’re doing differently, because I think that can be much more powerful. The readmission might be a good thing, you see. It isn’t necessarily a bad thing.

(RED0014, Consultant)

Enablers and barriers

The qualitative data were used to identify the organisational factors, processes, resources, and staff training necessary for the successful implementation and delivery of the Redthread service. Table 5 lists the enablers and barriers that were raised in interviews with both Redthread youth workers and UCLH staff with regards to service implementation.

Statistical analysis

The mean score for each risk dimension was reported separately and statistical tests of the differences in risks ‘before’ and ‘after/during’ the follow-up were carried out using McNemar–Bowker and paired Student’s t-test. McNemar–Bowker tests were then conducted within subject to look at young people’s feelings of safety at the initial risk assessment compared with the ‘after’ risk assessment. Paired samples t-test were conducted to identify whether there was any statistically significant change in safety ratings pre- and post-intervention.

Consequences

The information on the initial (before) risk assessment for young people using YVIP for the period April 2020 to November 2021 is shown in Figure 7 (and in Appendix 5, Table 13). Young people were identified as facing a number of potential risks, particularly in relation to experiencing and participating in further harm, experiencing and participating in criminal behaviour, engaging in education, employment or training, not willing to engage with other services and not maintaining positive relationships with family.

FIGURE 7.

Risk factors in the initial risk assessment for young people who used the Redthread service in the period April 2020 to November 2021.

Figure 7 shows that, among all the assessed risks, the higher proportions of young people were identified as ‘high risk’ for (1) not maintaining positive relationships with family (22%) and, (2) not engaging in education, training or employment (19%). Figure 8 shows the range of assessed risks post-engagement. Broadly, the proportion of young people assessed to be facing high risks decreased or remained the same for all categories except for the risk of not willing to engage with other services, which rose from 2.8% to 5.6%.

FIGURE 8.

Risk factors in the end risk assessment for young beneficiaries who used the Redthread service in the period April 2020 to November 2021.

McNemar–Bowker tests of symmetry were then carried out to identify whether there were significant changes in risk factor ratings from the initial assessment to the end assessment (Table 9). The test revealed that the reduction in risks pre- and post-intervention were statistically significant for just two of the risk categories: risk of experiencing further harm (χ²(3) = 8.40, p = 0.0384) as well as for risk of young people not maintaining positive relationships with family (χ²(3) = 5.77, p = 0.050). The results show that the majority of the young people remained in the low-risk category and that changes from high to medium risk were more usual than remaining in the high-risk category.

A paired samples t-test was also conducted to identify whether there was a significant change in the overall score on how safe the young person felt right now. While there was a slight improvement on this overall measure (before: M = 7.11, standard deviation 2.18; after: M = 7.77, standard deviation 1.90), this change was not statistically significant.

Costs

Table 10 provides information on the mean costs for those engaging with the Redthread programme. Those engaged in a longer-term programme and short-term crisis supported young people combined and the mean costs per finished consultant episode for a group of patients matched to two characteristics of potential Redthread users: age (11–24 years) and HRG codes for inpatient stays selected on the basis of a range of possible treatments actual Redthread users may undergo at UCLH. ED costs were based on those reported by UCLH for 11- to 24-year-olds across all 12 HRG codes collected for EDs.

The mean costs per young person (n = 169) for the Redthread intervention were calculated using information provided by Redthread for the 21-month period April 2020 to December 2021. Consumer price index inflation rate for 2020–21 was applied to 2020 prices. 102 All the resources used by Redthread were assumed to be spent on delivering the YVIP services for all the four activities (front line, service development, training to the medical staff and administration). All the costs are reported separately between these activities based on time Redthread staff spent on each activity (provided by Redthread).

Table 10 shows that the main cost for the Redthread service was for front-line services (50% of total costs), followed by service development (26%), administration (14%) and training activities (10%). The total mean cost per supported young person was £1865. Table 10 also provides the mean costs for the constructed group of actual patients treated at UCLH. For the inpatient costs, based on a set of diagnosis codes relevant to Redthread referral reasons (see Appendix 3, Table 12), between February 2020 and August 2021, 161 11- to 24-year-olds were anonymously identified as having a nonelective admission at UCLH. UCLH’s patient level information and costing system data were used to specify unit costs, which were then applied to each of the 161 patient spells via the spell HRG code. In practice, this reduced the sample to 119, as 42 had a ‘null/unknown’ HRG code. Based on this sample, the mean cost per patient was £5789 (range £473–73,837). The mean cost for an ED attendance was £203 (range £82–455) and was based on an average number of attendances of patients aged 11–24years of 19,501 per year across all types of coded attendance.

Principal findings

The analysis of assessed risks for young people engaged with the Redthread YVIP for the period April 2020 to November 2021 showed that the only risks for which we observed a statistically significant decrease after the intervention were the risk of ‘experiencing further harm’ and the risk of ‘not maintaining positive relationships with family’. Our results may be on the conservative side when compared with other studies evaluating YVIP in London. For example, one has reported that improvements in 6 of the 13 risk factors were statistically significant. 101

The average Redthread costs for Redthread cases was estimated to be £1865. However, following the reasons described in Chapters 5 and 7, it was not readily possible to link Redthread cases with their corresponding hospital records to track activity and, hence, their actual hospital costs. The costs of hospital treatment were therefore based on an artificial group that mirrored as best as possible Redthread cases in terms of age and diagnosis groups.

The mean cost for the emergency inpatient treatment of an artificially constructed group, similar to those likely to be referred to Redthread in UCLH were as estimated to be £5789, while the mean cost per attendance at the ED was £203. The estimated average costs of Redthread YVIP are around one-third of the costs of inpatient treatment for a similar group and around nine times larger than the mean ED attendance cost. Based on the 169 young people involved in Redthread as part of this analysis, and total Redthread costs of £315,197, this suggests that the Redthread programme would break even if, for every three people engaging with the service, at least one emergency inpatient admission was avoided.

Another important economic aspect related to programmes targeting youth violence is that some, and possibly the most significant part, of the costs and benefits materialise outside of health and social care. These include costs associated with the criminal justice system, as well as education, the welfare system in general and costs associated with disability or lost productivity as opposed to the generation of income from employment and tax. Other studies that have tried to quantify the costs of crime have estimated that the total unit costs from homicide amount to more than £3.2 million, for violence with injury £14,100, rape £39,400 and other sexual offences £6500. 103 Health-care services have been estimated to be only a small fraction of the total costs (especially in relation to lifetime effects of physical and emotional harm) amounting to just 0.03% of total costs for homicide or 3–6% for crimes with injuries, rape or sexual offences. 103

Overall, our evaluation showed that the Redthread YVIP intervention has the potential to reduce some risks of harm for young people and to be potentially self-financing if the intervention could avoid future use of health care. However, our research was limited by the small sample of patients, lack of a control group, inability to evaluate effectiveness and the subjective assessment of risks based on Redthread personnel (see also the limitations of the study).

Comparison with other studies

Global research on violence has identified the burden and costs that violence places on health and social prospects across the life course. 104 Violence affects not only health outcomes, quality of life and social choices that people make, but also has wider consequences on families, communities and society. 105 The acute treatment costs of penetrating trauma injury in England and Wales vary by patient, cause and body region of injury and treatment characteristics. The reported median cost of index admission for 532 patients aged 16+ years admitted to a MTC following knife-related injuries in a major UK city in 2018 was £4375. 95 This was lower than the costs reported previously for England and Wales, which were estimated at £7983 in 2008 by a Trauma Audit Research Network study, which focused on more severely injured patients and also included cost of prehospital care. 106 An earlier study, reviewing the cost of 187 gunshot injuries treated at a teaching hospital in Manchester, reported a mean cost of £2698 for patients admitted for inpatient care. 107 A study by Christensen et al. (2008),106 which examined the acute treatment costs of penetrating trauma injury in England and Wales, estimated average hospital costs of £7196 per penetrating injury due to stabbing. 106 In most of the studies, the ED costs are not identified separately from the total health-care costs. A study conducted in South Wales estimated that about 30.0% of assaults resulted in a visit to ED at a cost of £200 per attendance in 2019. 105

Our findings are consistent with other studies in England and Wales, showing that inpatient health-care costs are considerable and much higher than the costs of youth violence interventions. In fact, a 2020 study of the service, which looked at both health and social costs and benefits of Redthread’s YVIP,108 concluded that the programme may have a positive net economic and social benefit that was calculated at up to £4.90 economic and social benefit per £1 spent.

Strengths and limitations

While this study provides some insights on the main costs and risk reductions resulting from Redthread YVIP, it also has limitations which need to be taken into account. One limitation was that the information on costs from Redthread could not be directly linked with hospital data, and therefore we had to rely on an artificially constructed group of potentially similar patients to provide indicative, plausible inpatient costs for young people using the Redthread services. A second limitation was that findings from our analysis were based on a relatively small group of individuals (an impact of COVID-19 on hospital services) and were also limited from the lack of a comparable group with assault-related injuries.

The analysis of consequences was based on Redthread’s own risk assessment process. The answers to the risk assessment involved professional judgements of youth workers and their interpretation and therefore should be considered in this context.

Implications

Our analysis showed that young people attending Redthread YVIP report lower risk of experiencing further harm and a lower risk of not maintaining positive relationships with their family after completing the intervention programme. The cost of delivering the Rethread intervention per young person is considerably lower than the estimated inpatient costs for a comparable group.

Future studies need to collect detailed information on the costs of intervention, costs of the health-care treatment as well as the wider cost for society. Linking the data gathered on the young people attending the Redthread YVIP with hospital data and identifying a suitable control group will help to complete a full economic evaluation of the Redthread YVIP.

Period of evaluation

The first year of data collection coincided with the COVID-19 lockdown and the service provided by Redthread was different in nature to the service that was implemented from April 2021 (see Chapter 1), and not one that they considered to be a model for an ongoing service. From an evaluation perspective, if we chose to focus on the more consistent service from April 2021, the numbers would be too few to observe any impact. If we included data from the previous year, there would be a question about the consistency of the service being evaluated.

Limitations of the emergency care dataset

The HES and ECDS to which we had access over the evaluation period could not be linked because the ECDS had no individual patient identifiers. This also meant that multiple attendances at ED for the same individual could not be identified.

Defining an appropriate measure of hospital attendance

Whichever option is chosen, identifying appropriate hospital attendances from the routine data is important, whether for measuring outcomes, past attendance histories or, in some options, selecting cases and controls. Ideally, relevant attendances would include those that are related to a person’s eligibility for YVIPs and which are potentially avoidable. 109 Although we can establish a list of codes for identifying relevant activity from routine hospital data, this can only be an approximation and, historically, UCLH has not coded ‘injury intent’ within the ‘chief complaint’ field in their ECDS. This indicates that an injury was sustained from an adverse event (as opposed to an accident) and is important for young people who are more likely to be exposed to violence and abuse. The completeness of this field is being addressed within the hospital.

One UK study of routine hospital inpatient data found that recurrent admissions were more common in young people with adversity-related injury than those with accidental injuries. 110 While it would be possible to identify these injuries in many cases within the admitted patient HES data, the available information in ECDS would make it harder for this distinction to be made. Also, the chances of detecting an impact will depend on adequate specificity (i.e. the chances that unavoidable reattendances are not selected). Possible codes that can be used to identify such activity within HES inpatient data and ECDS are shown in Appendix 3, Table 12.

Data requirements

Cases would be identified from Redthread’s database and their records linked to routine hospital data. Reattendances would be extracted from links to national inpatient and emergency care data sets, namely HES and the ECDS for the ED. Identifying reattendances at other hospitals would be particularly important for UCLH services, where we estimate that of people aged 11–24 years who attend their ED, nearly half (47%) of their subsequent visits to ED (where they have one) will be at another hospital (see Appendix 2 and Figure 12). A link to national hospital data would also be important to enable us to assess past attendance histories, which, again, could be at other hospitals, and is likely to be an important matching variable. Linkage to HES and ECDS would need to be approved and carried out by NHS Digital.

FIGURE 12.

Proportion of subsequent ED reattendances by English acute trust between April 2018 and March 2020 following an initial ED visit to UCLH over the same period. Only the first reattendance for each individual is included.

Redthread does collect its own data on reattendances up to two years after someone starts engaging with the programme. Staff gather these data retrospectively each month by identifying anyone who became engaged with Redthread in the same month one or two years previously and then searching the hospital records for any reattendances since. These reattendances are classified by Redthread as adversity or nonadversity related but are only picked up if they occur at UCLH.

An alternative means of collecting data might be to ask consenting individuals to report their own data, although this would rely on sufficient numbers consenting and maintaining reliable data returns.

Consent

Redthread’s typical consent process with the service user does not include sharing personal information beyond their own organisation (even if it is anonymised or pseudonymised). Therefore, sharing of individual user records from Redthread would not be possible, unless Redthread collect retrospective consent from young people, which is unlikely to get a high response rate, or we apply to use Section 251 of the NHS Act 2006. Section 251 involves the use of confidential patient information without the need for patient consent, subject to the approval of the confidentiality advisory group, and this would take several months. In addition, Redthread do not hold NHS numbers on their records, and this is a required identifier used by NHS Digital to carry out linkage.

Controls or comparators

The comparator group could be a cohort of patients attending hospitals in London without a similar violence prevention scheme, whose characteristics suggest that they might be eligible for, and likely to engage with, Redthread services. These patients would need to be identified from routine administrative hospital data. Although we could match on patient characteristics and reasons for hospital attendance, there are several noncoded features that could be highly relevant to the reason why the young person has been considered eligible for Redthread, which would make it difficult to establish the accuracy of any matching.

An alternative control group could be people who are referred to the Redthread service but do not engage with it. Again, there could be key differences in person characteristics that are not identifiable in the data and may introduce important biases, but if referrals are coded within the routine hospital data this may be a more achievable option.

Chances of detecting an impact

Based on evidence of the potential impact of similar services, we estimate that 110 people who engaged with the service would need to have been followed for six months to have an 80% chance of detecting a 10% reduction in reattendance. With longer follow-up and a greater assumed impact, the sample size could be smaller. Further details of this analysis, including projected sample sizes under different assumptions, are shown in Appendix 4 and Figure 13. From February 2020 to December 2021, a total of 59 young people engaged with the full longer-term programme rather than shorter-term crisis support; with an engagement rate of about five per month, we estimated that a total of 110 would be reached by the end of October 2022. If the aim were to evaluate only the service that had been operating since April 2021, then it would take somewhat longer. These calculations were for measuring the impact on those engaged in the longer-term programme in terms of hospital reattendance. If we included people receiving shorter-term crisis support, then it might take a shorter time to reach the required cohort size, although the required cohort size could be larger if the assumed impact of such support is less.

FIGURE 13.

Estimated sample sizes required under different assumptions of baseline reattendance rates and the effect of the Redthread intervention to achieve 80% power.

Summary of advantages and disadvantages of this option

This option is the only one that follows up cohorts of matched cases and controls and may be the most robust design of the options being considered, also offering most statistical power. However, being able to identify an appropriate control group remains an important barrier and the approach would have to accommodate the consenting process. Moreover, it does not account for the positive impact Redthread service users might have on others in the community. For example, by indirectly influencing a reduction in the numbers of hospital attendances of others.

Data requirements

Referrals cannot be reliably identified from the hospital’s own data unless referrals are better coded within the local hospital records, for example, by using a SNOMED code for violence reduction services, such as SCTID: 410236002, Violence control education (procedure), (see Chapter 5). There is a process for using such codes within UCLH but this was not applied in at least 85% of cases. If such codes were reported in the national ECDS there would be no need to link local to national records, an analysis could be carried out with ECDS and inpatient HES data. However, this relies on accurate clinical coding and may miss patients who are not referred through ED or are picked up through Redthread’s own data searches.

Consent

Once a referral is coded, all analysis would be based on routine hospital records, although the importance of consent is unclear, given that not all referrals will be engaging with the Redthread service. As a service evaluation conducted by UCLH with the intention of improving service delivery there may be no need for consent, although ethics approvals would still be required.

Controls or comparators

For this option, the control group would have to come from other hospitals, with associated challenges as described under option 1a.

Size of cohort

Since not all the cohort would be receiving the intervention, a larger sample would be needed to achieve appropriate statistical power. Also, if the referral code is only applied to ED records, people referred from departments other than ED may not be picked up.

Summary of advantages and disadvantages of this option

This option may circumvent the need to gain consent from people engaging with Redthread, since Redthread’s own data would not be used, but this is currently unclear. The other issues associated with option 1a around selecting appropriate controls and measuring benefits within a community remain. Also, people not referred through ED may be missed. Between February 2020 and December 2021, 43% of referrals were engaged with the service either in the longer-term programme of work (15%) or shorter-term crisis support (28%) so any impact of the intervention over that period would only be observed among a minority of cases and would therefore be harder to detect. However, since after the COVID-19 restrictions the service has been operating more as intended, engagement rates are expected to be higher which raises the power of this approach.

Data requirements

Data would all come from routine hospital records.

Selecting cases

There would be a major challenge in identifying who could be eligible for Redthread from the routine data. As mentioned above, when discussing the problems with identifying control groups under option 1a, there are several noncoded features that could be highly relevant to the reason why the young person has been considered eligible for Redthread. For example, it may not be obvious that a young person presenting at hospital is at risk, but a clinician may have reason to suspect they are a victim of an adverse event and would benefit from Redthread support.

Based on the list of clinical codes we identified within the ECDS, (see Appendix 3 and Table 12) we found 3275 patients aged 11–24 years who may have been eligible for Redthread attending ED between February 2020 and August 2021. As a comparison, over the same period, far fewer young people were referred to Redthread: 252 in all, of whom 43 became engaged in a longer-term programme with the service. There may be ways of improving the criteria for selecting potentially eligible people with more detailed analyses of hospital data but, as mentioned above, the data for UCLH are incomplete for some of the key fields that would help.

Consent

Since we would not know which of the cases actually engage with Redthread, the consenting issue for Redthread users would not be an issue.

Controls or comparators

Control groups derived from routine hospital data would be a better match than under options 1a and 1b, since the same fields would be used to identify both cases and controls. There would also be no data linkage required between information held on Redthread’s system and routine hospital data.

Size of cohort

Because a potentially eligible cohort derived in this way is so much larger than numbers referred to or engaging with Redthread, analysis of that cohort would be very likely to miss any direct impact of Redthread. This analysis is better viewed as an assessment of the service on the whole cohort of eligible young people attending UCLH.

Summary of advantages and disadvantages of this option

This option circumvents the need to gain consent from people engaging with Redthread, since Redthread’s own data would not be used. Also, it would be possible to obtain a better-matched control group. However, the issue associated with option 1a around selecting appropriate controls and measuring benefits within a community remains. Unless it becomes possible to identify eligible young people from the routine data with a reasonable degree of accuracy, any impact will only be observed among a minority of cases and therefore harder to detect.

Data requirements

Data would all come from routine national datasets.

Consent

This analysis would only use routine data sets and would not be able to identify people engaging with Redthread, so no further consent would be needed.

Controls or comparators

Comparator hospitals would not be operating a YVIP over the same period. Ideally, these would be within London but, given the increasing prevalence of these programmes, it may be challenging to find such hospitals. However, since this is a before–after comparison, we could pick hospitals where they have such a service but where its use over the study period has remained stable.

Size of cohort

As mentioned under option 1c, we found 3275 patients aged 11–24 years who may have been eligible for Redthread attending the ED at UCLH between February 2020 and August 2021. However, this analysis relies on being able to identify groups of patients attending emergency services that would be specific enough for the evaluators to notice any impact Redthread might be having. This is unlikely to be achieved unless better eligibility criteria can be derived from the routine data.

Summary of advantages and disadvantages of this option

As with options 1a, 1b and 1c, if Redthread users become less likely to be involved with violence themselves, they would be indirectly influencing wider numbers of admissions, which would not be measured using this approach. Also, as with option 1c, unless it would be possible to identify eligible young people from the routine data with a reasonable degree of accuracy, any impact will only be observed among a minority of cases and therefore harder to detect.

As young people could be attending several different EDs in London, an analysis that only focuses on one hospital would lose any impact of Redthread services on attendances elsewhere. As mentioned above, just under half of reattendances to ED at English NHS hospitals following a visit to the UCLH department are at another hospital.

Supporting evidence for this approach

A previous literature review included the finding that violence in London is highly geographically concentrated, and a London-wide picture can mask significant variation between areas. 20 This report also confirms that many forms of violent crime are heavily clustered in a small proportion of lower super-output areas, so when responding to the problem of high instances of violence, a localised approach is valuable. If the Redthread intervention reaches individuals who are involved in gang-related violence, these areas may see an overall reduction in hospital presentations.

Data requirements

Areas where people engaging with Redthread live could be provided by Redthread themselves, thereby avoiding any need to link to hospital records. Other areas with similar characteristics where no one is engaging with Redthread should also be selected. Data on hospital attendance would come from HES and ECDS as before, although we would need to identify a suitable area of residence. Neighbourhoods could be defined as middle super-output areas (MSOA), census areas of mean population size of around 7200, which are recorded by Redthread. However, even though a MSOA field is available in ECDS, it is not currently populated. Other area fields are populated, however, such as electoral ward, but they would need to be used consistently across datasets.

Consent

Redthread would need to provide information about where their service users lived, which would be the only data required from them. Although there would be no explicit means to identify Redthread users in the hospital data, knowledge about UCLH attendance and area of residence might be enough to identify them in routine data. Where there were sufficiently high numbers within an area, aggregated numbers would be sufficient. Where there were small numbers, it might be possible to exclude those areas, given that the local impact might be small anyway, although the number of areas that are thus removed would need to be low in relation to those that are included.

Controls or comparators

Because this analysis is investigating relationships between levels of engagement in an area to outcomes there is no need for a comparator.

Likelihood of detecting any impact

Earlier in the project, we received data from Redthread that showed that of 28 people who were engaged in a longer-term programme, they were resident in a total of at least 16 MSOAs, and due to low numbers, some MSOAs had to be combined. This suggests that they may be too dispersed among different MSOAs and too low in number in comparison to relevant hospital attendance activity within the area, thus reducing the ability of such an analysis to detect Redthread’s impact. Since the MSOA field has not been populated in the national emergency care data, this is difficult to determine. However, we recommend improved coding or using areas that are recorded, such as electoral ward, so that the feasibility of an area-level option can be tested further.

Summary of advantages and disadvantages of this option

This option would not need to explicitly identify control groups and is the only option that takes account of any wider positive influence engagement that Redthread may have on local communities.

This approach assumes that any equivalent service at other hospitals remains stable for both before and after periods, so that any differences in hospital activity can mostly be attributed to the Redthread intervention at UCLH. To mitigate any problems, we could restrict to areas where this is more likely to be the case – so that the dominant change in service is that at UCLH.

Evidence reviews and current evaluation evidence (Chapter 3)

1. Available evidence indicates that young people who present in EDs from physical violent injury (e.g. arising from gun and knife crime) are at significant risk of repeat injury. Moreover, children and young people are vulnerable to a variety of risks in the community and can therefore re-present to EDs due to different forms of harm: physical assault interpersonal violence, substance misuse and severe mental health problems related to safeguarding issues.

2. To address young people’s exposure to risk and safeguarding harms, hospitals have introduced youth violence interventions within EDs and trauma centres that are focused on behavioural change. Hospital-based YVIPs are one approach and aim to reduce a young person’s overall risk of harm and reinjury. Most of the evidence base on these interventions comes from the United States. These programmes typically focus on gun injuries in the American health system context with injury recidivism tracked as a primary outcome, using hospital data.

3. Evidence suggests that YVIP can reduce reinjury. However, a problem with assessing the evidence base is the heterogeneity found across youth-based hospital interventions as a whole in terms of the nature of the intervention and how outcomes are reported. Some are brief behavioural interventions, some involve hospital social workers as opposed to youth workers, programmes can target different age ranges (e.g. young adults up to 30 years of age in some cases) and studies can include relatively small sample sizes.

4. In the context of high health treatment costs in the United States for gun injury, the evidence suggests youth violence interventions can be cost-effective where they help to prevent violent incidences.

5. The literature also highlights ongoing issues with data sharing across hospitals, meaning that measuring outcomes for repeat injury across different geographies and health providers, or systems, remains difficult.

6. Our review confirmed there is limited evidence from the UK health system, in particular about the impact of youth violence and youth worker hospital interventions focused on prevention.

Assessment of programme theory and implementation at UCLH (Chapter 4)

1. Within UCLH’s adolescent and paediatric services, the Redthread service is viewed positively and as filling a gap in service provision. Redthread’s status as an independent, non-statutory service was viewed as a factor which enabled vulnerable young people to share information with a youth worker. This information was often useful to clinicians and could help inform patient care. A youth worker could also advocate for young people within the hospital setting, and bridge services within the community (e.g. housing, specialist support services, social care and education). This underscores the importance to front-line clinicians of having stronger links with the community to support vulnerable young people following their discharge from hospital, especially those for whom child safeguarding protocols no longer apply (e.g. those aged over 18 years).

2. Paediatricians, child safeguarding experts and child and adolescent psychiatrists are particular champions of the Redthread service, perceiving that it not only fulfils a need for young people not met by standard clinical care, but that youth workers can help a young person to better engage in their medical care and treatment. The service is not viewed as duplicating other services or roles at UCLH (e.g. hospital social workers, playworkers or child safeguarding).

3. In practice, Redthread’s service goes beyond the ‘teachable moment’ (which provides an opportunity for dialogue with a young person about their health risk and personal motivation to introduce behavioural changes in their life). For example, Redthread has been engaged in a range of activities at UCLH, including:

   1. long-term case management work (e.g. this could be up to six months, or more, engagement with a youth worker). This is intensive support for young people at highest risk of harm, commonly with complex social and family situations and who have experienced trauma

   2. on-site short-term crisis support within the hospital (e.g. helping the person to get food, providing reassurance)

   3. education and training for hospital and external staff (e.g. on contextual safeguarding); for example, with respect to safeguarding young people aged 11–24 years and how to engage positively with a young person in distress

   4. awareness-raising and other communications within UCLH, for example attending safeguarding meetings, and activity to promote to staff how to make a referral.

4. The service was severely disrupted by the COVID-19 pandemic, and this impacted on the embedding of the Redthread service beyond paediatrics and adolescent services, affecting its influence within the adult ED in particular. Redthread is normally dependent on youth workers’ physical presence in EDs, however, youth workers were not on site at UCLH for some of the period, from March 2020 to April 2021. While the charity continued to provide case management support to young people virtually (which many young people were reported to like due to their habitual use of mobile devices), a lack of physical youth worker presence will have resulted in lower staff awareness of the service.

5. By winter 2021/22, Redthread was perceived to be well embedded in the paediatric ED and adolescent services (including wards), and there was increasing awareness of the service in outpatient departments which saw vulnerable young people with chronic illness. Redthread and clinical staff noted that more could be done to raise staff awareness of Redthread across the trust, especially among nurses, junior doctors and other staff working in adult ED. This is particularly important due to staff changes and turnover (e.g. rotations).

6. Identification of eligible young people was not solely dependent on youth workers being in ED, either paediatric or adult ED. Referral routes at UCLH were diverse and could come from other inpatient services. Changes had arisen due partly to the impact of COVID-19, therefore the identification of eligible referrals went beyond ED services.

7. Reasons for referral were by no means limited to young people with knife injuries or those who had been subject to sexual or physical assaults, but also included substance misuse, suicidal ideation and mental health crises. Staff observed a particular increase in young people presenting with severe mental health problems and distress on account of disruption caused by the COVID-19 pandemic (e.g. domestic violence and family instability). This raised questions about the most appropriate referral criteria for the service, given that Redthread is not a mental health crisis service.

8. The key implementation barriers faced were:

   1. the impact of the COVID-19 pandemic, with paediatric and adolescent services being temporarily moved away from UCLH to neighbouring London trusts

   2. UCLH staff changes and turnover during the implementation period which necessitates regular staff training to ensure all staff are aware of the service and how to refer

   3. lack of physical space for Redthread (e.g. office space to ensure their presence and have confidential discussions with young people)

   4. difficulties engaging young people aged over 18 years presenting in the adult ED, unless a clinician was aware of the Redthread service and had obtained consent prior to their discharge.

9. Staff suggested the following enablers were key to successful embedding of the service:

   1. ‘invested’ senior clinical and operational staff to champion the service locally and across departments (e.g. senior clinicians and ED managers)

   2. integration of Redthread youth workers within hospital processes for identifying vulnerable young people and discussing their care (e.g. MDT inpatient reviews, safeguarding meetings, safety planning)

   3. physical space near ED for engaging with young people (e.g. for confidential discussions)

   4. electronic health-care record integration (e.g. Epic) which has been especially useful for enabling referrals to be made directly by clinicians when youth workers are off site; indeed, some staff even suggested a prompt might be helpful to remind clinicians they can refer an eligible young person to Redthread.

   5. clear standard operating procedures have been agreed between the hospital and Redthread.

Description and review of data used to manage the Redthread service at UCLH (Chapter 5)

1. Redthread collect a wealth of data on individuals, particularly on those who engage longer term with the service. Data on engagement rates support the qualitative finding that a greater proportion of 11- to 17-year-olds referred to Redthread receive some form of support from the service (59%) compared with those aged 18–24 years (23%).

2. Describing the service users and measuring outcomes was challenging in the absence of linkage between Redthread data and routine health-care data. The main limitations to using Redthread data in isolation included:

   1. Redthread’s consent model did not allow individual-level records to be shared beyond their organisation, and we could only analyse aggregated data

   2. aggregated data could not be analysed with respect to multiple characteristics at the same time, so we could not establish whether there were inequalities in engagement levels within specific groups

   3. data on hospital activity of service users were not regularly captured by Redthread and, when collected, only showed activity within UCLH. This is in the context of a hospital where nearly half of all 11- to 24-year-olds who reattended after an initial ED attendance did so at another hospital trust.

3. There were also limitations to only using data from UCLH’s patient administration system:

   1. Redthread referrals were not consistently flagged, and where they had been recorded, there was no indication as to whether the individuals had accepted or declined support

   2. UCLH did not systematically record information relating to the intent behind injuries, which would have helped us to understand whether an individual might have been eligible to receive Redthread services.

4. To improve identification of the eligible cohort, referrals to Redthread services should be recorded in a dedicated data field, and the injury intent emergency care data field should be completed, so that potentially relevant hospital attendances can be drawn out from the data more easily.

5. With respect to new hospital-based YVIPs, we recommend that prior to service roll-out, sufficient time is earmarked for setting out data requirements and to ascertain the feasibility of reasonable options for robust quantitative evaluation.

Costs of service at UCLH (Chapter 6)

1. We were unable to conduct a CEA due to a lack of good quality evidence describing the impacts of YVIPs on subsequent use of hospital services.

2. A CCA showed that there was a statistically significant decrease in assessed risks for young people engaged for longer-term YVIP support for the period April 2020 to November 2021 for the following risks:

   1. ‘experiencing further harm’

   2. ‘not maintaining positive relationships with family’.

3. The mean cost per person engaging with Redthread support, over a 21-month period was calculated to be £1865.

4. The mean cost for the emergency inpatient treatment of an artificially constructed group, similar to those likely to be referred to Redthread in UCLH, was estimated to be £5789 while the mean cost per attendance at the ED was £203.

5. Interpretation of these results should be made with caution due to our research being limited by the small sample of patients, lack of a control group, inability to evaluate effectiveness and the subjective assessment of risks based on Redthread personnel.

Feasibility of quantitative evaluation of service at UCLH (Chapter 7)

1. Redthread aims to have a positive impact across a wide range of aspects of young people’s lives, for example in education, crime, housing, benefits and employment, as well as reducing attendance at hospital. We focused on the latter as we considered it easier to quantify and monitor given the available data.

2. We considered a number of possible options for a rapid quantitative evaluation of the impact of the service but concluded that none of them would be currently feasible. The reasons for this are due to combinations of the following, some of which overlap with the data limitations described above:

   1. relatively small numbers of young people have so far engaged in the longer-term programme with Redthread at UCLH, leading to likely insufficiently powered analyses of the impact of that particular programme over the period of the research project

   2. lack of consent to enable access to individual person-level data for Redthread users to link to hospital administrative data

   3. likely difficulty in being able to detect the impact of the service among wider cohorts (such as in people attending ED with potentially eligible presentations)

   4. a lack of key information recorded in UCLH ED records; for example, information about the most likely human intent of the injury, and patient engagement with the violence intervention programme

   5. our inability to link national hospital inpatient and emergency care records due to the lack of linkable patient identifiers across the datasets

   6. the difficulty in identifying comparable control groups from routine hospital data.

3. We contend that, were data and governance barriers able to be resolved, the two most promising options for quantitative evaluation would be ones:

   1. comparing the data of young people engaged with Redthread to a set of appropriately matched control individuals, and

   2. studying the impact of Redthread at small area level.

4. We have therefore made recommendations to facilitate future evaluation:

   1. For analysts to use reattendance as one measure of impact, and work with clinicians and Redthread to develop criteria from routine hospital records that can be used to identify which attendances are potentially avoidable.

   2. Similarly, develop criteria for identifying groups of patients attending other hospitals from which control groups can be selected.

   3. For Redthread and acute hospital partners to consider mechanisms by which information from the service, as to who chooses to engage with different levels of support, and who chooses not to engage, can be linked to the routine data for analysis purposes.

   4. For clinicians to routinely record when a Redthread referral has been made by using the relevant code in the patient administration system, rather than only using free text. If a person is identified by Redthread themselves, then for this to be also flagged in the routine patient data.

   5. That suitable geographical areas are decided upon between Redthread and UCLH, making it feasible to test an area-level approach.

Service implementation

Several findings from this evaluation at UCLH align with observations from Redthread evaluations at other NHS trusts.

First, NHS staff are receptive to youth workers being on site and value Redthread as an independent, expert, professional service which fills a gap within current NHS provision and has potential to bridge secondary health-care and community services.

Second, the critical importance of timely consent of young people within emergency settings, given the ‘opt-in’ model that underpins the charity’s programme theory and ethos (i.e. young people must voluntarily engage with the service). However, compared with MTCs in London (e.g. St Mary’s Hospital, St George’s Hospital), it appears that young people referred to the service at UCLH comprise a different population profile going beyond the local geographical boundaries (because of transport links) and have different physical injuries to those commonly arising from assault and gang violence (e.g. gun shot and knife wounds, which are usually directed to London MTCs). Therefore, even if gang involvement is directly or indirectly suspected in a case, clinicians speak of complex presentations and treating young people who are victims of sexual abuse, domestic violence, have severe mental health problems or a long-term condition that has flared up due to social factors (e.g. peer intimidation, family conflict). A youth worker’s case load at UCLH might therefore extend beyond short-term crisis support work to providing longer-term support beyond six months, and to helping young people who are not from the local borough and who require advocacy across statutory services both inside and outside of London. This suggests that the Redthread service model for local hospitals (such as UCLH) requires adaptation to reflect differences from MTCs, for example, more diverse referral pathways beyond adult ED including outpatients and specialist paediatric and adolescent services.

Finally, this NIHR RSET evaluation concords with an evaluation in the Midlands that found that services such as Redthread rely on ‘highly motivated individuals to promote adoption by giving access to NHS organisations, leveraging personal and professional networks and internal lobbying’. 80

Quantitative impact

Consistent good-quality evidence on the impact of violence intervention schemes on subsequent health-care use and other related outcomes has been lacking (see Chapter 3). As discussed in Chapters 5 and 7, we have not been able to contribute our own quantitative assessment of impact but have made recommendations for future evaluations. However, our reflections are in accordance with observations made in a qualitative evaluation of Redthread’s YVIP expansion into sites in the Midlands,35 where one of several ‘potential threats and barriers’ was identified as being ‘A failure to value and integrate routine data to describe how YVIPs are supporting the work of NHS and other staff, ensuring safety after discharge and securing engagement with community-based services’.

Costs

The acute treatment costs of penetrating trauma injury in England and Wales vary by patient, cause and body region of injury and treatment characteristics. Our findings are consistent with other studies in England and Wales showing that health-care costs are considerable and much higher than the costs of YVIP interventions. The costs reported for England and Wales, in 2008 by a Trauma Audit Research Network study, which focused on more severely injured patients and also included cost of prehospital care were estimated at £7983. 106 Another study, reviewing the cost of 187 gunshot injuries treated at a teaching hospital in Manchester, reported a mean cost of £2698 for patients admitted for inpatient care. 107 In most of the studies, the ED costs are not identified separately from the total health-care costs. A study conducted in South Wales estimated that about 30.0% of assaults resulted in a visit to the ED at a cost of £200 per attendance in 2019. 105

Another important economic aspect related to programmes targeting youth violence is that some, and possibly the most significant part, of the costs and benefits materialise outside of health and social care. These generally refer to costs associated with the criminal justice system or incarceration, but also costs associated with disability or lost productivity (as opposed to the generation of income from employment and tax). Health-care services have been estimated to be only a small fraction of the total costs (especially in relation to lifetime effects of the physical and emotional harm) amounting to 0.03% of total costs for homicide or 3–6% for crimes with injuries, rape or sexual offences. 103

Global research on violence has identified the burden and costs that violence places on health and social prospects across the life course. 111 Violence affects not only health outcomes, quality of life and social choices that people make but also have wider consequences on families, communities and society. 105

Increased spending on violence prevention would help to reduce the economic burden on the health-care system, the huge burden that arises from the long-term health impacts of adverse childhood experiences and would also increase the savings that would accrue outside of the health-care system.

Strengths

Our study was carried out as a rapid, mixed-methods evaluation, integrating qualitative and quantitative methods. Components included a review of the evidence, interviews, meeting observations, and analysis of national hospital and local service activity and costs datasets. We were aided throughout by close collaboration with a variety of stakeholders including Redthread and UCLH staff, and external advisers. These relationships helped us to gain access to key people, information and datasets.

Despite the challenging circumstances brought about by the COVID-19 pandemic for UCLH and Redthread, we were able to conduct confidential interviews with front-line staff, Redthread managers and youth workers, and a small number of meeting observations, largely on account of working in close collaboration with clinical leads at UCLH, and general staff enthusiasm for the service.

The approach we brought to our quantitative analysis was thorough and broad in its scope. In trying to determine the feasibility of evaluation of impact, we considered several different perspectives: that of the individual young person, the hospital, and the community. This work has led to recommendations to improve future efforts at evaluation which will be helpful beyond the service at UCLH.

This study provides valuable reflections on the main costs of, and potential risk reductions resulting from Redthread YVIP service. It also gives an indication of the success rate Redthread might need to achieve in terms of preventing readmission to recoup the costs of its intervention.

Limitations

The implementation of the service was significantly disrupted by the COVID-19 pandemic, and this had consequences for how the service embedded itself at UCLH. Just as this had an impact on the traction of the Redthread service outside of paediatric and adolescent services, it also had an impact on our ability to engage with staff outside of those services. In particular, it remained particularly difficult to gain traction with the evaluation from staff based in the adult ED at UCLH and those with lower awareness of the Redthread service. Future evaluations or research studies will need to engage with staff treating young people in different parts of the trust beyond paediatric and young people’s services.

The uniqueness of the implementation of the service at UCLH (due to its particular context) also meant that our study would have benefitted from direct comparison with another ED in London. We recommend that multisite case studies are conducted at trust level in future to allow for systematic comparisons across hospitals, at similar time points. In particular, multiple case studies could be used to generate knowledge about how Redthread implementation processes, and youth violence services more generally, should be adapted and tailored given the types of services a trust provides, the staff skill mix and types of clinical presentations most commonly seen within EDs.

The research team were not able to approach young people who had engaged with Redthread to ask directly about the impact of the service for practical and ethical reasons. In addition, because of the pandemic and the necessary shift to remote data collection, we were not able to carry out observations of clinician–patient interactions. We may have therefore missed out on important insights into how the service was received by young people and its wider impacts. While we attempted to mitigate this limitation by seeking to interview Redthread’s youth ambassadors, this latter route also ultimately proved impractical.

While in one respect (as noted above) our analysis of the feasibility of measuring quantitative impact covered a variety of perspectives, for the evaluation of a service whose aims encompass many aspects of a young person’s life and experience, our focus on measuring impact solely via subsequent reattendances or re-admissions was a limitation. Moreover, it is possible that a positive consequence of engaging with Redthread could be an increase in their engagement with health-care services and hence their likelihood to reattend hospital, even for harm-related incidents, but we have no evidence as to the extent to which this might happen.

It is understandable that Redthread has taken utmost care to prevent data of supported young people being shared via strong consent processes, but it must be acknowledged that this has also meant that possibly the strongest options for analysis were not possible. Other routes to analysis were also difficult to realise due to a lack of specific data being collected by the hospital, in addition to relatively small numbers of young people having been engaged for longer-term support over the evaluation period.

The lack of linkable Redthread and hospital data meant that our analysis of costs was not based on directly measured costs of service recipients; we instead had to rely on a constructed group of patients with broadly similar characteristics and diagnoses. However, our success in constructing Redthread-eligible or Redthread-like populations from national hospital data was difficult to appraise. In addition, the constructed group were based only on those admitted as emergencies, and the relatively high cost of treatment per admission reflects this fact.

The analysis of consequences was based on data from Redthread’s risk assessment process. While informative, and based on the professional judgement of youth workers, this process was subjective, and should be considered with appropriate caution. The analysis was carried out with data from a very small number of young people, and the scale from 1–3 used for measuring risk indicator may miss on granularities of the distribution of the individual risk continuum.

Participant representation

Redthread’s service was concerned with providing support to potentially vulnerable or at-risk young people aged 11–24 years. The service’s setting was an inner-city London hospital, close to major national and regional transport links. Of those who received a longer-term service from Redthread and whose ethnicity was known, approximately two-thirds were from nonwhite ethnic groups.

As such, this study reflects on a service which aims to improve the care and prospects of potentially underserved groups of individuals.

While it was not possible to conduct interviews with young people who had experienced the Redthread service for ethical and practical reasons, we interviewed Redthread youth workers and clinical and other staff whose responsibilities and experiences covered the care of potentially vulnerable younger people.

Ovid MEDLINE(R) ALL <1946 to 14 February 2022>

1. Domestic Violence/ or Violence/ or Intimate Partner Violence/ or Gun Violence/ 43545

2. assault.mp.  12686

3. ‘Wounds and Injuries’/ 80409

4. human trafficking.mp. 819

5. domestic violence.mp. or Domestic Violence/ 11958

6. Crime Victims/ or knife crime.mp. 11703

7. Wounds, Penetrating/ or stabbing.mp. 13277

8. (gang violence or gang).mp. 1459

9. gang exploitation.mp.  1

10. sexual exploitation.mp.  650

11. Adolescent/ or adolescent.mp.  2194729

12. Young Adult/ 978523

13. children.mp. or Child/ 2225512

14. juvenile.mp. 90983

15. Emergency Service, Hospital/ 80576

16. urgent care.mp. 2766

17. (Major Trauma Centre or Major Trauma Center).mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub-heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 634

18. emergency department.mp. 99941

19. Hospitalization/ 125139

20. Pediatrics/ 57154

21. youth worker.mp. 24

22. Adolescent Development/ 5155

23. teachable moment.mp. 593

24. youth services.mp. 183

25. hospital-based violence prevention.mp. 10

26. intervention.mp. 723390

27. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 156293

28. 11 or 12 or 13 or 14 3898408

29. 15 or 16 or 17 or 18 or 19 or 20 308825

30. 21 or 22 or 23 or 24 or 25 or 26 728825

31. 27 and 28 and 29 and 30 554

32. limit 31 to (english language and yr=‘2012 - 2022’) 283

Embase <1980 to 2022 Week 06>

1. Domestic Violence/ or Violence/ or Intimate Partner Violence/ or Gun Violence/ 66651

2. assault.mp. 20248

3. ‘Wounds and Injuries’/ 119213

4. human trafficking.mp. 950

5. domestic violence.mp. or Domestic Violence/ 14920

6. Crime Victims/ or knife crime.mp. 4016

7. Wounds, Penetrating/ or stabbing.mp. 10183

8. (gang violence or gang).mp. 1786

9. gang exploitation.mp. 1

10. sexual exploitation.mp. 1004

11. Adolescent/ or adolescent.mp. 1640407

12. Young Adult/ 445356

13. children.mp. or Child/ 2235948

14. juvenile.mp.  147699

15. Emergency Service, Hospital/ 7609

16. urgent care.mp. 4413

17. (Major Trauma Centre or Major Trauma Center).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword heading word, floating subheading word, candidate term word] 1159

18. emergency department.mp. 153578

19. Hospitalization/ 439434

20. Pediatrics/ 80304

21. youth worker.mp. 62

22. Adolescent Development/ 4264

23. teachable moment.mp. 778

24. youth services.mp. 230

25. hospital-based violence prevention.mp. 9

26. intervention.mp. 1106006

27. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 215902

28. 11 or 12 or 13 or 14 3415021

29. 15 or 16 or 17 or 18 or 19 or 20 656973

30. 21 or 22 or 23 or 24 or 25 or 26 1110762

31. 27 and 28 and 29 and 30 497

32. limit 31 to (english language and yr=‘2012 - 2022’) 366