This chapter contains reproduced material from Aslam et al.1 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) licence, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The text below includes additions and formatting changes to the original text.
Study setting
STRETCHED was designed as a mixed-methods ‘natural experiment’ evaluation, based on anonymised linked routine outcomes and qualitative data in four UK ambulance services with one intervention (‘case management’) and one control (‘usual care’) site in each service. The East of England Ambulance Service NHS Trust (EEAS), London Ambulance Service NHS Trust (LAS), Welsh Ambulance Services NHS Trust (WAST) and West Midlands Ambulance Service University NHS Foundation Trust (WMAS) were approached and selected following a survey of practice across the UK, which identified where both case management and traditional models of care were in place in different areas (‘sites’) within an ambulance service’s catchment area.
Study population: natural experiment
The study population was defined to include adults meeting nationally agreed criteria for classification as ‘frequent caller’ by ambulance services during 2018. Thus, people were eligible for inclusion in STRETCHED if:
- they were eligible for inclusion on a participating ambulance service’s frequent callers list during 2018 (made 5 or more emergency ambulance service calls in 1 month, or 12 or more times in 3 months)
- were aged 18+ at the time they became eligible
- lived in a study site at the time they became eligible.
People were only added to the study cohort the first time they were eligible for inclusion on a frequent callers list during 2018; two ambulance services included patients classified as frequent callers at the start of 2018.
Participant recruitment: interviews
Service providers
We sought to interview a wide range of service providers, including front-line staff (paramedics, call handlers), partner health and social care providers, commissioners and managers (n ~ 36).
Patients receiving case management
To gain a strong picture of patient experience and circumstances, we sought to conduct largely unstructured interviews with people in intervention areas of each service referred for case management (n ~ 32).
People self-identifying as frequent callers (STRETCH-UP)
In an addition to the original protocol, we sought to boost the numbers of interviews exploring patient experience and circumstances by an open call recruitment process via third-sector organisations and social media (n ~ 20).
Interventions
Describing the intervention
Details of ambulance service resources used in the intervention implementation were obtained from relevant documents and through discussion with and surveys of staff involved in case management at the study sites. Documents reviewed included a survey of 12 UK ambulance services regarding the management of the care of people who make frequent emergency calls during the study period,11 TIDIER checklists describing the intervention in the four study sites, and transcripts of the qualitative interviews with commissioners, frequent caller leads and care providers were undertaken as part of this study.
East of England Ambulance Service
In the EEAS intervention site, case management was initially organised by a clinical co-ordinator and a frequent caller lead. The team then evolved, comprising a frequent caller lead managing three clinicians and an administrator. Based on monthly reports, the frequent caller lead contacted patients who were identified as frequent callers. Patients were discussed at the monthly multidisciplinary high-intensity user group (HIUG) meeting, which agreed on individualised action plans, based on patient needs and linked with services involved in a patient’s care [e.g. their general practitioner (GP), social services, mental health teams, one-on-one out-of-hours, police in some cases]. The clinical co-ordinator, frequent caller lead and management team liaised to put in place a management plan to reduce ambulance callouts by meeting unmet care needs.
London Ambulance Service
In the LAS intervention site, the frequent caller lead reviewed callers meeting the frequent caller criteria on a monthly basis and referred eligible patients to the most appropriate external care service based an assessment of needs. Patients were contacted within 1 week and their care needs discussed at MDT meetings where actions and a future care strategy were devised. This initiative was led by a GP, with patient contact largely undertaken by a specialist nurse and non-clinical care co-ordinator, both working part time on the initiative. Patients were monitored on a monthly basis if they continued to call.
Welsh Ambulance Service
In the WAST intervention site, a Clinical Support Officer (CSO) provided clinical and management support for 6 ambulance service frequent caller leads, working closely with 13 frequent attendee leads within hospital EDs. The CSO analysed available data monthly to identify frequent callers. Ambulance frequent caller leads and ED frequent attendee leads jointly managed patients identified via letters, case meetings with GPs, and home visits to assess patients’ needs. Patients were discussed at ED-based MDT meetings, and referred to the appropriate agency with care plans created and shared with other agencies to provide a network of support for the patient.
West Midlands Ambulance Service
In the WMAS intervention site, the Ambulance Service frequent caller lead was head of clinical practice and WMAS consultant, with a varied portfolio of tasks, including the high-intensity service user or frequent caller portfolio. They provided strategic overview, data and information to all partners including 24 Clinical Commissioning Groups (CCGs), most with a high-intensity service user group working on behalf of the GPs. The CCG lead received the list of frequent callers from the ambulance service frequent caller lead on a monthly basis and contacted and/or visited identified patients to assess need. They also liaised with community services to provide required services, participated in MDTs and devised care plans to ensure that patients had access to the most appropriate care pathway.
Characteristics of care at control sites in the four ambulance services participating in STRETCHED are outlined in Table 1, together with a summary of key features of case management.
| Control site: usual care | Intervention site: case management | |
|---|---|---|
| Aim | To discourage people classified as frequent callers from further calling | To identify and manage triggers and causes that lead to the patient making frequent calls (e.g. adverse childhood events, mental health, drug and alcohol dependency, frailty) |
| Care model: delivery | Within-service management | MDTa cross-sector case management |
| Model elements | CSO contacts patient by letter stating they have called the emergency ambulance service more than normal and should seek help from their GP; contact number provided within letter for patient to talk to ambulance service manager11CSO calls or sends letter to patient’s GP to make them aware patient is calling the emergency ambulance service frequentlyIf calls persist, CSO may contact other services to intervene and support patientCSO writes individual care plan which is shared with the call centre clinical team. When the patient calls, he/she is triaged to a clinician in the call centre rather than an ambulance being sentIf caller persists, he/she may be referred to the police and/or court | CSO speaks to GP or practice manager to assess patient’s case and needsPatient is discussed at monthly MDT meetingPatient is interviewed to assess unmet needsPatient is allocated to appropriate agency to lead on care planning and provision. A care plan is created and shared so that any agency contacted by the patient knows what has been agreedIf calls persist or patient has an antisocial behaviour order, the CSO visits the patient along with a police officer or representative from the ED |
Intervention implementation costs
The health economic analysis compared the implementation cost of case management to the potential benefits of ‘avoiding expenditure’ in subsequent health and social care resource utilisation (where available). The implementation intervention costs included a descriptive summary of resource use, cost tables and an overview of incurred costs of intervention implementation at different study sites (where data availability and quality allowed). Although the model of delivery of these services can differ between ambulance services and within ambulance service sites and change over time, in all cases, case management was available in addition to usual care (even though the scope and nature of usual care differs). Usual care was therefore either non-existent or remaining unchanged and, for this analysis, was assumed, following Research Management Group (RMG) discussion, to be the same for patients receiving case management and those who do not. The intervention implementation cost included in the cost–consequences analysis (CCA) is therefore based on the difference in costs between the usual care support provided by the ambulance service throughout all services and sites (if available) and the new additional resources and associated costs required to provide the intervention of enhanced case management support for frequent callers. This includes the additional funding required to operate the case management intervention, which is predominantly driven by costs of staff operating and providing the case management intervention. This cost is expressed in relation to the number of patients supported by the scheme as the monthly cost per patient case-managed.
Logic model
A logic model is a linear graphical representation of the theory of how an intervention produces its intended outcomes; it can help to prioritise and structure data collection and analysis to explore the main aspects of an intervention.
We organised a stakeholder event to bring together patient representatives and professional staff involved in commissioning, planning and delivering case management for people who call emergency ambulance services frequently. We presented an overview of STRETCHED, an explanation of what a logic model is and its purpose, and then broke into small, mixed groups to discuss the four parts of the logic model – components, mechanisms, impacts and context. We then developed our logic model in the light of these discussions.
Outcomes
We finalised study data items following completion of the logic model, specifying items which would enable us to obtain data on:
- further emergency contacts:
- emergency ambulance calls
- emergency department attendance
- emergency admissions to hospital
- declassification/reclassification as ‘frequent caller’
- effects on other health and social care services
- adverse events, as available
- deaths
- injuries
- serious medical emergencies
- police arrests
- costs of intervention and subsequent use of health and social care
- patient experience of intervention.1
We further specified data items related to patient demographics, case mix and patterns of calls, for example, ‘out-of-hours’ (evenings/nights/weekends/holidays). Historical data about healthcare utilisation immediately prior to follow-up allowed us to adjust analyses for differences between cohorts, strengthening this study design.
Primary outcome
The primary outcome in STRETCHED is a composite hierarchical indicator of mortality; emergency hospital admissions; ED attendances; and emergency ambulance service calls.
Data collection
Blinding
STRETCHED was not blinded.
Natural experiment
Ambulance services provided NHS Digital and Digital and Health Care Wales (DHCW) (formerly, the NHS Wales Informatics Service; NWIS) with identifiable information on study patients, enabling matching to anonymised data from multiple sources. In parallel, the ambulance services submitted clinical and operational data on study patients to the SAIL databank. After linkage and anonymisation processes were complete, we created a single integrated study database for analysis within the SAIL Gateway, a Trusted Research Environment.
Patients could request that their data were not used in the study by contacting the appropriate ambulance service. Data ‘opt-out’ requests to NHS Digital and the SAIL databank were upheld.
Sample size
We expected to identify 158 eligible patients per study site (316 per ambulance service; 632 per arm; 1264 cases total). Assuming that no more than 5% of cases would be lost to follow-up (e.g. via opt-out requests to NHS Digital), we expected to include in analysis 300 patients per ambulance service (n = 1200), with 90% power to detect a standardised statistical effect of 0.2 at the 5% significance level.
Health economics
In costing the implementation of the intervention, ambulance services were asked to provide additional information on the delivery of the case management intervention that could not be retrieved from other sources using a resource use questionnaire sent by e-mail. The questionnaire was designed to provide an understanding of the composition of the team delivering the intervention and the percentage of staff contracted hours required to operate the enhanced case management intervention beyond usual care (if provided) and pay bands (to calculate costs). A simplified follow-up questionnaire was then sent to sites containing specific questions about the main staff previously identified. See Report Supplementary Material 1 for further details on the two questionnaires.
Staff time required was converted into staff costs using the mean annual basic pay per full-time equivalent (FTE) for staff using NHS Agenda for Change pay scales50 where the staff pay band was known. The mean annual basic pay per FTE for non-medical occupational groupings for NHS England was used for English sites and Wales NHS Agenda for Change pay scales 2018–9 for the Welsh Ambulance Service. Where no pay bands were available but professional occupation was known, NHS staffing costs were derived from the Unit Costs for Health and Social Care 2018–9,51 which covers the 12-month period of earning estimates from May 2018 to April 2019. Where pay band information was unavailable but professional occupation was known (e.g. nursing clinician or administrator), the mean annual basic pay per FTE for non-medical occupational groupings for NHS England was used for English sites and Wales NHS Agenda for Change pay scales 2018–9 are used for the Welsh Ambulance Service.
Once all intervention components were costed, costs of all components were added up to arrive at a case management intervention implementation cost per site.
Healthcare resource utilisation and costs
Data on resource utilisation and the resulting costs to health care were collected for the period between 2017 and 2019 and assessed over a time horizon of 6 months from when a patient first became eligible for inclusion on the frequent caller list. Data were obtained from routinely collected healthcare data available within the Hospital Episode Statistics (HES) data sets for England and the SAIL databank for Wales. The healthcare resource use data available for analysis comprised emergency ambulance service calls, ambulance dispatches, ED attendances, emergency admissions, elective inpatient stays and outpatient visits.
For emergency admissions, all finished consultant episodes (FCEs) with length of stay of 1 day/overnight stay were considered non-elective short stay, whereas FCEs of 24 hours or more were costed as non-elective long stay. For non-elective hospital episodes with length of stay of zero, a short stay FCE cost was applied. ED attendances were costed based on Healthcare Resource Group (HRG) codes where possible. Similarly, emergency ambulance service calls were costed using the appropriate HRG codes.52 In cases where it was impossible from the available data to distinguish between calls and hear and treat or refer cases, a weighted average unit cost of £10.29 was applied.
Unit costs were taken from the NHS Reference Costs;52 HRG codes were requested for all healthcare components and data sets.
Costing resource use included in the primary outcome
As per the clinical effectiveness analysis, resource use included in the primary health economics outcome comprised emergency admissions, ED attendances and emergency ambulance service calls.
Secondary outcome resource use costing
Resource use collected for the secondary outcomes of the study included ambulance callouts, outpatient appointments and elective inpatient admissions.
As with the primary outcome resources, resource use for secondary outcomes was costed using HRG codes (if supplied with the data) or weighted average unit costs.52 Ambulance callouts were costed using the appropriate HRG codes.52 Outpatient data retrieved from HES were costed using the HRG codes in conjunction with speciality service code supplied. Speciality service codes were used to cost Welsh data (retrieved from the Patient Episode Dataset for Wales, PEDW). For elective admissions, FCEs of 1 day or more were costed as elective inpatient stays, with episodes with length of stay of 0 considered to be day cases.
Total healthcare costs
Once costed, the total number of healthcare resources used and the total costs accrued over the 6-month follow-up period for primary and secondary outcomes and for both combined, as well as the mean resources used per patient and mean cost per patient (and standard deviation, SD), were calculated for intervention and control arms using SPSS (SPSS Inc., Chicago, IL, USA) and EXCEL (Microsoft Corporation, Redmond, WA, USA).
Costs were expressed in 2018–9 UK Great British pounds (GBP). No discounting was applied to either costs or consequences as the follow-up period did not exceed 1 year.
Qualitative elements
We collected qualitative data at the intervention site within each participating ambulance service to explore perceptions of how the intervention worked, what created its effect (if any), why it might function differently in different settings and for different groups of people, and any challenges to implementation and delivery of the intervention. In addition, we recruited people who identified as frequent callers via third-sector organisations and social media, as part of the STRETCH-UP substudy.
We conducted in-depth interviews with a range of service users who had been referred for case management within one of the ambulance service areas; although we originally planned to interview patients referred for case management in all ambulance service areas, we were not able to recruit in the other three sites. In order to gather additional user data on the user perspective, we added a supplementary qualitative study, STRETCH-UP, which recruited interview people with experience of frequent calling to ambulance services across the country. Data from STRETCH-UP are presented alongside data collected in line with the original STRETCHED protocol.
Service providers
We planned to conduct up to eight interviews (remotely via video or telephone) with a range of service providers in each ambulance service intervention site. Interview guides covered case management delivery processes, barriers and facilitators to changed working, perceived impact for patients, issues around diversity and terminology, strengths and weaknesses of the approach and wider organisational impact across health economies, such as information sharing, communication and continuity of care.
Patients
We aimed in each intervention site to interview people who had been referred for case management, to explore their own narrative about their circumstances, experiences and views regarding their needs, service use and care received before and after the case management intervention, and the intervention itself, including terminology used. We worked with participating services to identify and invite callers to this key patient-focused element of the study. We intended purposively to include a wide range of views and experiences, selecting individuals with differing demographic characteristics, length of time they have required care and management approach provided for them to include typical and atypical patient stories. We specifically sought to include patients from ethnic minorities.1
Interviews were conducted by a member of the study team (AK). With respondents’ consent, we audio-recorded and transcribed all individual interviews.
People self-identifying as frequent callers (STRETCH-UP)
In an addition to the original protocol, we sought to boost the numbers of interviews exploring patient experience and circumstances by an open call recruitment process via third-sector organisations and social media. We recruited participants with the help of our existing networks, including the SUPER group and third-sector organisations such as Diabetes UK. We produced an illustrated flyer and asked organisations to share it on their social media accounts [e.g. Twitter (Twitter, Inc., San Francisco, CA, USA)], attach it to routine e-mailings to members, and feature it on their websites, as appropriate. Potential participants were invited to make contact directly with the study team; they would then be sent a more detailed information sheet and invited to discuss the study and see if they were eligible to take part, before being formally recruited and consented. Interviews were conducted remotely (via video) by members of the study team (AP, BAE, AT) and with respondents’ permission were recorded and transcribed in full.
Please see Report Supplementary Material 2 for consent forms and participant information sheets and Report Supplementary Material 3 for focus group and interview topic guides.
Data analysis
Natural experiment
Our primary analyses used the treatment allocated (the ‘intention to treat’ principle), so patients that met the criteria for ‘frequent caller’ status in an intervention (case management) site were included in analysis in that arm of the study whether they were offered or received any intervention. The outcome measure time point was defined as 6 months from a patient’s first eligibility for inclusion on a ‘frequent callers’ list and applied to patients at control sites as well as intervention sites. This time interval was regarded as appropriate, as case management is a targeted, time limited intervention designed to work within 6 months.
Arms were compared using two-sided tests at the 5% significance level, adjusted for patient characteristics (ambulance service, age in years, sex, ethnicity, Townsend deprivation quintile) and service use (number of emergency ambulance service calls, emergency admissions, elective admissions and outpatient appointments) during the 6 months before patients became eligible. No adjustment for multiplicity was required as the primary outcome had a pre-specified order of analysis. Ninety-five per cent confidence intervals (CIs) are reported where appropriate. Analysis was conducted using SPSS hosted within the SAIL Gateway, a Trusted Research Environment maintained by Swansea University.
We summarised patient recruitment via a CONSORT53 flow chart, and provided descriptive data summaries for patients, by ambulance service and by site, for demographics, call patterns and volume. Numbers of events were dichotomised where contextually useful (for instance, to declassify or reclassify patients), and converted to rates when appropriate.
Outcomes for the 6 months after patients first became eligible for inclusion in STRETCHED were compared between arms using generalised linear models (logistic models for binary variables; negative binomial models for count variables; linear models for raw and transformed measurement outcomes) with appropriate linking functions. From these models, we obtained adjusted comparisons of outcomes in patients at intervention sites versus those in patients receiving usual care at control sites.1
Using linked routine data allowed us to gather retrospective data on service use for callers included in each cohort, strengthening comparisons by enabling adjustment for historical and contemporaneous differences in service use, case mix, and demographics. Reflecting an initial focus on comparing (any) case management approach with the within-service list approach, coding for sites initially assumed that outcomes by study arm were similar across ambulance services.1 We also presented background information on ambulance service interventions and descriptive summaries of outcomes by ambulance service to test this assumption. The adjusted comparisons incorporated information on covariates and factors based on demographic and case-mix data.
We formalised our planned analyses in a combined Statistical and Health Economics Analysis Plan (SHEAP), compliant with the relevant Swansea Trials Unit (STU) standard operating procedures (SOPs); the SHEAP (see Report Supplementary Material 4) detailed conventions on model fitting (including inclusion and exclusion rules for covariates and factors), management of missing data and the reporting of outcomes.
Health economic evaluation
The health economic analysis framework consisted of a CCA of the costs and outcomes/consequences of case management from an NHS and Personal Social Services perspective.
A CCA is a form of economic evaluation where the full range of disaggregated costs and outcomes are reported separately in a format which allows a comprehensive summary of the different costs and negative and positive effects, covering health and non-health, both to patients and other parties. A CCA thus tends to take a broad perspective, and is considered to be not only more transparent and comprehensive but also easier to understand and apply for decision-makers, when compared with other methods of health economic evaluation.54 The analysis follows a descriptive approach which presents effectiveness results (primary and secondary outcomes) separately to that of the costs, providing a disaggregated overview of return on investment. The CCA approach is recommended for complex interventions that look to observe a broad range of effects which are difficult to measure in a single common unit and is considered particularly appropriate in observational studies,55,56 particularly when collection of the data on quality of life required for a cost–utility analysis is not practical or possible.
Using the CCA framework, we compared the implementation cost of case management approaches to the potential benefits of ‘avoiding expenditure’ in subsequent health and social care resource utilisation from a public sector perspective. The overall resource implications for case management approaches for frequent callers included intervention implementation costs, costs of further emergency ambulance service calls and the costs arising from utilisation of other health and care services in the 6-month follow-up period.
The utilisation of other health and care services by the study cohorts in the control and intervention sites was captured using routine data sources – NHS Digital for the three English ambulance services and the SAIL databank for WAST – and costed using appropriate published unit costs. The inclusion of mental health and social care resources were subject to the availability of routine data and assessment and validation of its quality. Where HRG cost codes were available, the accompanying unit cost per HRG code was applied on a per-event basis using a look-up algorithm (created in SPSS syntax) for the healthcare resource use costing for the study. If HRG codes were not available for some resource components (e.g. hospital inpatient stays), the available data were compared to the required field list for the NHS Digital Reference Cost Grouper57 which can generate HRG codes for each activity record if certain mandatory fields are available within the data. In cases where no HRG codes were available and use of the Digital Reference Cost Grouper was not possible due to missing data in mandatory fields, overall weighted average unit costs across all HRG codes (excluding paediatrics) were used for costing.
We used linear models to adjust for baseline total number of healthcare service contacts and total baseline cost, as well as age, deprivation quintile, ethnicity (collapsed into white/non-white) and gender commensurate with the statistical analysis and to calculate adjusted mean difference with 95% confidence intervals (CIs) and p-values between the two patient groups. Healthcare costs associated with the primary and secondary outcomes were also analysed by site to assess differences in consequences of the various ways case management was employed in practice on secondary and emergency care use. The combined SHEAP (Report Supplementary Material 4) specified conventions on model fitting for health economics outcomes, consistent with the modelling of clinical outcomes in inclusion and exclusion rules for covariates and factors, management of missing data, and reporting of outcomes.
Cost–consequences analysis
The CCA approach separates costs and consequences and reports them in a disaggregated manner. A cost–consequences table was compiled to report the disaggregated costs and outcomes associated with case management compared to usual care for people who call the emergency ambulance service frequently comparing cumulative costs over the full period of follow-up against the outcomes.
Qualitative analysis
Analysis of focus group and interview transcripts was carried out by members of the research team with the two patient and public involvement (PPI) representatives, alongside input from the research paramedics. We removed all identifiable data from interview transcripts before analysis.1 We used a data-driven thematic approach to analysis which generated themes from the implicit and explicit ideas within participants’ accounts.58 We followed six stages of analysis:59 data familiarisation, generating initial coding, searching for themes, reviewing themes, defining and naming themes, producing a report. During analysis, we checked emerging coding and themes with the wider research team and members of the patient panel. We also assessed whether we had an adequate range of participants reflected in our purposive sample.1 Analysis of data (transcripts, field and observation notes) generated by the interviews was supported by use of NVivo (QSR International, Warrington, UK), computer-assisted qualitative data analysis software.
Synthesis
The STRETCHED logic model (WP1) informed our synthesis and reporting of quantitative and qualitative findings. Quantitative data were used to draw conclusions about comparative costs and effects; qualitative data helped us to understand and interpret these results and to generate theories about how the new models of care were working. We gathered key themes from across all WPs on the effectiveness, attitudes, barriers and facilitators to case management. We interpreted overall effectiveness and cost-effectiveness results in the light of analysis from the qualitative data about which components of case management – for example, care plan, timing of interventions, shared decision-making – were perceived to work well and for whom. We used the logic model to inform the synthesis of results, considered and interpreted at a joint meeting of the RMG and the Lived Experience Advisory Panel (LEAP).
Changes to study design
The natural experiment in the four ambulance services was undertaken within the framework initially established; we extended the intervention sites in two ambulance services when it became clear we had initially overestimated the number of eligible callers within original sites.
Recruitment of patients for the qualitative work proved more challenging, particularly in the three English ambulance services – in Wales, we benefitted from a concurrent and related research initiative. We therefore extended our original design to include a small supplementary study (STRETCH-UP), in which we sought further interviews with people with experience of frequently calling ambulance services, identified via peer support/third-sector organisations. We intended to approach a variety of organisations, including National Users Survivor Network, National Voices, British Lung Foundation, British Heart Foundation, National Centre for Mental Health cohort. Participants could be self-selecting; they may or may not have had any experience of case management, but able to provide insight into why they made high-intensity use of services.
Public and patient involvement
We worked extensively with public contributors to develop a comprehensive approach to active involvement of patients and the public at all study stages.60 We used a layered approach enabling people to be involved at strategic and local levels in line with their interest, experience and health. We aimed to enable active and meaningful involvement throughout to enhance research quality, rigour and ethical standards, in line with the STU SOP on service user involvement61 and best practice.62 Our public co-applicants (PG, BE) were actively involved in study design and shaping the proposal. They were also involved in proposal design to undertake interviews with patients, challenging biases and assumptions in the team and helping choose proper language to describe people who phone the emergency ambulance service.1 They were members of the RMG and equal partners in decisions about study implementation and dissemination.
We also convened a patient advisory panel (subsequently retitled the Lived Experience Advisory Panel, LEAP) comprising individuals recruited through community groups, support agencies and third-sector networks. The panel reflected the range of people who make frequent calls to the emergency ambulance service including older people, those with chronic illness and people from lower socioeconomic levels. We sought to ensure that public contributors within STRETCHED matched as closely as possible the diverse population of people making frequent calls. LEAP provided a less formal route for people to contribute to specific tasks such as advising on patient-facing materials, sense-checking patient results and devising dissemination materials.62–64
We recruited a further two public members to the independent Study Steering Committee (SSC) to bring patient perspectives to oversight and scrutiny decisions. We offered honoraria for all involvement and reimbursed incurred expenses. We provided training for all public contributors, including induction at start of involvement and training in specific skills, such as Good Clinical Practice, analysis, meeting skills and dissemination skills, as required.64,65
Research and governance, ethical approvals and permissions, study oversight
STRETCHED was sponsored by Swansea University. We received approval from the Health Research Authority and R&D permissions at all participating NHS organisations. The Health Research Authority approval included a favourable opinion from an NHS Research Ethics Committee (19/WA/0216) and Confidential Advisory Group (CAG) support for accessing patient data without consent.
With this CAG support, it was not necessary to approach patients for consent to participate in the main effectiveness study (WP2); instead, STRETCHED used identifying information held within ambulance services only (and not shared with the wider research team). With information governance permissions in place at NHS Digital and the SAIL databank, retrospective routine data were linked using a split file approach, and an integrated study data set for analysis created within the SAIL Gateway.66
The relatively small sample of current patients calling frequently were invited by participating ambulance services to give consent and participate in one-to-one interviews. We recognised the potential vulnerability of these patients, and we took steps to ensure that interviews were conducted in a sensitive manner, at a place of the interviewee’s choice or by telephone/video and by appropriately trained researchers at each site. We obtained Swansea University Research Committee approval for the supplementary STRETCH-UP extension, in which we sought to recruit further participants for the qualitative work.
We convened an independent Study Steering Committee, with members from: a non-participating ambulance service; the Frequent Caller National Network (FreCaNN); a range of third-sector healthcare-related organisations; the police service; and academia. Two members offered a PPI perspective; the SSC chair had links with a carers’ trust.