Prehospital video triage of potential stroke patients in North Central London and East Kent: rapid mixed-methods service evaluation

Optimising access to organised stroke care in the NHS in England

Stroke is a major cause of death and disability in the UK and internationally. 2 Evidence from urban settings suggests that ‘system optimisation’ of stroke services, whereby they are centralised into a small number of hyper-acute stroke units (HASUs), is linked with better care delivery and outcomes. 3–6 HASUs (which in future will be termed acute stroke centres, or comprehensive stroke centres if they deliver mechanical thrombectomy) offer rapid access to stroke specialist assessment and treatment, including intravenously administered ‘clot-busting’ therapies, if appropriate.

Such centralised services rely on effective collaboration between multiple stakeholders, including hospital stroke services and ambulance services, to ensure the appropriate conveyance of patients to a HASU. 7–9

In recent years, the NHS in England has sought to improve patient access to organised stroke care in several ways. The NHS Long Term Plan reinforces the role of networked stroke systems at a regional level to improve care delivery and clinical outcomes. 10 Integrated Stroke Delivery Networks (ISDNs) are made up of multiple health agencies, including ambulance trusts, and aim to ensure that NHS stroke services comply with 7-day quality standards for stroke care and National Clinical Guidelines for Stroke. In addition, there is support to scale up technologies that improve the quality of stroke services, such as the potential use of artificial intelligence to interpret computerised tomography (CT) and magnetic resonance imaging (MRI) scans and the implementation of telehealth. 10

Minimising unnecessary conveyance to specialist stroke units using telemedicine

Because of the limited specificity of screening tools for stroke [e.g. the Facial drooping, Arm weakness, Speech difficulties and Time (FAST) tool, or variations, which is used by most ambulance clinicians and has been heavily promoted in the national press], acute stroke services commonly manage large numbers of patients who, although suspected to be having a stroke (e.g. because they are ‘FAST positive’), turn out to have non-stroke conditions (so-called ‘mimics’). 11 Remote specialist stroke assessment via telemedicine has been found to support accurate triage of patients12 and has the potential to identify patients who do not need urgent treatment in a specialist unit. However, to date, the piloting and implementation of such technologies has been limited in England. Reported obstacles to adoption include technical issues (e.g. reliable video-call signal quality) and cultural barriers (e.g. ambulance clinicians’ concern regarding the benefits of potentially increasing on-scene time to seek specialist secondary care stroke expertise).

International evidence for ‘telestroke’ and mobile stroke units

Elsewhere, there has been movement towards the uptake of telemedicine in stroke care, particularly in the USA and Germany (see Chapter 3). 16 There is emerging evidence about telemedicine’s safety and cost-effectiveness,13,14 indicating that the uptake of telemedicine in stroke care can provide neurological expertise in real time effectively and within the tight time window necessary for hyper-acute stroke treatment. 12

However, from clinicians’ perspectives, there may be issues with the usability of new telestroke systems or mobile facilities that rely on visual cues. Reviews of the evidence on implementing telestroke describe several obstacles to and enablers of adoption. Major barriers include unfamiliarity with the technology and how it conflicts with cultural norms; technical issues with audio-visual quality; lack of staff confidence in systems; lack of information technology (IT) support; and poor communication between clinicians. 15,16 Reviews also note the potential impact of prehospital video triage on decision-making, for example through the addition of stroke specialist expertise. 15

A recent scoping review16 describes a small but growing evidence base on how remote technologies may support ambulance clinicians in triaging potential stroke patients, including the use of prehospital biomarkers and imaging and mobile telemedicine. The review16 reports no UK-based research on ambulance telemedicine systems to support remote assessment by stroke clinicians. However, the international research reported indicates that such systems are viewed positively by staff and can result in reduced time to care interventions, and that prehospital remote diagnosis can be as accurate as hospital-based diagnosis. 16 The review also noted relatively few data to suggest that such systems result in more appropriate conveyance of patients, or about their impact on outcomes such as patient safety. 16

Understanding implementation of digital innovations in health-care systems

There is increasing interest in how innovative digital technologies come to be adopted by and used in health-care systems and may shape clinical practices and workflows. For example, recent World Health Organization guidance17 suggests that digital health should be understood in terms of an ongoing process of development, whereby digital interventions evolve from early piloting and prototyping to digital maturity, requiring ‘real-time’ monitoring of both technical functionality and stability in addition to health outcomes.

NHS service context for the pilot services

As outlined above, there are many areas (including implementation, impact and experience) where knowledge is limited in relation to prehospital digital triage services.

In this evaluation, we studied two pilot schemes that introduced prehospital video triage for suspected stroke patients in North Central (NC) London and East Kent. These pilots were introduced in response to the COVID-19 pandemic. Below, we provide contextual information on these two areas and summarise briefly the local responses to COVID-19.

The NC London pilot took place in the North Central sector of London, covering a population of 1.2 million people. 18 The area is served by a single HASU, hosted by University College London Hospitals NHS Foundation Trust (UCLH) and the London Ambulance Service NHS Trust (LAS). This sector is part of the wider London ‘hub-and-spoke’ acute stroke service model, implemented in 2010, whereby all suspected stroke patients are eligible for initial treatment in a HASU (hub) and, if required, ongoing acute care in stroke units (SUs), which act as ‘spokes’, offering specialist acute rehabilitation services nearer home. 8 In response to the pandemic, the UCLH HASU was relocated from the main UCLH hospital site [where it had been co-located with an emergency department (ED)] to the nearby National Hospital for Neurology and Neurosurgery (NHNN), a dedicated neurological and neurosurgical hospital (which did not have an ED).

The East Kent pilot covered the area served by East Kent Hospitals University NHS Foundation Trust (EKHUFT) and the South East Coast Ambulance Service NHS Trust (SECAmb). EKHUFT serves a population of just under 700,000 people. Pre pandemic, this area was served by two stroke units, at the William Harvey Hospital and Queen Elizabeth the Queen Mother Hospital (QEQM). However, in response to the pandemic, hyper-acute stroke services were moved from these sites to a single routinely-admitting service at Kent and Canterbury Hospital (Canterbury, UK). As with the revised model in UCLH, this reorganisation moved the HASUs away from co-location with the trust’s ED and to a new location that did not have immediate ED support.

Pre-hospital video triage services implemented in North Central London and East Kent

In 2020, NHS services in NC London and East Kent introduced prehospital video triage services in response to the developing COVID-19 pandemic. These were ‘on-scene’ digitally supported systems that let ambulance clinicians contact acute stroke clinicians for remote clinical assessment using digital communication platforms [i.e. FaceTime (Apple Inc. Cupertino, CA, USA)] using communications devices [i.e. smartphones and iPads (Apple Inc.)]. The aim of the video assessments was to establish whether or not a patient was suitable for conveyance to a HASU or if they should be on a different care pathway, thus minimising unnecessary conveyance or delays. The anticipated benefits of this system were to:

• Support appropriate referrals to HASUs or other pathways [e.g. local ED, General Practitioner (GP) or transient ischaemic attack (TIA) clinic] during the COVID-19 pandemic. This would contribute to streamlining of care for different patient groups and protect vulnerable older patients from unnecessary conveyance to hospital, where there was a risk of exposure to COVID-19.

• Ensure the timeliness of treatment for optimal patient outcomes, in accordance with best practice guidelines.

• Help services to run as efficiently as possible (e.g. by providing decision support to ambulance clinicians and reducing the number of unstable non-stroke patients being brought to a HASU without co-located ED support).

Table 1 provides an overview of the key events in the development and early implementation of prehospital video triage in NC London and East Kent.

The evaluation of these services represented an important opportunity to build an understanding of the acceptability and safety of prehospital video triage for suspected stroke patients. Key examples of potential learning were establishing the acceptability of the services to their users (i.e. ambulance and stroke clinicians); factors influencing usability (e.g. signal quality and environment); the impact of the service on patient destination, travel times and the delivery of clinical interventions; and factors influencing the implementation of these services (e.g. governance and training).

In 2020 and 2021, the authors engaged with ambulance and stroke services across the UK, establishing that, at the time, no equivalent prehospital video triage services were active. Since our evaluation commenced in July 2020, two other areas of Kent have launched their own prehospital video triage services and a number of other areas of the NHS in England have indicated that they are interested in potentially implementing a service of this kind.

Phase 1: review of existing reviews

First, we identified existing systematic, scoping, or rapid reviews on this topic published in the last five years using key words (e.g. telemedicine, telestroke, stroke, ambulance/paramedic, triage and review). The search was limited to peer-reviewed outputs published in English. We searched a range of databases and resources [e.g. The Cochrane Library, Association for Computing Machinery (ACM) Digital Library, Web of Science, Epistemonikos and PROSPERO]. This search was conducted in June–July 2020.

The findings from this phase are summarised in Chapter 3 and were shared with clinical collaborators in an umbrella ‘review of reviews’ paper. In addition, this review informed the design of the rapid systematic review conducted in phase 2 (thus helping to avoid any unnecessary duplication of existing reviews) and helped shape interview topic guides for the qualitative workstream.

Phase 2: rapid systematic review

The rapid systematic review built on the results of phase 1 and followed recent guidance and recommendations on conducting systematic reviews. 24 We aimed to identify conceptual frameworks and social science perspectives on the implementation of digital and communication technologies used for remote assessment and triage and assisting two-way communication between ambulance and stroke clinicians. In addition, we aimed to identify any knowledge and evaluation gaps, such as whether or not any economic impact analyses had been conducted about similar service innovations. We searched for an inclusive range of empirical studies [including randomised controlled trials (RCTs), feasibility studies, pilots, and service evaluations], using several relevant databases [including MEDLINE, Cochrane Library, Cumulative Index to Nursing and Allied Health Literature (CINAHL), EMBASE and Web of Science]. Our search terms were developed in collaboration with clinical co-authors and a librarian who specialises in literature searches. Search results were screened for duplicates and relevance against inclusion/exclusion criteria at title and abstract stages. The papers included in the review were then assessed for quality using the Mixed Methods Appraisal Tool (MMAT)25 and relevant Critical Appraisals Skills Programme (CASP)26 tools.

Data extracted from the included papers were then summarised in a table detailing the study type and methodology, population, outcomes and findings, and then written up using narrative synthesis27 and reported following Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) recommendations. 24 Data extraction was completed in July–August 2021.

Further details of this review can be found in Chapter 3 and the PROSPERO protocol for this review [URL: www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42021254209 (accessed 15 July 2022)].

Recruitment/consent

Participation in qualitative aspects of this service evaluation took place only with fully informed consent. Potential interviewees were sent an information sheet and had at least 48 hours to consider participation and ask any questions about the project. Interviews were conducted and recorded only with informed consent. Interviewees were free to withdraw at any time, up to and including during the interview. Participant details were anonymised for analysis and write-up.

For observations, meeting members were sent an information sheet in advance, alongside other meeting papers. The chairperson announced the evaluation team’s presence at the beginning of the meeting and confirmed that no names would be used in the final report. Although the situation did not arise, the evaluation team agreed that they would ‘dial out’ if issues were being discussed that those present at the meeting preferred not to be observed.

Data collection and analysis

Non-participant observations were conducted between August 2020 and May 2021. Interviews were conducted between October 2020 and May 2021.

Interviews were recorded using digital audio recorders and transcribed by a professional transcription service. Observations were recorded as handwritten or electronic field notes. Interview transcripts and observation field notes were imported into NVivo version 12.0.0.71 (QSR International, Warrington, UK) and analysed using framework analysis.

Emerging data (including from interviews, observations and documents) were captured using Rapid Assessment Procedures (RAP) sheets (see Report Supplementary Material 2). 21 These sheets supported the iterative nature of the evaluation by facilitating parallel data collection and analysis. The RAP sheet was structured to reflect the interview and observation guides and was refined over the course of the evaluation in response to the data. Team members involved in data collection met weekly to discuss progress and ensure consistency in data collection and analysis. The RAP sheets were updated regularly over the course of the study to facilitate quick and ongoing analysis and feedback with stakeholders. 28

Ambulance conveyance data

We analysed routinely collected ambulance journey data for NC London and East Kent, as provided by our partner organisations. These patient-level data were anonymised. They covered journey destination (e.g. which HASY or non-HASU the patient was conveyed to) and the timings of different stages of the ambulance job cycle (i.e. time from first call to dispatch, time from dispatch to scene, time at scene and time from scene to hospital). As outlined in Chapter 6, these data covered only the areas in which prehospital video triage was implemented during the period April–September 2020, that is, the months following the initial implementation of prehospital video triage. Therefore, a limitation of these data was that we had no comparators (either geographic or historic) against which to analyse performance.

National stroke clinical audit data

We analysed national stroke clinical audit data via the Sentinel Stroke National Audit Programme (SSNAP). We accessed publicly available data, aggregated at a team-level. We collected data covering teams in the participating areas (NC London and East Kent), and compared with figures for the rest of England (RoE) as a cross-sectional control. SSNAP aggregate data are reported quarterly and, at the time of analysis, only two quarters of post-implementation data were available: July–September 2020 and October–December 2020. Therefore, to permit understanding of whether performance changed over time, we analysed July–December quarterly data for 2018 and 2019 as historical controls.

Quantitative analysis

These data were analysed to assess the following outcomes.

Data requests

We requested local data via participating NHS organisations. We submitted a request for team-level aggregate SSNAP audit data, but the data were put in the public domain in sufficient time to permit our analyses.

Qualitative data

Stakeholder interviews were conducted only with fully informed consent. All interviews were recorded using encrypted digital audio recorders. They were transcribed by a professional transcription service, and data were transferred to and from this service via a secure File Transfer Protocol (FTP) system. Interview transcripts were pseudo-anonymised and stored on a secure drive to which only the evaluation team had access, located on the password-protected UCL system. All non-participant observations were recorded as handwritten or electronic fieldnotes and stored in the same secure UCL system.

Quantitative data

Ambulance and SSNAP data were stored as Microsoft Excel^® (version 2204, Microsoft Corporation, Redmond, WA, USA) spreadsheets on a secure drive to which only the evaluation team had access, located in the password-protected UCL system.

Survey

Survey data were collected via the secure UCL Opinio platform [version 7.18, UCL Information Services Division, London, UK; URL: www.ucl.ac.uk/isd/services/learning-teaching/e-learning-services-for-staff/e-learning-core-tools/opinio (accessed 27 July 2021)]. 37 Survey results were extracted from the software platform and stored on a secure drive to which only the evaluation team had access, located in the password-protected UCL system.

What was already known?

• There is a well-established evidence base on telestroke networks (which link stroke centres to rural or community hospitals using telemedicine) and mobile stroke units (MSUs) (highly equipped vehicles with CT scanners, laboratory equipment and telemedicine equipment, as well as stroke staff).

• Less is known about prehospital ‘mobile’ stroke telemedicine, which connects emergency clinicians/paramedics to hospital-based stroke clinicians.

What this chapter adds

• We conducted a systematic review into ‘mobile telestroke’ or ‘in-ambulance telemedicine’, a process that enables communication between ambulance crews and hospital-based doctors in cases of suspected stroke.

• Mobile telemedicine systems in stroke care have been piloted and found to be technically feasible. These systems enable two-way audio and video communication between emergency personnel and stroke clinicians, provided that there is good network coverage [e.g. fourth-generation broadband cellular network technology (4G)]. Indeed, reliable two-way audio and video communication is deemed especially important for remote clinical assessment of stroke.

• Several factors support successful implementation. These include staff training, compatibility with existing systems, clear communication, reliable network coverage, co-design with staff, high-performing microphones and a stable audio connection (especially to avoid clinicians having to repeat information or misunderstandings).

• Barriers to implementation include background noise, poor usability and weak or inconsistent network coverage.

• Few process evaluations or RCTs have been conducted on in-ambulance telemedicine for stroke compared with studies of telestroke networks and MSUs, especially with regard to the impact on destinations (including patients remaining at home in the case of mimics) and treatment times.

• Little is known about cost-effectiveness, patient privacy and patient experience in mobile telestroke in prehospital settings.

Phase 1: umbrella ‘review of reviews’ – a rapid, exploratory search

To guide this rapid evaluation and quickly identify key terms and relevant topics, a rapid search for existing reviews was undertaken from June–July 2020. According to the Joanna Briggs Institute (JBI) Manual for Evidence Synthesis, umbrella review permits consideration of multiple wide-ranging aspects of a given issue. 42 We deliberately broadened the search to quickly identify existing systematic, scoping, narrative or rapid reviews on the use of innovative technologies to support remote triage of stroke and suspected stroke patients by ambulance and paramedic crews on site.

The aims of the rapid umbrella review were to:

• identify relevant reviews on the topic (systematic and other types, e.g. scoping reviews)

• identify conceptual frameworks or theories used to understand the implementation of digital interventions in this context

• identify gaps in research or evaluation knowledge

• determine how our rapid evaluation and future research might address these gaps (and avoid duplication).

The following headings were used to guide the search:

• population – patients suspected of stroke, ‘stroke mimic’ or TIA

• phenomenon of interest – paramedic and ambulance crews (i.e. EMS clinicians) using technologies to triage and treat stroke (or suspected stroke) patients at the prehospital stage

• context – ambulance clinicians triaging patients in ambulances or in patients’ homes or care settings (i.e. non-hospital or hyper-acute clinical settings)

• outcomes – for example efficiency, safety and outcomes compared with usual practice and stroke management pathways.

Phase 2: systematic review

In addition to placing this evaluation and its conclusions in the context of a wider literature, it was decided to conduct a systematic review for four reasons. First, the most recent and relevant review identified in phase 1, Lumley et al. ,16 focused on a variety of prehospital interventions in the stroke pathway (e.g. biomarkers), whereas we were interested primarily in a very specific intervention covered in that scoping review: ‘mobile telemedicine’. This meant that we needed to focus specifically on technologies that enabled communication between ambulance clinicians and stroke clinicians and centres. Second, following discussion of the phase 1 search results and new guiding questions with an information specialist at UCL, the published search strategy used in Lumley et al. 16 was discussed and it was noted that there were some medical subject index headings (MeSH) and key words that had not been included but which would be important in the context of this particular evaluation (e.g. we added terms such as ‘bidirectional communication’, ‘iPad’, ‘clinical informatics’ and ‘HASU’). In addition, Lumley et al. 16 had framed part of their review on process outcomes (e.g. ‘detect’ and ‘diagnosis’) and the information specialist advised that this may have missed some literature of relevance to this review. We therefore wished to employ a fine-grained search strategy focused on mobile telemedicine, the prehospital emergency context and communication between health professionals, requiring a modified design. We also wanted to identify any social science or human–computer interaction literature associated with this topic to provide an interdisciplinary focus, requiring new terms to be added (see Report Supplementary Material 3 for the detailed MEDLINE search we developed). Third, the phase 1 search revealed a relatively large literature on MSUs. Because we were interested in evidence about potentially lower-cost digital technologies that support two-way communication in the field, we required a different strategy from that for other reviews, and to go beyond the literature on MSUs. Finally, as the two pilots evaluated here were implemented in the context of COVID-19, it was important to capture any more recent studies that had been implemented and written up during the last year (2020–21), although it was noted that, owing to publication time delays, more recent empirical studies and reports might be small in number.

The final systematic review search strategy involved close working between a researcher (JL) and a UCL information specialist (Debora Marletta, UCL librarian) and extensive trialling of key terms and Boolean operators across multiple databases. The review was guided by PRISMA 2020 recommendations24 to ensure that the approach taken was both transparent and explicit, and a final protocol was published on PROSPERO following team discussion and review. 43 Collaboration between the researcher and the information specialist was essential to maximise the sensitivity of the search and to sense-check returns, helping ensure that the final records included insights about the implementation of digital and communication technologies within prehospital acute stroke pathways. The information specialist checked all final search strategies that had been refined until consensus was reached. In addition, a ‘test list’ of potentially eligible papers was used drawing on phase 1 in order that JL could confirm that the final search strategy was picking up relevant papers.

We developed a specific title and research questions for the systematic review in autumn 2020, which were discussed and agreed on as a team. The phase 2 review aimed to identify implementation factors associated with digital and communication technologies that could impact on health-care service quality and health outcomes, and any conceptual frameworks and social science perspectives that could help explain these relationships. The search strategy was framed around the participants, intervention, and health context and setting, but not limited by outcomes in order to capture the wider literature. Instead, we were interested in a variety of outcomes of interest, such as implementation factors, safety and security of digital and communication systems, usability, user experience, acceptability (staff and patients), cost-effectiveness, overall clinical effectiveness, and ambulance destination outcomes (e.g. faster treatment times, but also issues of usability, staff experience, enhanced communication and safety). We also wanted to confirm any evaluation gaps, such as whether or not economic analyses had been conducted.

The final review title was ‘The implementation of digital interventions and communication technologies in emergency care pathways to support the remote assessment and triage of patients suspected of stroke: a systematic review’.

The review questions were as follows:

• Which human, technological, and usability factors are associated with the implementation of digital and communication technologies in ambulance settings (e.g. mobile telemedicine) that enable two-way communication between paramedics and hospital physicians in the care of stroke and potential stroke patients?

• Which system and local contextual factors are important for the implementation of these kind of technologies in the care of stroke and potential stroke patients?

• What service-level outcomes (e.g. clinical, financial and resource impacts) are associated with the use of these kind of technologies in the care of stroke and potential stroke patients?

• What evidence exists about the safety and security of these kind of technologies and systems when used at the prehospital admission stage? Relatedly, are any adverse effects reported in the literature?

• What (if any) conceptual frameworks have been used to understand the implementation of these kind of technologies and systems?

Database search strategies and specific operators were developed and piloted by a researcher and information specialist during the first part of 2021. Six databases were used for the systematic review:

1. the Cochrane Systematic Reviews Database

2. The University of York Centre for Reviews and Dissemination [covering the Database of Abstracts of Reviews of Effects (DARE), the NHS Economic Evaluation Database (NHS EED) and Health Technology Assessment (HTA) database]

3. MEDLINE (via Ovid)

4. CINAHL

5. EMBASE (1980–present)

6. Web of Science Core Collection.

The final searches were run across the databases from July to August 2021. Searches were saved so that the main reviewer received automatic alerts about new records. Report Supplementary Material 3 provides details of the search strategies executed across a selection of these databases.

Data extraction and screening

A total of 4577 records were downloaded from the databases into EndNote [version 20, Clarivate Analytics (formerly Thomson Reuters), Philadelphia, PA, USA], which was used to identify duplicate entries (Figure 2). The process used for deduplication was recommended by the information specialist. 45 Duplicates were automatically identified by EndNote and manually checked by one researcher (JL) using this method. Records were next uploaded to the software Rayyan.ai for screening of titles and abstracts, a package that also facilitates team collaboration. To ensure reliability in the screening of titles and abstracts, a screening tool was developed (see Report Supplementary Material 4) and piloted on the first 200 records, and the results were discussed with a second reviewer (AIGR). This led to a validated final screening protocol for searching all titles and abstracts, which was led by one reviewer (JL). If any papers were identified in which eligibility was unclear, the title and abstract were discussed by two reviewers and an agreement was reached. This process was important for screening a large number of papers that addressed telestroke ‘hub-and-spoke’ configurations, but excluded discussion about emergency care pathways and ambulance/paramedic teams.

FIGURE 2.

Systematic review: PRISMA flow diagram. Based on PRISMA 2020 – flow diagram for systematic reviews which include searches of databases and registers only). 44 CRD, Cochrane Systematic Reviews Database; WOS, Web of Science Core Collection.

One reviewer (JL) identified the 288 records for review against the inclusion criteria and validated screening tool. This resulted in the identification of around 60 relevant reports for inclusion. A second reviewer (AIGR) was involved in screening this final group, using Rayyan.ai. 46 Any reports where it was uncertain whether or not they should be included were discussed until consensus was reached. For example, a number of studies incorporated a prenotification telephone call from emergency clinicians to stroke clinicians or units; these were excluded following discussion because they did not involve the explicitly use of video or two-way communication to support remote triage and diagnosis. Two reviewers (JL and AIGR) approved the final group of 47 included reports.

Phase 2: systematic review

The systematic review identified 47 papers that met the inclusion criteria (32 primary studies39,41,59–88 and 15 reviews16,39,40,47,48,53,89–97; for details, see Appendix 1, Table 11). One record was a trial registration;98 this led to the identification of a research paper,80 which was included in place of the trial registry record. The publication date ranges suggested a fairly slow research trajectory, with signs of growth from 2019 onwards. The majority of studies originated in the USA and Germany, with a small number of studies found from across Scandinavia, Asia and mainland Europe. The UK studies were based in Scotland and no studies were based in the NHS in England. Below we provide a summary of the findings. We included other reviews to ensure that no underlying studies were missed, to provide continuity with phase 1 and to confirm any evidence and knowledge gaps.

Principal findings

Our review confirmed a limited, yet growing evidence base about telemedicine use within ambulances for the diagnosis of stroke. This can be viewed as a growing field of research compared with a traditional focus on MSUs, telestroke networks and other hospital-to-hospital telemedicine systems, which are common in the USA and Europe where rural areas require access to neurological expertise that is otherwise unavailable.

This systematic review provided an important update on a scoping review by Lumley et al. 16 identified in phase 1. We conducted a systematic review containing more key terms and medical subject headings to hone in on the emerging topic of within-ambulance mobile telemedicine (Lumley et al. 16 included a number of other innovations in their scoping review, such as biomarkers, which was not the focus of our study). Despite not including conference proceedings, we were able to identify new studies as our search was highly sensitive and extended beyond 2019 to 2021. As with Lumley et al. ,16 a meta-analysis was not attempted owing to the diversity of studies found. We also included existing literature reviews to get a wider overview of the topic and growth in the field and to ensure that we captured all relevant studies.

Our systematic review found limited data on the impact on outcomes, although these, including data showing the impact on treatment and destination times, are emerging. However, there was generally limited information about other service outcomes of importance to understanding prehospital video triage, such as the impact on decision-making, patient destination and avoidance of unnecessary hospital admissions (thus helping reduce demands on hospital services).

It has already been observed that ‘ambulance-to-hospital teleconsultation may result in markedly reduced delivery times for thrombectomy without delaying intravenous thrombolysis’;84 further research and evaluation is required to support this assessment, although there are early indications of positive impacts on thrombolysis and destination times from a small number of observational and cohort studies. Yet, clearly, more insights are needed about not just transfer and treatment times, but destination end points within the stroke care pathway. With regard to the technical specifications and usability of these systems, there was a gap concerning the influence of different professional roles, such as whether there are advantages of ‘paramedic–neurologist’ communication over ‘paramedic–nurse’ communication. Joseph et al. 66 provided an especially helpful study into communication between different professional groups, taking a human factors and task-based approach and highlighting teamwork and a number of practical considerations (e.g. background noise).

There was a lack of insight about the overall safety of these systems (although clinical safety is covered by reference to the use of diagnostic scales and stroke protocols, e.g. FAST and NIHSS). We found one study on overall cost-effectiveness that confirmed what some authors have noted elsewhere: that mobile telemedicine in ambulances is a lower-cost option than introducing MSUs. Notably, there were few RCTs, although a number of studies attempted comparisons with standard acute stroke care and emergency pathways. Finally, more needs to be understood about the impact of different health system models and stroke service configurations on implementation. For example, in some cases designated telestroke centres may field all calls from paramedics/EMS clinicians; in other cases, it may be neurologists based within specific hospitals. The impact of different service models, therefore, requires further investigation.

Strengths and weaknesses

Strengths of the review included a transparent and explicit search strategy, specifically following PRISMA for Scoping Reviews (PRISMA-ScR) 2020 guidelines in phase 2 where multiple databases were searched by an experienced researcher with expert information specialist input. Collaboration was used to design and execute a systematic search strategy (i.e. with tailoring to the different databases). We developed and validated a screening tool to operationalise the inclusion and exclusion criteria with a second reviewer sense-checking findings and final inclusion decisions through a process of using software (Rayyan.ai’s collaboration function) and meeting to discuss interpretations. The wider team and co-investigators also provided ongoing advice and feedback into the design and findings.

Owing to the timeframes of this rapid evaluation, no risk-of-bias assessment or detailed critical appraisal was performed, although this will take place prior to publication of the systematic review in a peer-reviewed journal, following PRISMA-ScR 2020 recommendations. 24 Given the limited evidence on this topic, the review could be further strengthened by undertaking additional steps that were not feasible within the timeframe of a rapid evaluation (e.g. additional citation searching and checking pre-print servers and recent conference proceedings). In this chapter, we have summarised the main findings with a focus on the issues most relevant to the empirical findings and conclusions of this report, and the most recent empirical evidence found.

Implications

There were several implications from this search of the literature, especially from the findings from the systematic review. First, mobile telemedicine was found to involve some examples of well-equipped ambulances with high-definition cameras and audio equipment, and there are suggestions in the literature that these may mitigate background noise and capture non-verbal cues, which are important in stroke teleconsultations. Given a small but growing evidence base in this area, there is need for further evaluations and clinical trials of these systems that take into consideration the specific equipment used, its functionality, its costs and the implementation factors that shape usage, for example understanding which contexts are optimal for performing remote triage and assessment and which are not, and why. Second, only one cost-effectiveness analysis was identified in the systematic review, suggesting a need for more economic studies, especially given the variations found between health systems internationally and regionally and different system set-up costs. Finally, there was a lack of detailed exploration about the patient experience, with most studies focusing instead on the perspectives of health-care professionals and advantages for health providers.

Overall, we found that less is known about stroke teleconsultations that are on scene (such as in a person’s home), away from a more highly equipped ambulance, and involve emergency clinicians and remote stroke neurologists. In particular, more knowledge is needed about how prehospital telemedicine systems affect stroke destination decisions (including the identification of non-stroke patients or ‘stroke mimics’), and this may help to ensure that patients are directed to the most appropriate point of care or service. This is also necessary given the potential of mobile telemedicine to help with managing high levels of demand on stroke and emergency services, and the possibility of costs savings (e.g. by reducing unnecessary ambulance journeys and diagnostic scans). Very few studies discussed in any detail bypassing certain hospitals as a result of remote triage decision-making between ambulance and stroke clinicians.

More understanding is also required about the contribution of mobile telemedicine over prehospital stroke notification systems that involve a stroke alert to the hospital and data transmission (e.g. vital signs). Future studies may also wish to consider which types of service outcomes are most appropriate for measurement when evaluating pilot services. For example, there may be only marginal gains of using mobile telemedicine for prehospital triage if onset-to-treatment times are already excellent and occur within a well-specified destination pathway. As one paper80 noted, there has been much attention on studying and improving onset-to-treatment times for decades, yet only in recent years has there been a focus on prehospital stroke care, including how much time is spent on scene by emergency clinicians.

What was already known?

• Prehospital video triage for suspected stroke patients was introduced in early 2020 in NC London and East Kent.

• Gaps exist in the evidence related to patient outcomes, experience and satisfaction, communication between professionals, and impact on decision-making.

• Little use has been made of social science concepts to understand such services.

What this chapter adds

• There were contrasts in aspects of the triage services and how they were implemented in both areas. However, stroke and ambulance clinicians in both areas saw the interventions as highly usable and acceptable. A barrier to acceptability and sustainability was that acute stroke physicians delivered triage assessments alongside their other professional duties, placing pressure on them and potentially limiting the quality of communication.

• Leaders of the triage services were conscious of risks to patient safety and set processes in place to monitor and manage potential adverse incidents. Local data collected through these processes indicated few reported safety incidents, and there was a broader perception among ambulance and stroke clinicians that these services are providing safe care.

• Several factors relating to theory about implementation and sustainability of innovations (including contextual factors, innovation characteristics, and implementation approaches) helped explain what drove change rapidly, how widespread ownership of the services was achieved, and which further actions might facilitate sustainability of these services.

• In terms of context, the COVID-19 pandemic created a ‘burning platform’ for change, which interacted with long-established challenges around appropriate conveyance of suspected stroke patients. These factors combined to encourage more facilitative governance processes at local system and service levels.

• In terms of innovation characteristics, stroke and ambulance clinicians found prehospital video triage straightforward to use, suggesting it offered important advantages over conventional approaches to assessing and conveying suspected stroke patients.

• In terms of implementation approaches, service leaders worked across organisational and professional boundaries to ensure rapid and effective development of the services, approval by local organisations and systems, and uptake by frontline clinicians. Training (in terms of its availability and intensity) was seen as an important influence on clinicians’ awareness of and confidence in using prehospital video triage.

Overview of findings

Our results are organised as follows. First, we present the data collected and analysed in this analysis. Second, to address important gaps identified in the literature, we present the analysis of ambulance clinician and stroke clinician perspectives on the usability, acceptability, and safety of prehospital video triage. Third, we describe some factors related to implementation of these services, guided by the Nolte framework17 for understanding implementation of digital innovations in health-care settings (see Figure 3). Throughout, we discuss issues that appeared specific to clinician groups and areas, but also draw out common themes.

Data analysed

We analysed interviews conducted remotely with 27 local stakeholders in NC London and East Kent (Table 4). Interviewees included stroke consultant physicians and ambulance clinicians who had experience of using prehospital video triage, and local clinical leads of the prehospital video triage services based in ambulance and stroke services in both areas. We analysed observation data on 9 events, including governance meetings, training sessions and engagement events (which we attended remotely). Finally, we analysed 23 documents, including meeting minutes, service pathways and protocols, and training documentation.

Clinician perspectives on usability

Ambulance and stroke team clinicians in both areas reported overall that they found prehospital video triage usable, in terms of the platform used (FaceTime), reliability of connection, and sound and image quality.

Clinician perspectives on acceptability

Clinician perspectives on patient safety

Interviewees reported relatively few concerns about patient safety. The main (and interrelated) considerations identified related to whether conducting the prehospital video triage assessment contributed to unwarranted delays in patients receiving care and whether patients were conveyed to the appropriate service.

Communications and decisions

Communication was central to prehospital video triage. This reflected the fact that (without access to brain scanners in the ambulance) both ambulance and stroke clinicians had limited tools available to aid diagnosis:

I think the FAST test was quite basic and that’s all the kind of . . . We don’t have any other tools to kind of establish what’s going on.

ST008, ambulance clinician, NC London

The key interactions were between stroke and ambulance clinicians, and between clinicians and patients.

Factors perceived as influencing implementation of prehospital video triage

In this section, we discuss factors that emerged as important to the implementation and delivery of prehospital video triage in NC London and East Kent. This is arranged around the factors identified by Nolte (Figure 4). 17 In each case, we consider how aspects of these factors (e.g. relative advantage in relation to ‘innovation characteristics’) might help explain the implementation and impact of prehospital video triage.

FIGURE 4.

Overview of factors perceived as influencing implementation. Italic text covers obstacles to implementation; roman text covers facilitators. Adapted from Nolte E, How Do We Ensure that Innovation in Health Service Delivery and Organization is Implemented, Sustained and Spread?, Box 2, Page 11, Copyright (2018) WHO Regional Office for Europe. URL: https://apps.who.int/iris/bitstream/handle/10665/331980/Policy-brief-3-1997-8073-2018-eng.pdf?sequence=5&isAllowed=y (accessed 1 September 2022). 17

Adapted from Nolte E, How Do We Ensure that Innovation in Health Service Delivery and Organization is Implemented, Sustained and Spread?, Box 2, Page 11, Copyright (2018) WHO Regional Office for Europe. URL: https://apps.who.int/iris/bitstream/handle/10665/331980/Policy-brief-3-1997-8073-2018-eng.pdf?sequence=5&isAllowed=y (accessed 1 September 2022). 17

Principal findings

There were several contrasts in the prehospital video triage implemented in NC London and East Kent, and in how they were put into action. However, there were many common findings on usability, acceptability and safety. The first of these concerned the usability of prehospital video triage, in terms of the platform used, reliability of connection, and sound and image quality: stroke and ambulance clinicians in both areas saw the services as highly usable. Clinicians reported that the processes concerned did not represent a significant change to how patients were assessed, and the technology used was perceived as familiar and straightforward. Ongoing governance processes permitted rapid adaptation to address obstacles to usability, for example by enabling access to local Wi-Fi or providing headphones to reduce environmental distractions.

In terms of acceptability to users (i.e. ambulance and stroke clinicians), both clinician groups found the services acceptable (despite concerns that ambulance clinicians may have been reluctant to lose decision-making agency, we only found pockets of this in East Kent, where responsibility shifted more firmly to stroke physicians). A potential obstacle to clinician acceptability was that stroke physicians found delivering assessments alongside their other professional duties a significant challenge. In terms of safety (as perceived by ambulance and stroke clinicians), the new services represented an important change to the patient pathway in both areas. Therefore, service leads were highly conscious of risks to patient safety and set processes in place to monitor and manage any incidents. Observations of local governance meetings suggested these reviews identified very few safety incidents, and each incident was analysed to support further development of the service.

Several factors related to theory about implementation of innovations (innovation characteristics, contextual factors and implementation approaches) helped explain what drove the decision to change, how development progressed so rapidly, how widespread professional ownership of the services was achieved and which further actions might facilitate greater sustainability of these services. Key among these were national and local contextual factors, characteristics of the innovation itself, and the implementation and dissemination approaches employed.

Contextual factors (at system and local levels) played an important and interrelated role in implementation and delivery of prehospital video triage. The ‘burning platform’ of the pandemic and longstanding challenges of appropriate conveyance of suspected stroke patients combined to form a highly receptive local context for change. This was reflected in more facilitative governance processes and high local buy-in for change. In addition, other service changes implemented locally to manage the pandemic (specifically, moving stroke units so that they were no longer co-located with ED services) may have been important levers to increase the uptake of prehospital video triage. There was a strong influence of adopter characteristics: clinicians’ and local systems’ prioritisation of quality and safety of care facilitated rapid uptake by frontline clinicians. It also ensured that processes to monitor and address potential adverse incidents or other issues were central to governance of the pilots. Organisational antecedents also played a role: limited resources in the form of stroke team capacity represented a significant source of clinician stress; this may have important implications for the sustainability of prehospital video triage services.

The innovation characteristics of prehospital video triage were viewed positively by their intended users. Stroke and ambulance clinicians found prehospital video triage straightforward to use and reported that it offered important advantages over conventional approaches for assessing and conveying suspected stroke patients.

Implementation and dissemination were pivotal to the development and delivery of prehospital video triage. Clinical leaders took active and collaborative approaches to negotiate and drive change: they worked across several boundaries, including organisational (working beyond local acute and ambulance trusts to engage the wider system) and professional (with close collaboration between stroke and ambulance clinicians in both areas). This supported rapid development, approval and implementation of the prehospital video triage services. Regular governance meetings played a key role in monitoring delivery of prehospital video triage and providing important assurance and learning opportunities about patient conveyance and patient safety. Contrasts in training approaches employed in NC London and East Kent suggest that more active approaches were preferred by staff and may act to form stronger interprofessional collaboration in future.

Strengths and weaknesses

A strength of this analysis was that we were able to study two similar innovations implemented at the same time in two regions of the NHS in England . However, there were also several contrasts, including patient eligibility, decision-making processes and training approaches. Therefore, this evaluation represented a ‘natural experiment’, and was thus a valuable learning opportunity. Several features of the cases evaluated may make them relevant to other parts of the NHS in England. The innovation itself relied on established technology (i.e. iPads and FaceTime) that are increasingly common in NHS ambulance services. In terms of contexts, we were able to study implementation of prehospital video triage in NHS settings during the COVID-19 pandemic in contrasting urban and rural settings. Another important strength of this analysis was that it was coproduced with stroke and ambulance leads in both pilot areas but also informed by an up-to-date review of the evidence, meaning that the focus of the evaluation reflected both local expertise and the wider evidence base. Using remote data collection methods [i.e. Microsoft Teams^® (Microsoft Corporation) videoconferencing for interviews] increased the evaluation team’s flexibility, meaning that interviews could be conducted or reorganised to accommodate the requirements of extremely busy clinicians.

There were several limitations, chiefly related to data collection. We were unable to recruit similarly large numbers of clinicians in both areas; this meant that our sample from East Kent might not reflect the same range of clinician perspectives as in NC London. Furthermore, we only interviewed stroke consultant physicians, and future research would benefit from capturing the perspectives of less senior doctors who also delivered prehospital video triage. There were several other clinical groups we were unable to interview for this evaluation. Towards the end of the project, we approached a number of nursing and ward management staff in the hope of obtaining a broader perspective on the impact of prehospital video triage on stroke teams. However, given wider pressures and the limited time available, we were unable to conduct any of these interviews, and therefore to gain their insights we shared our interim findings with these individuals for their feedback. To mitigate the gaps in our clinical stakeholders, we shared interim findings with a wider range of clinical stakeholders in NC London and East Kent, all of whom felt that the themes presented rang true. Finally, we were unable to interview patients and carers about their experiences of these services; this was due to our project’s status as a rapid service evaluation. We prioritised obtaining clinicians’ views of patient experience, but recognise this as a key limitation that flows from the nature of projects of this kind.

Comparison with other studies

As outlined at the beginning of this chapter (and in Chapter 3), there is a small but growing evidence base related to the implementation and outcomes of prehospital video triage for suspected stroke. We believe that our analysis both complements and extends this evidence base.

Previous research on acceptability of prehospital video triage for stroke has described staff being satisfied in terms of the quality, safety and reliability of services. 16 Our analysis confirmed broad satisfaction with services while also noting some important underlying benefits. Examples of benefits reported by ambulance clinicians were educational (i.e. learning more about the nature of stroke) and improved experience of delivering this care pathway (i.e. feeling more confident that hospital services would not query the patient conveyance decision).

Our findings on usability aligned with previous research in this context. For example, our analysis confirmed the previously-identified importance of signal stability and audio-visual quality to ensuring a clear assessment process. 16 However, whereas previous simulation-based research has focused recommendations on improving audio-visual technology,69 some of the interventions identified in our analysis (e.g. the value of an audio headset or a separate office to reduce ambient distraction) may reflect the advantages of studying service delivery in the (frequently overcrowded and noisy) real world.

Our evidence on interprofessional communication contrasts somewhat with previous research on prehospital video triage for stroke. Whereas past work has presented the role of the paramedic as ‘the ears and eyes of the neurologist’,66 in our analysis we found a strength in a more balanced dynamic. Placing ambulance clinicians’ evaluations at the heart of the overall patient assessment was felt to increase buy-in from these clinicians; furthermore, it gave stroke clinicians an increased appreciation of ambulance clinicians’ expertise.

Previous research has commonly noted the importance of training ambulance clinicians to ensure effective use of prehospital video triage. 16 Our analysis confirmed the contribution training made to clinicians’ confidence in using the system. In addition, some participating ambulance clinicians also described the potential for joint training to bridge interprofessional boundaries between stroke and ambulance clinicians. This issue, related to engendering shared ownership of such services, aligns with findings from previous reviews on implementing telestroke networks. 15

Our analysis demonstrated the value of applying social science theory to the analysis of implementation factors, which our systematic review (see Chapter 3) found to be a gap in the literature. Nolte’s framework17 helped us identify several important factors related to contextual factors (at national, regional and service levels), implementation approaches (e.g. collaborative leadership, governance of change and training approaches), and the innovation itself (i.e. its perceived straightforwardness and advantage over ‘business as usual’).

Implications

Prehospital video triage as implemented in NC London and East Kent was seen as usable and acceptable by clinicians. However, greater capacity (i.e. more stroke clinicians available to conduct assessments) may increase the acceptability, experience and sustainability of these services. Rapid implementation of change was enabled by the global pandemic (its effects were also likely to have been strengthened by increased facilitation of change by local governance processes at system and service level). Clinical leadership in both stroke and ambulance services was key to implementation. The collaborative approaches taken by these leaders to work across organisational boundaries and to engage frontline clinicians (e.g. by designing processes that reflected professional expertise and where training ensured clear understanding of the new services) was important in ensuring uptake of the new services.

What was already known?

• Prehospital identification of stroke can be challenging. Prehospital video triage may improve identification of stroke, thus supporting more appropriate patient conveyance decisions, better-informed stroke teams in receiving hospitals and by extension improved time to treatment and better patient outcomes.

• There is little published evidence on how ambulance clinicians experience prehospital video triage and what influences their decision-making; furthermore, the majority of such research has been conducted in simulation (as opposed to ‘real-world’) settings.

• Our qualitative analysis (see Chapter 4) included a limited number of ambulance clinician perspectives, but identified several key themes related to prehospital video triage, including its usability, influence on conveyance decisions and patient safety, and implementation.

What this chapter adds

This chapter presents a survey of 233 ambulance clinicians who had experience of using prehospital video triage in NC London and East Kent. Responses identified several factors related to major themes of this evaluation, including usability, impact on conveyance decisions and patient safety, and implementation (specifically training); ambulance clinicians also provided views on the future use and development of prehospital video triage:

• Usability – although ambulance clinicians in both areas found prehospital video triage usable, a higher proportion of NC London clinicians were of this view. This pattern was mirrored in clinician views of audio-visual signal, where again NC London clinicians were more likely to agree that signal was of sufficient quality. Underlying these patterns may be differing likelihood of experiencing difficulties with internet/Wi-Fi coverage in the studied areas.

• Conveyance decisions – ambulance clinicians in both areas felt that prehospital video triage influenced conveyance decisions occasionally.

• Patient safety – NC London clinicians were more confident about the safety of prehospital video triage than clinicians in East Kent. Safety concerns included potential for misdiagnosis and delayed or inappropriate conveyance.

• Training – ambulance clinicians in NC London were much more likely to feel that they had received sufficient training. This is likely to reflect the different training approaches employed in NC London and East Kent.

• Future use and development – ambulance clinicians overall saw prehospital video triage as an improvement on previous prehospital approaches and wanted it to continue. Again, this was felt more strongly in NC London than in East Kent.

Survey tool development

The survey was developed by members of the evaluation team (led by JL) alongside clinical collaborators in NC London and East Kent. It was designed to be short and as easy as possible to complete, to reduce burden on participating clinicians (who were at the time dealing with substantial service pressures during the COVID-19 pandemic) and thus maximise survey completion.

Examples of collaborator feedback included clarifying items on conveyance decision-making and training. The final version of the survey was tested and signed off by clinical collaborators and built using the software package Opinio (which is hosted on the UCL information services infrastructure and meets UCL information governance standards).

The survey consisted of 18 questions, including consent, and was designed to take approximately 10 minutes to complete (see Report Supplementary Material 5 for the final survey questions). Most questions were multiple choice, with a ‘free text’ option for some questions, for example those covering patient safety, training and recommendations for future development.

Sampling

Our sample was ambulance clinicians who had experience of using prehospital video triage for stroke, piloted in NC London and East Kent. To ensure that respondents had some experience of using prehospital video triage, we asked about frequency of use of the system. Beyond this, our survey used a convenience sample of ambulance clinicians based in the NC London and East Kent areas. As a result, the extent to which our findings can be generalised to other ambulance clinicians and services piloting such services in other clinical or geographic contexts is limited.

Distribution of survey and data collection

The survey link was distributed to 550 ambulance clinicians who had used prehospital video triage in NC London, and to 424 ambulance clinicians in East Kent. Our clinical collaborators sent potential participants a survey invitation by e-mail; the e-mail included a participant information sheet and a link to the survey. Reminders were sent out at regular intervals to promote survey participation. In East Kent, the survey was also promoted to staff on a private Facebook (Facebook, Inc., Menlo Park, CA, USA) page.

Informed consent was taken at the beginning of the survey. The survey invitation made it clear that the survey should only be completed by ambulance clinicians who had used the prehospital video triage services (e.g. specifying FaceTime on iPads). To ensure confidentiality, no names, e-mail addresses or other personal data were collected. Respondents were therefore able to complete the survey completely anonymously using the link provided. Participation was on a completely voluntary basis.

The survey was left open for just over 5 weeks, from 21 April to 30 May 2021.

Data analysis

The raw survey data were downloaded to Microsoft Excel for analysis; we used the ‘pivot table’ function to collate overall scores and percentages for the overall sample and disaggregated by participating areas (i.e. NC London and East Kent). For each survey item, a chi-squared test of independence was performed on the raw data to compare the patterns of responses from NC London and East Kent ambulance clinicians.

Response rates

A total of 309 staff accessed the survey, but 76 responses were found to be incomplete and therefore removed from the sample. There were 233 total respondents across both regions, with 159 fully completed responses in NC London (representing a response rate of 159/550, 28.9%) and 74 in East Kent (representing a response rate of 74/424, 17.5%).

Overview of findings

Detailed survey responses are presented in Appendix 1, Table 12. In the sections that follow, we report in-depth survey responses, covering ambulance clinicians’ perspectives on the following topics:

• clinician experience and usage of prehospital video triage

• usability of prehospital video triage

• technical performance of prehospital video triage

• responsiveness of stroke teams

• influence on conveyance decisions

• impact on patient safety

• sufficiency of training

• future use and development of prehospital video triage.

Clinician experience and frequency of use

As noted in Appendix 1, Table 12, the majority of respondents across both regions had ≥ 3 years’ experience of working as ambulance clinicians (76%, n = 177) and there was no significant difference between the levels of experience reported in the two areas [χ²(3, N = 233) = 3.40; p > 0.05]. Usage patterns were broadly similar across both regions at an aggregate level, with the majority (93.1%) of all respondents having used the system three or more times. However, there was a significant difference in the pattern of responses from NC London and East Kent ambulance clinicians [χ²(3, N = 233) = 12.97; p < 0.01]: Figure 5 shows that a higher proportion of respondents in East Kent (49%) reported using the triage system 10 times or more than in NC London (26%).

FIGURE 5.

Frequency with which respondents reported using prehospital video triage. Baseline, N = 233 (LAS, n = 159; SECAmb, n = 74).

Usability of prehospital video triage

The majority (89%) of respondents in both areas reported that they were satisfied with ease of use of prehospital video triage. However, the pattern of responses in NC London and East Kent differed significantly [χ²(3, N = 233) = 35.33; p < 0.01]: Figure 6 shows that although most respondents in both areas agreed, strong agreement was higher in NC London (94%) than in East Kent (78%, with 16% either disagreeing or strongly disagreeing). It is possible that a number of factors (discussed in Chapter 4) may have contributed to this, including signal strength and coverage, image/sound quality, responsiveness of stroke clinicians, and training. In the following sections, we present ambulance clinicians’ views on all these potential contributing factors.

FIGURE 6.

Proportion of respondents who felt prehospital triage was easy to use. Baseline, N = 233 (LAS, n = 159; SECAmb, n = 74).

Technical performance of prehospital video triage

The majority of respondents across both regions (77%) found the sound and video quality sufficient for the task of remote triaging suspected stroke patients. The pattern of responses in NC London and East Kent differed significantly [χ²(3, N = 233) = 17.48; p < 0.01]: the level of agreement was higher in NC London (84%) than in East Kent (62%). Where technical problems had been encountered (see Appendix 1, Table 12), the most common issue reported among all respondents was poor sound quality, followed by calls not being answered. In NC London, 37.7% of respondents reported no issues with the system, while in East Kent 18.9% reported no issues.

Thematic analysis of 22 ‘other’ text comments indicated system performance issues centred around three main areas:

1. lack of signal/connectivity (e.g. unable to connect to FaceTime application and poor video and sound)

2. background noise (at the HASU end)

3. iPad device tricky for some older patients to engage with (e.g. those with dementia).

Responsiveness of stroke teams

The majority of all respondents (91%) reported that their call was answered by a stroke clinician after one or two attempts. However, the pattern of responses in NC London and East Kent differed significantly [χ²(3, N = 233) = 33.02; p < 0.01], with 98% of NC London respondents requiring one or two attempts, while 77% of East Kent respondents required one or two attempts, and 23% of East Kent respondents required more attempts (Figure 7). When asked what they did in instances where they did not receive an immediate response from a stroke doctor, the majority of respondents in both areas (overall 54%) suggested that they would attempt contact up to two times, then revert to the standard patient conveyance procedure. A minority in both areas (6.9% in NC London and 18.9% in East Kent) reported that they would keep trying until the call was picked up.

FIGURE 7.

Number of attempts to have call answered by a hospital stroke team. Baseline, N = 233 (LAS, n = 159; SECAmb, n = 74).

Free text responses were provided by the 24% who responded ‘other’. Many used this to report that calls were picked up straight way or that respondents had not encountered any issues (e.g. ‘Always had response’ and ‘We have been rung back almost immediately’). Others described reverting to the usual conveyance process, for example if the patient was FAST positive (e.g. ‘Depending on how critical or how obvious the stroke is I will only try a maximum of twice before conveying’).

Influence on conveyance decisions

A majority of respondents reported that triage influenced decisions about where to convey patients (e.g. convey to HASU or local ED, referral to TIA service, or remain at home) infrequently: this suggests that prehospital video triage tends to confirm initial stroke-related assessments by ambulance clinicians based on FAST. However, the pattern of responses in NC London and East Kent differed significantly [χ²(3, N = 233) = 8.72; p < 0.05]: Figure 8 suggests that a larger proportion of NC London respondents reported that prehospital video triage influenced decisions frequently, while a larger proportion of East Kent respondents felt prehospital video triage never influenced their decisions.

FIGURE 8.

Frequency that respondents felt prehospital video triage influenced conveyance decisions. Baseline, N = 233 (LAS, n = 159; SECAmb, n = 74).

Impact on patient safety

A clear majority of respondents had no concerns about the safety of prehospital video triage. However, the pattern of responses in NC London and East Kent differed significantly [χ²(3, N = 233) = 30.48; p < 0.01]: Figure 9 shows that 91% of NC London respondents reported no concerns about safety, compared with 62% in East Kent. This suggests that more ambulance clinicians in East Kent than in NC London had concerns about the safety of prehospital video triage.

Thematic analysis of 35 free-text responses submitted alongside this question highlighted four overlapping safety concerns. Below, we provide quotations to illustrate each theme.

FIGURE 9.

Proportion of respondents who had concerns about patient safety. Baseline, N = 233 (LAS, n = 159; SECAmb, n = 74).

First, ambulance clinicians noted concerns about delays to treatment, either through time spent on scene or through inappropriate decisions:

Occasionally the person who answers will not listen to a handover or look at the pt [sic – ‘patient’] and will make a decision that is not suitable. This leads to pts [sic] being conveyed to inappropriate hospitals and then delaying treatment.

Ambulance clinician, East Kent

My concern is that conveyance is delayed in order to make the call therefore shortening the window for treatment.

Ambulance clinician, East Kent

I feel sometimes this could delay patient care, time is brain after all.

Ambulance clinician, NC London

The process hinders efficient conveyance, the line is hard to hear, it questions our judgement as clinicians.

Ambulance clinician, East Kent

Second, ambulance clinicians raised concerns about misdiagnosis or suboptimal clinical assessment (especially where a paramedic and neurologist may disagree):

Sometimes it can be very hard to handover the patient via FaceTime, and I feel on only a couple of occasions doctors have disregarded a patient I have felt has clearly been a stroke positive.

Ambulance clinician, East Kent

Missed diagnosis of stroke. Dismissive attitude of consultants towards paramedics. Dangerous practice of some doctors saying it is a TIA with ongoing stroke symptoms.

Ambulance clinician, East Kent

Third, a small number of comments were made about stroke clinicians having a ‘dismissive’ or ‘unfriendly’ response on calls with ambulance clinicians:

Some doctors do not trust clinicians or are quick to dismiss symptoms such as grip strength imbalance.

Ambulance clinician, East Kent

Very occasional but I have had occasions where I thought the team were dismissive of the possibility of a stroke and told us to convey to the nearest [hospital].

Ambulance clinician, NC London

Respondents also provided some reflections on how ambulance clinicians could revert to standard approaches if prehospital video triage did not work, and that positive interactions with stroke clinicians also took place:

This system only adds to our options. If for some reason the call is not successful, we can still continue as we did prior to the systems [sic] introduction.

Ambulance clinician, NC London

The HASU team have been able to speak and see the patient and any concerns are discussed in handover.

Ambulance clinician, NC London

Sufficiency of training

As noted in Chapters 3 and 4, training is a potentially important component of implementing and disseminating a new way of working. However, the pattern of responses suggests significantly different perceptions of training in NC London and East Kent [χ²(3, N = 233) = 66.54; p < 0.01]: Figure 10 shows that 91% of NC London respondents reported that they felt they had received enough training, whereas in East Kent only 42% reported that they had received enough training and 47% of ambulance clinicians in East Kent felt that they had not received enough training.

FIGURE 10.

Proportion of respondents who felt they had received sufficient training. Baseline, N = 233 (LAS, n = 159; SECAmb, n = 74).

Free-text responses provided by ambulance clinicians who indicated that they had not received training suggested that, although some staff found the system intuitive to use (thus perhaps not feeling a need for training), not all did. Respondents identified several topics for training, which included finding and accessing FaceTime and what to do when calls are not answered. They also wanted greater clarity on the handover process and decision-making, including stroke clinician expectations and decisions to leave a patient at home or convey to a local ED (as opposed to a HASU) when a patient appears to be FAST-positive. Respondents also noted that stroke clinicians might benefit from training to increase their awareness of ambulance clinicians’ knowledge and skills.

As noted above, most concerns about training were raised by ambulance clinicians in East Kent. Many of the topics identified above are in fact covered in the flow charts on SECAmb’s NHS Service Finder portal, which is provided through NHS Digital and offers ambulance clinicians access to key decision-making and service provision information. That many clinicians remained unaware of this information suggests that more active approaches to dissemination and training on prehospital video triage would be valued by clinicians. For example, further free-text responses indicated clinician preferences for in-person or online delivery of training, whereas others indicated that a team leader should deliver the training.

Future use and development of prehospital video triage

The majority of all respondents (86%) agreed or strongly agreed with the statement that overall, the new telemedicine triage system was an improvement on the previous triage process for suspected stroke patients. However, the pattern of responses in NC London and East Kent differed significantly [χ²(3, N = 233) = 40.49; p < 0.01], with 94% of NC London respondents stating the new system was an improvement but only 68% of East Kent respondents doing so (see Appendix 1, Table 12).

Similarly, the majority of respondents thought that prehospital video triage should continue. However, the pattern of responses in NC London and East Kent differed significantly [χ²(3, N = 233) = 32.86; p < 0.01]: Figure 11 suggests that support for continuation was felt more strongly by ambulance clinicians in NC London (96%) than in East Kent (70%), with 13.5% of East Kent respondents indicating that the service would require improvements if it were to continue.

FIGURE 11.

Proportion of respondents who felt prehospital video triage should continue. Baseline, N = 233 (LAS, n = 159; SECAmb, n = 74).

In relation to prehospital video triage’s potential use in other areas [e.g. myocardial infarction triage and sharing electrocardiogram (ECG) readings], there was a significantly different pattern of responses in NC London and East Kent [χ²(3, N = 233) = 41.15; p < 0.01]: while 79.9% of NC London respondents supported its use in other clinical contexts, only 39.2% of East Kent respondents were supportive (see Appendix 1, Table 12).

Below we provide a sample of free-text comments from respondents when asked if they would like prehospital video triage to continue in their region. Their tone reflects the generally positive view presented in the survey responses:

This is a very safe system, with highly experienced specialist neurologists directly impacting and improving patient care and outcomes, not only with strokes but with some more complex neurological cases.

Ambulance clinician, NC London

I think that this is a valuable tool to assist clinicians in their decision making. It is less stressful knowing that when in doubt we do have a specialised line that we can call for a second opinion.

Ambulance clinician, NC London

Used it yesterday, managed to leave a severely frail patient at an appropriate rehab facility following a likely TIA with the help of the consultant on the call. Really in-depth assessment.

Ambulance clinician, East Kent

I think it is helping our patients drastically.

Ambulance clinician, NC London

My concern, is that less subtle signs of CVA [cerebrovascular accident] are being missed e.g. for posterior CVA. And therefore, the pt [sic] is not being treated in a timely manner.

Ambulance clinician, East Kent

When it works it works well. I believe it would be better to have the stroke teams at all A&E to avoid misdiagnosis and delayed management.

Ambulance clinician, East Kent

Principal findings

Our survey of ambulance clinicians in NC London and East Kent found broad acceptance and good usability for the new triage system overall. However, a common pattern throughout survey responses was that although both areas were broadly supportive, ambulance clinicians in NC London were more positive about prehospital video triage than ambulance clinicians in East Kent.

Considering usability and acceptability of prehospital video triage, a higher proportion of ambulance clinicians in NC London found prehospital video triage to be usable than in East Kent. This was paralleled – and perhaps partly explained – by NC London clinicians being more likely to report reliable technical performance, including audio and visual signals. Underlying this may be stronger and more reliable internet/Wi-Fi connections in NC London than in East Kent.

Ambulance clinicians reported differing impacts of prehospital video triage in NC London and East Kent. Respondents in both areas reported mainly that prehospital video triage influenced conveyance decisions infrequently. This potentially reflects our qualitative analysis (see Chapter 4) which suggested that triage often confirms ambulance clinicians’ initial stroke-related assessments and provides clinicians with a degree of reassurance and confidence about their decisions. In contrast, ambulance clinicians in NC London and East Kent diverged in their views of prehospital video triage’s impact on patient safety, with NC London clinicians clearly more confident about its safety than those in East Kent. Particular concerns raised included delays owing to the time taken to conduct the remote triage assessment and the potential for misdiagnosis and thus inappropriate conveyance.

In terms of implementation and dissemination of prehospital video triage, we found that ambulance clinicians in NC London were clearly more satisfied that they had received sufficient training. This may reflect our qualitative findings (Chapter 4), which indicated that NC London took a more active approach to training, employing in-person and online approaches, whereas East Kent’s approach was more passive, sharing information through e-mail and established online resources.

Turning to future use and development of the services, most ambulance clinicians in both areas felt prehospital video triage was an improvement on what had gone before it, and were supportive of its continued use in future. However, perhaps reflecting the previously mentioned points on usability and safety, these views were more pronounced among ambulance clinicians in NC London than in East Kent.

Strengths and weaknesses

This survey was a valuable opportunity to collect individual perspectives from ambulance clinicians across NC London and East Kent at scale, using a survey tool codesigned with stroke and ambulance clinicians. In addition, this analysis complemented our qualitative analysis (see Chapter 4) by providing further ambulance clinician perspectives on such central issues as usability, acceptability and training.

There were several limitations. First, our analysis was based on a convenience sample (although we asked respondents to confirm that they had used prehospital video triage). Second, to promote ease of use and rapid distribution, the survey was not distributed with a unique invitation link or password; therefore, it is possible that people submitted more than one completed response. This means that there are limits to the extent to which findings can be generalised to other ambulance clinicians and services piloting such systems nationally. There were also variable response rates across the two regions, resulting in a higher level of representation from LAS than SECAmb.

Comparison with other studies

To our knowledge (see Chapter 3), research on prehospital video triage for stroke (and associated interventions) has predominantly focused on single approaches to delivering such services; furthermore, the bulk of past research has relied on simulation methods. By surveying a large number of ambulance clinicians in two areas that have implemented prehospital video triage in different ways, we provide a substantial data set reflecting clinicians’ experiences of ‘real-world’ implementation and delivery of these services.

Given that research suggests that diagnosing stroke is important but potentially challenging in the prehospital environment,15,16,100 this may explain why many staff responded so positively overall to the introduction of the new triage system. In connection with this, our survey demonstrated that ambulance clinicians overall value the contribution of prehospital video triage in the context of the COVID-19 pandemic. 93

Our survey confirmed findings in the literature on telemedicine in ambulances concerning technical issues with bandwidth, audio and video quality. 13,16,48,93 Furthermore, it confirmed recent simulation research suggesting background noise in stroke clinicians’ environment might be influential. 66

As noted in Chapter 3, very little research has been conducted on the safety of prehospital video triage. 16 Our findings suggest that the perceived safety of similar prehospital video triage services may vary as a result of other factors that have been analysed heavily (for example audio-visual quality).

Previous research on prehospital video triage (and associated interventions) has noted that training may be required to support delivery of such services. 59,62,65,79 Our findings on ambulance clinician perceptions of the training they received suggested that more active approaches to training may make a difference, with more active approaches (as used in NC London) preferred by clinicians.

Implications

Ambulance clinicians overall found prehospital video triage usable and acceptable; furthermore, they considered it an improvement on previous arrangements for prehospital assessment of stroke and wanted it to continue to be used in future. However, NC London clinicians reported these positive views more strongly than East Kent ambulance clinicians. A number of contrasts between the areas might help explain these differences in views. In particular, ambulance clinicians in NC London were more likely to report stable audio-visual signals and high responsiveness of stroke clinicians; furthermore, they were much more likely to report feeling that they had received sufficient training. Finally, NC London ambulance clinicians were less likely to report feeling concerned about the safety of prehospital video triage. Throughout, survey responses identified issues that planners and service leaders might think about if they plan to implement prehospital video triage sustainably in these and other contexts. Key considerations include stable internet/Wi-Fi coverage to support a suitable audio-visual signal, high responsiveness from stroke clinicians in addressing triage calls and sufficient training to ensure that ambulance clinicians are confident in using the services.

What was already known?

• There is a small but growing evidence base on the impact of prehospital video triage for stroke on delivery of care.

• Some evidence suggests that prehospital video triage may improve access to key interventions (e.g. a timely brain scan and thrombolysis).

• There is little evidence on the safety of prehospital video triage for suspected stroke.

What this chapter adds

• We analysed locally collected ambulance conveyance data for ambulance journeys over the period April–September 2020 (n = 1400). Our analysis demonstrated that, of the patients who underwent prehospital video triage, 39.1% in NC London and 51% in East Kent were transferred to the local HASU. Over 95% of stroke patients conveyed to a HASU arrived within recommended time thresholds in both NC London and East Kent.

• National stroke data reporting time from symptom onset to arrival at hospital and stroke services suggested that, despite additional time spent on scene, prehospital video triage was delivered while supporting timely patient conveyance.

• We analysed data for stroke patients submitted to the national stroke audit for the July–December periods of 2018, 2019 and 2020 (n = 137,650). Our DiD analysis suggested that, in both areas, delivery of important stroke clinical interventions (time to brain scan, stroke unit, thrombolysis and swallow screen) either improved significantly or did not change following introduction of prehospital video triage (relative to changes in RoE). These effects may be explained in part by other stroke service changes that took place concurrently in the studied areas.

• Our findings suggest that prehospital and acute stroke care in NC London and East Kent, supported by prehospital video triage, either improved or was sustained during a time of extreme system stress.

Setting

As outlined in Chapter 1, during the early stages of the COVID-19 pandemic, alongside the introduction of prehospital video triage there were other changes to how services were organised in NC London and East Kent. Notably, the location(s) of stroke services operating as HASUs changed: in NC London, the HASU relocated from the UCLH main hospital site to Queen Square. In East Kent, stroke services at William Harvey Hospital and QEQM hospital stopped admitting stroke patients routinely and the new Kent and Canterbury Hospital (K&CH) HASU opened.

Data

We analysed two data sources, summarised in Table 5:

1. Ambulance conveyance data: patient-level data routinely collected by ambulance services for suspected stroke patients in NC London and East Kent, for the period April–September 2020.

2. Stroke national audit data: publicly available, team-level clinical audit data from SSNAP for the period July 2018–December 2020.

The ambulance journey data recorded patient destination and conveyance time. These patient-level data were provided by our collaborating ambulance services, and covered the period April–September 2020, that is, the period immediately following introduction of prehospital video triage. The absence of data covering pre-implementation in NC London and East Kent and conveyance in areas that had not introduced prehospital video triage limited the extent to which we could analyse the impact of prehospital video triage on ambulance journeys.

The SSNAP data were analysed separately from the ambulance data and used to analyse delivery of key clinical interventions for stroke patients before and after the introduction of prehospital video triage. These data were available in the form of quarterly reports. At the time of analysis, two ‘quarters’ covering the period following implementation of prehospital video triage were available: July–September 2020 and October–December 2020. As a pre-implementation comparator, we analysed the July–September and October–December quarters for 2018 and 2019 (these quarters were selected to minimise the influence of seasonal variations). We used RoE as a regional comparator.

Analysis: patient destination

To understand effects on patient destination, we analysed the proportion of patients undergoing prehospital video triage who were then conveyed to HASUs (which admit stroke patients routinely) and other services in NC London and East Kent.

First, we analysed local ambulance data descriptively, calculating the proportion of patients conveyed to HASUs and other services in NC London and East Kent. We also analysed SSNAP data on the proportion of stroke patients being treated by HASUs and non-routinely-admitting teams (which generally provide ongoing hospital-based stroke rehabilitation stroke care for patients, following transfer from the local HASU, but may also provide care to late-presenting stroke patients who are no longer eligible for HASU care).

In addition, we analysed SSNAP data using a between-region DiD regression analysis to analyse whether the proportions of patients admitted to HASU and non-routinely-admitting teams changed over time in NC London and East Kent relative to the changes over time in RoE (see explanation in Chapter 2, Difference-in-differences regression).

Analysis: timely transfer of stroke patients to stroke services

To understand whether stroke patients were transferred to HASUs in a timely fashion, we analysed local ambulance data to assess whether patients who were conveyed to HASUs arrived within recommended time thresholds. The National Stroke Configuration Guidance recommends that journey times in urban areas should be ideally 30 minutes but no more than 60 minutes. 30 In the absence of regional or historic comparators, we present only descriptive results for this period of time, comparing journey times against these thresholds.

In addition, we examined SSNAP data reporting median time from symptom onset to arrival at hospital and arrival at a HASU, in NC London, East Kent, and RoE, before and after the introduction of prehospital video triage. Because these public data only presented median times, we were unable to combine quarterly data or conduct any formal data analysis.

Analysis: timely delivery of clinical interventions

National guidance for stroke recommends that stroke patients should undergo a range of interventions upon arrival at hospital, including brain imaging, swallow screening, and assessment by stroke specialists. 106 We analysed SSNAP data to examine whether or not delivery of such interventions changed in the period following the introduction of prehospital video triage services in NC London and East Kent relative to RoE.

To understand changes in delivery of clinical intervention over time, we used a between-region DiD regression analysis, comparing the changes over time in NC London and East Kent with the change over time in RoE (see explanation in Chapter 2, Difference-in-differences regression). Patient characteristics were included separately in the estimations (as the small number of hospitals did not permit inclusion of more predictors than cases): these are reported as a robustness check in Appendix 1, Table 15. A statistically significant DiD estimator in this case would confirm the difference in effect over time between the compared regions. Delivery of interventions analysed was timed relative to ‘clock start’, which refers to the point when patients arrived at hospital services, and were as follows:107

• proportion of stroke patients directly admitted to a stroke unit within 4 hours of clock start

• proportion of stroke patients scanned within 1 hour of clock start

• proportion of stroke patients given thrombolysis

• proportion of stroke patients assessed by a stroke specialist.

Destinations of patients who underwent prehospital video triage

This section presents our analysis of ambulance patient destination data for NC London and East Kent. We found that, following introduction of prehospital video triage, a substantial proportion of suspected stroke patients who may traditionally have been conveyed to a HASU were instead (following specialist assessment) referred to alternative acute services (e.g. a local ED) or recommended to stay home.

Table 6 presents local ambulance data for the period April–September 2020, covering 929 patients in NC London and 471 patients in East Kent who underwent prehospital video triage. From these patients, 181 (19.5%) in NC London and 61 (13.0%) in East Kent were not conveyed. It is important to note that the data included all calls for patients who underwent prehospital video triage – that is, including patients who were initially suspected of having stroke, but many of whom were assessed and found likely not to be having stroke.

Our analysis shows that in NC London during the observed period the National Queen Square HASU was a major destination, with 338 patients (36.4% of the total) conveyed there. A small proportion of patients was conveyed to neighbouring HASUs (with Northwick Park HASU and Royal London HASU each admitting 1.2%). The median time from the 999/111 call to the ambulance reaching the hospital for all patients was 84.6 minutes [interquartile range (IQR) 34.8] and the median time from scene to patient arriving at the hospital was 20.2 minutes (IQR 17.6).

In East Kent, most patients were conveyed to K&CH HASU (50.7%), while a smaller proportion was conveyed to QEQM (85 patients, 18%) and William Harvey Hospital (82 patients, 17.4%). The median time from the 999/111 call until the ambulance reached hospital for all cases was 63.3 minutes (IQR 34.7) and the median time from leaving scene to patient arriving at the hospital was 24.3 minutes (IQR 13.3).

Overall, our data show that a large proportion of the triaged patients in both NC London and East Kent was sent to non-HASU services. As noted in Chapter 4, there were very few cases where patients who were transferred to non-HASU services turned out to have had a stroke.

Proportion of stroke patients directly admitted to a hyper-acute stroke unit

An aim of prehospital video triage was to ensure appropriate patient destination. This section presents our analysis of SSNAP data (covering 137,650 stroke patients) to examine the proportion of stroke patients directly admitted to HASUs and non-HASUs before and after the introduction of prehospital video triage.

Table 7 suggests that in NC London the number of stroke patients admitted to a HASU was smaller in 2020 than in 2019, but comparable to 2018. For comparative purposes, the numbers of routinely admitted patients at a national level did not vary substantially during the same periods (see Table 7). The proportion of admissions to non-HASUs in NC London was higher in 2020 (10.6%) than in 2019 (7%), but again similar to 2018 (9.2%); in 2020, non-HASU admissions were highly localised around one SU (Barnet).

In East Kent, the majority of patients were conveyed to the new K&CH HASU (65%). Despite stopping as routinely-admitting units, a proportion of stroke patients were still conveyed directly to QEQM (14.5%) and William Harvey (20%) following this change (Table 8). The changes in stroke patients directly admitted to a HASU before and after the pandemic were less clear in East Kent and this was possibly confounded by the fact that K&CH started as a HASU and replaced QEQM and William Harvey at the start of the pandemic.

We used DiD regression analysis to compare trends in the numbers of patients admitted to HASUs and non-HASUs in NC London and East Kent relative to RoE before and after the introduction of prehospital video triage. Results showed that the number of patients admitted to non-routinely admitting teams decreased (p < 0.05) in these regions relative to RoE for the period July–December 2020 (see Appendix 1, Table 14).

Stroke patient conveyance times from scene of stroke to hospital

The majority of patients conveyed to a HASU had journeys that fell within nationally recommended time thresholds (see Table 7). In NC London, the median time from leaving scene to arrival at hospital was 29.2 minutes (IQR 17.6); 47.1% of the patients fell beyond the 30-minutes threshold and 5.0% of them fell beyond the 60-minutes threshold (see Table 7). In East Kent, the median time from leaving scene to arrival at hospital was 24.3 minutes (IQR 10.8); 25.9% of the patients fell beyond the 30-minutes threshold and only 3.3% of them fell beyond the 45-minutes threshold, with no patients arriving beyond the 60-minutes threshold. Locations of patients whose conveyance times exceeded thresholds are reported in Appendix 1, Table 13.

Time from stroke symptom onset to arrival at hospital and arrival at stroke unit

To examine whether prehospital video triage contributed to increased patient transfer times, we looked at SSNAP data on median time from stroke symptom onset to stroke patients’ arrival at hospital (Figure 12) and median time from stroke symptom onset to stroke patients’ arrival at a routinely admitting stroke unit or HASU (Figure 13). These data suggest that prehospital video triage in NC London and East Kent did not result in an increase in overall time from symptom onset to arrival at hospital or a HASU. In NC London, median times from symptom onset to arrival at hospital appeared slightly lower in July–December 2020 than they were in July–December 2018 and 2019; median times from symptom onset to arrival at a HASU were clearly lower in July–December 2020 than they were in July–December of 2018 and 2019. In East Kent, median times from symptom onset to arrival at hospital and arrival at a HASU in July–December 2020 appeared broadly consistent with median times in July–December of 2018 and 2019. Times for RoE for both measures appeared consistent over 2018, 2019 and 2020.

FIGURE 12.

Median time from stroke symptom onset to arrival at hospital in NC London, East Kent and RoE, July–December, 2018, 2019 and 2020.

FIGURE 13.

Median time from stroke symptom onset to arrival at stroke unit in NC London, East Kent and RoE, July–December, 2018, 2019 and 2020.

The numbers and characteristics of stroke patients treated by hyper acute stroke units

Table 9 presents our analysis of SSNAP data on the numbers and characteristics of patients treated by HASUs in NC London, East Kent and RoE for the period July–December from 2018 to 2020. These data help explain whether or not the characteristics of patients treated changed during the pandemic (e.g. whether or not only more severe strokes were being conveyed to a HASU). In addition, we used these characteristics as risk adjustments to assess whether or not these characteristics made any difference to the likelihood of receiving clinical intervention.

Table 9 shows that the median age of stroke patients treated by HASU teams was higher for East Kent than NC London. This may be because East Kent’s population has a higher population age than NC London overall. 108 The demographic structure of the population may also explain why the proportion of female stroke patients in East Kent is higher (as, on average, female individuals live longer) than NC London. There were no clear trends in stroke type over time for NC London and East Kent, although both regions generally admitted a lower proportion of ischaemic stroke patients with infarction than RoE.

The NIHSS quantifies impairment caused by stroke. NIHSS status varied on average between four and six (the only exception being QEQM hospital in July–September 2018, which had an average of 7.5). The proportion of HASU-treated stroke patients with two or more comorbidities was higher for NC London than for East Kent and more comparable with RoE for all three compared years. The proportion of patients who were conveyed to a HASU by ambulance was generally higher for NC London than for East Kent.

Delivery of clinical interventions before and after the introduction of prehospital video triage

Using SSNAP, we analysed whether there were any changes in delivery of key stroke clinical interventions in NC London and East Kent HASUs relative to RoE before (July–December 2018 and 2019) and after (July–December 2020) the introduction of prehospital video triage. The changes before and after DiD estimators are presented in Table 10. As a robustness check we also included patient characteristics (including stroke type, sex, age, NIHSS and number of covariates) as covariates in our analyses; we report these separately in Appendix 1, Table 15.

The proportion of stroke patients admitted to a HASU within 4 hours (see Table 10) increased significantly relative to RoE in East Kent following introduction of prehospital video triage (40.95%, p < 0.05).

The proportion of patients receiving a brain scan within 1 hour (see Table 10) did not change significantly in NC London relative to RoE following introduction of prehospital video triage, but increased significantly in East Kent (23.15%, p < 0.05).

In terms of thrombolysis, we found no significant changes following implementation of prehospital video triage in NC London and East Kent relative to RoE for the proportion of stroke patients thrombolysed, stroke patients eligible for thrombolysis or patients thrombolysed within 1 hour (see Table 10). However, we found a significant reduction in the proportion of patients eligible for thrombolysis in East Kent following introduction of prehospital video triage (–7.25%, p < 0.05).

The proportion of patients assessed by a stroke physician within 24 hours of clock start (see Table 10) increased significantly relative to RoE in both NC London (7.4%, p < 0.05) and East Kent (32.4%, p < 0.001) following introduction of the prehospital video triage.

The proportion of patients assessed by a stroke nurse within 24 hours of clock start (see Table 10) increased significantly relative to RoE following introduction of prehospital video triage in both NC London (3.7%, p < 0.001) and East Kent (16.25%, p < 0.001).

The proportion of eligible patients who underwent a swallow screen within 4 hours (see Table 10) increased significantly relative to the rest of England in both NC London (20.30%, p < 0.05) and East Kent (31.55%, p < 0.001).

Principal findings

In this chapter, we aimed to analyse whether prehospital video triage supported appropriate and safe conveyance and treatment of suspected stroke patients. Our analysis of ambulance and SSNAP data on patient destinations found that (1) relatively few patients were being conveyed to HASUs outside the studied areas of NC London and East Kent; and (2) the number of stroke patients being treated in a HASU in NC London reduced somewhat in the early stages of the pandemic (results were unclear for East Kent because of the restructuring of the HASU services in this region). Although this suggests that the overwhelming majority of stroke patients were being conveyed to a HASU rather than their local ED, it reflects an observational analysis indicating a reduction in the number of milder stroke admissions from March through May 2020. 109 Likely contributing factors may include general reluctance of patients to attend hospital and clinical concerns about the safety of conveying patients between hospitals during the early stages of the COVID-19 pandemic.

The proportion of patients who arrived at a HASU by ambulance was higher for NC London than East Kent. In addition, a higher proportion of stroke patients was admitted to non-routinely admitting stroke units during the implementation period (i.e. 2020 data) in NC London, in East Kent, and at national level. Possible reasons for non-conveyance of strokes to the HASU need to be better explored but there may be several factors associated with this. First, during the time of analysis there were changes to stroke numbers reported nationally and locally because of the COVID-19 pandemic. These changes were mainly driven by the conservative management in local hospitals and nursing homes, patients presenting later or not at all, and the impact that the lockdowns and the pandemic had on the mobility of people in large metropolitan areas like London. 110 Second, some patients may have been conveyed to hospital by ambulance clinicians who were unaware of prehospital video triage service (e.g. private ambulance services). Third, some patients may have preferred to go to the nearest point of care (for example their local EDs). Finally, there may have been fewer interhospital ED to HASU transfers, primarily owing to factors associated with the COVID-19 pandemic, including clinician concerns about COVID-19 infection, ambulance capacity and availability of HASU beds.

Our analysis of ambulance conveyance times from scene of stroke to hospital found that the overwhelming majority of stroke patients reached stroke services within the recommended time thresholds. In addition, publicly reported median times from stroke symptom onset to stroke patients’ arrival at hospital and a HASU suggested that prehospital video triage did not result in increases in overall stroke patient conveyance times. These data suggest that prehospital video triage can be delivered while still supporting timely patient conveyance to stroke services. However, these patterns may in part have been facilitated by the relocation of HASUs in NC London and East Kent and reduced traffic during the pandemic.

Our analysis of clinical interventions over 2018–2020 found no evidence of significant reductions in delivery of key stroke clinical interventions (e.g. brain scan, admission to HASU and specialist assessments) in NC London and East Kent (relative to RoE). Rather, in several cases we found that the proportion of patients receiving timely interventions increased significantly (particularly in East Kent). This was despite the context of COVID-19, which placed unprecedented pressure on emergency services. However, it is likely that improvements may also be attributed to wider service changes conducted by the stroke teams in NC London and East Kent, for example, the introduction of new ‘front door’ processes or factors related to the COVID-19 context (e.g. there was less pressure on the system overall, leaving greater capacity in HASUs to attend to stroke patients).

Strengths and weaknesses

The main strength of the analyses was the unique regional data set we used, which contained detailed information on ambulance destination and conveyance times. We were also able to analyse a substantial (aggregate) data set covering delivery of stroke care for over 80,000 stroke patients over a 30-month period. The robust quasi-experimental framework we used is another strength of the study, permitting useful and informative comparison over time with RoE and ruling out other factors that could affect outcomes across the country during the same period. Having control over the trends over two different quarters also allowed us to account for seasonality in the trends.

Our study had several limitations that caution against interpreting our findings as purely causal effects. First, the local routinely collected ambulance conveyance data for stroke patients were only available for the period April–September 2020. This covered only the period after implementation of prehospital video triage, which continued to develop during this period. Furthermore, these data covered only the areas in which prehospital video triage had been introduced. Therefore, our analyses of these data lacked both historic and geographic controls, and we could not account for the impact of COVID-19 on patient behaviour and on the system (i.e. lower overall activity during this period, etc.). Second, we were unable to assess the impact of the new system on patient outcomes, such as long-term quality of life, disability or functional independence, as these measures are not collected with sufficient reliability in SSNAP clinical audit 6-month follow-up. Third, SSNAP data include only patients admitted to hospital; we were therefore unable to analyse impact on patients pronounced dead at home or while being conveyed to hospital. Fourth, the SSNAP data we analysed were aggregated at the team level: analyses of patient-level data would have permitted assessment of the relationship between patient characteristics and care delivery, especially in terms of analysing time to intervention. Fifth, in the SSNAP data we were not able to distinguish between the video and non-video assisted cases, meaning we could not assess any differential effect on delivery of HASU care. Sixth, we could not account for a series of factors that could influence the indicators analysed here, such as the number of ambulances, organisation of primary care, accessibility of nearer general hospital EDs and potentially public awareness of the FAST test in NC London or East Kent. Seventh, several system changes, including major system optimisation of stroke services, occurred during the analysed period in both NC London and East Kent and this made it difficult to attribute changes in, for example, care delivery to prehospital video triage.

Comparison with other studies

The evidence on prehospital video triage is small but growing, although with several important gaps. We believe that our analysis builds on or confirms some of the existing evidence base, but also begins to fill some of these gaps. As outlined in Chapter 3, research on prehospital video triage tends not to address the impact of such services on patient destination, in particular the potential for the avoidance of unnecessary conveyance to a HASU. In our analysis, prehospital video triage resulted in a substantial proportion of suspected stroke patients being conveyed to alternative destinations, such as the local ED, or remaining home. Taken alongside our qualitative findings (e.g. governance meetings investigating inappropriate patient conveyance), this suggests that specialist stroke assessment may have resulted in a substantial reduction in unnecessary patient conveyance to a HASU. Previous reviews in this area (see Chapter 3) indicate that the potential impact of prehospital video triage on patient safety is largely unstudied. 16 Our analysis of ambulance conveyance showed overwhelmingly that prehospital video triage did not result in conveyance times that went beyond recommended time thresholds. Our findings on improvements to delivery of clinical interventions reflect recent studies suggesting the potential of such interventions to support significant improvement of acute care for stroke patients, including administration of thrombolytic therapies. 41,111,112

Implications

Our analysis showed that a substantial proportion of suspected stroke patients who would traditionally be conveyed to a HASU were, following prehospital video triage assessment, conveyed to alternative destinations, thus avoiding unnecessary conveyance to a HASU. There was a slight reduction in the proportion of stroke patients being treated in a HASU during the early stages of the pandemic, reflecting wider evidence of obstacles to HASU conveyance nationally. Our analysis of the times from symptom onset to arrival at hospital and the HASU suggested that prehospital video triage can be conducted while also permitting timely conveyance of stroke patients to a HASU. We found several significant increases in the proportion of patients receiving timely stroke clinical interventions, and no examples of significant decreases. However, it is likely that these changes may be attributed at least in part to other service changes and COVID-19, in addition to prehospital video triage.

What evidence exists on prehospital triage services for suspected stroke patients, in terms of implementation, usability, safety and outcomes?

• There is a limited, but growing evidence base on remote prehospital video triage for stroke, which sits alongside more established and substantial literatures covering other aspects of telemedicine for stroke (e.g. systems permitting hospital-to-hospital communication and mobile stroke ambulance units). Much of this evidence was based on pilot or feasibility research, using both simulated and ‘real-world’ settings.

• In terms of usability, research suggested that stable network coverage and clear audio-visual signals were important to successful patient assessment. Communication between ambulance and stroke clinicians was also important in ensuring that stroke clinicians could access appropriate patient information.

• Training of both ambulance and stroke clinicians was identified as an important facilitator of effective prehospital video triage, for example using simulations to enable a clear understanding of new protocols and effective use of communications technology.

• Importantly, both reviews and primary studies on outcomes of prehospital video triage tended to focus on delivery of stroke clinical interventions, for example time from arrival at hospital to brain scan or thrombolysis. They presented little evidence on the impact of prehospital video triage on such key factors as appropriate patient destination, patient safety, and cost-effectiveness.

Were the prehospital video triage services piloted in North Central London and East Kent acceptable to their users (stroke clinicians and ambulance clinicians)?

• Our qualitative analysis suggested that in NC London and East Kent, some ambulance clinicians reported ongoing concerns about whether the potential benefits of accessing specialist secondary care stroke expertise might be outweighed by a delay in patient conveyance. However, most ambulance and stroke clinicians interviewed were supportive and accepting of the prehospital video triage services; for example, both groups cited perceived improvements in appropriate patient conveyance and potential reductions in pressure on services (e.g. unnecessary ambulance journeys and fewer ‘stroke mimics’ to manage in HASUs).

• Ambulance clinicians cited increased confidence and reassurance about their decisions on patient conveyance, and the view that they were learning more about stroke through their communications with stroke clinicians.

• Stroke clinicians noted that the service did not involve a significant change in practice, beyond having the opportunity to conduct assessments earlier in the pathway and gain advance knowledge of patients. However, they also described a barrier to acceptability in the common requirement to conduct triage assessments alongside their other professional duties, placing pressure on clinicians and potentially limiting the quality of communication.

• Our survey of over 200 ambulance clinicians in NC London and East Kent confirmed that a substantial majority of respondents (86%) found the prehospital video triage services an improvement on ‘business as usual’ and a similarly high proportion (88%) wanted the new services to continue to operate. However, these positive views were significantly stronger among NC London ambulance clinicians.

Were the services effective in terms of usability and image/sound quality?

• In our interviews with ambulance and stroke clinicians, image and sound quality were seen as sufficient to conduct prehospital video triage assessments. Ambulance clinicians noted that connections could be disrupted, as both built-up areas and highly open spaces had potential for limited Wi-Fi coverage. Under such circumstances, ambulance clinicians reverted to conveyance protocols that operated before the introduction of prehospital video triage. Issues with connectivity and ambient noise in hospital settings were addressed through connection to hospital Wi-Fi and by giving stroke clinicians audio headsets. Our survey confirmed that ambulance clinicians (77%) agreed that the prehospital video triage services were usable in terms of audio-visual quality.

• Ambulance and stroke clinicians reported that using the FaceTime communication platform was straightforward to use, although some ambulance clinicians suggested that training and ‘refresher’ courses could be beneficial. NC London’s approach to training was more active, with both face-to-face training and distribution of video information, whereas in East Kent protocols were distributed via e-mail and an online portal. In line with this, our survey found that a majority of respondents (89%) saw the services as easy to use. However, a higher proportion of NC London respondents (94%) rated the service as usable than in East Kent (78%): this may reflect the more active approach to training taken in NC London, where 91% of ambulance clinicians reported having received sufficient training, in contrast to East Kent, where 42% of ambulance clinicians reported having received sufficient training.

Did the services support the appropriate, safe conveyance and treatment of suspected stroke patients?

• Our analysis of ambulance journey data showed that only a small percentage of stroke patients did not reach hospital within 60 minutes of leaving the scene of stroke. Publicly reported national stroke audit data on times from symptom onset to stroke patients’ arrival at hospital and HASUs found that prehospital video triage did not result in increased median stroke patient conveyance times. This suggests that, despite the additional time spent on prehospital video triage consultations, prehospital video triage can be delivered while still supporting timely patient conveyance to stroke services.

• Our analysis of national stroke audit data found several significant increases and no significant reductions in delivery of stroke clinical interventions in NC London and East Kent following introduction of prehospital video triage (i.e. relative to changes observed in RoE over the same time period). However, it is possible that other factors, such as relocation of the hospitals’ hyper-acute stroke services in both areas, were significant contributors to these improvements.

• Our qualitative research suggested that leaders of the triage services were conscious of risks to patient safety and put processes in place to monitor and manage any incidents. Observations of meetings where these data were discussed suggested that there were few reported safety incidents, each of which was explored to identify potential improvements. Our interviews suggested that there was a broader perception among ambulance and stroke clinicians that these services were providing safe care.

• Our survey of ambulance clinicians confirmed that prehospital video triage influenced patient destination at least once for most (86%) respondents, while 82% of respondents had no concerns about the safety of the services. However, there was again a significant difference between responses in NC London, where 91% had no concerns about safety, and East Kent, where 62% had no concerns.

Which factors influenced uptake and impact of these services?

• Our qualitative research suggested that several factors helped enable the rapid development, implementation and uptake of prehospital video triage. These factors related to (1) the local and wider contexts into which prehospital video triage was introduced; (2) the interventions themselves; and (3) how they were put into action.

• Two national/international contextual factors helped drive these changes. First, there existed a longstanding challenge related to the limited specificity of screening instruments such as FAST, which led to the conveyance of a substantial proportion of non-stroke patients to stroke units. Second, the COVID-19 pandemic, by adding a significant patient safety risk to the issue of inappropriate patient conveyance, acted as a ‘burning platform’ for change.

• These drivers interacted with more local contextual factors. For example, they encouraged adoption of governance processes that were more facilitative of change at system and service levels. The pandemic also raised issues that played on values important to the ambulance and stroke clinicians who were to deliver prehospital video triage, specifically the desire to provide high-quality, safe care to stroke and non-stroke patients.

• In terms of the interventions themselves, the prehospital video triage services were attractive to many ambulance and stroke clinicians. They saw the process and interface as straightforward to use. Furthermore, reflecting their professional values (cited above), they indicated that it offered advantages over ‘business as usual’, in terms of getting the patient to the most appropriate service for the best possible care.

• Collaborative leadership was key to implementation: ambulance and stroke clinical leads worked in multiple directions to develop the new services. Within their organisations, clinical leads engaged with senior management to gain local endorsement for the services and worked actively with frontline clinicians (e.g. to encourage uptake, provide training and monitor progress of services). They also reached beyond their local organisations; for example, ambulance and stroke clinical leads worked together across organisational boundaries to develop services that would work for ambulance and stroke clinicians alike. They also engaged with wider system governance to obtain support for these changes and aid ongoing preparations for wider roll-out of the services, if successful.

Which aspects of these services should be retained post COVID-19 and which adaptations (if any) are required to support their implementation?

• The ambulance and stroke clinicians we interviewed were emphatic that the new prehospital systems represented should be retained. Many suggested that they should be implemented more widely, both in other parts of the country and other health-care specialties. This support for continuation was reflected in our ambulance clinician survey, where 88% of respondents indicated that they would like to see the service continue, and 67% suggested that similar services should be considered for other clinical settings.

• Broadly, interviewees suggested that relatively few changes were required to the services. A potential risk to sustainability of these services likely to require adaptation was ensuring sufficient capacity among stroke clinicians to deliver the assessments. The current arrangement, where stroke clinicians conduct assessments alongside their other duties, was felt to be disruptive and placed undue pressure on clinicians (which may have important implications for sustainability of prehospital video triage). Our findings on training suggest that more active approaches are preferred by staff: such approaches, if extended, may have potential to encourage increased collaboration between ambulance and stroke clinicians.

Implementation

We found that prehospital video triage can be developed and implemented rapidly. By drawing on a relevant theory of implementation and sustainability of innovations, we were able to establish that influential factors included context, implementation approaches and the characteristics of the prehospital video triage services themselves. These factors were interrelated; for example, the COVID-19 pandemic acted as a ‘burning platform’, encouraging more facilitative governance processes and local professional and organisational receptivity to new ways of working. The wider service reorganisations – whereby local stroke units were no longer co-located with ED services – may have encouraged staff to engage more with prehospital video triage. Ambulance clinicians saw training as an important means of becoming confident users of the new service; more active approaches to training are more likely to be viewed positively by clinicians. Collaborative leadership was key to implementation: ambulance and stroke clinical leads engaged with local senior management, frontline clinicians and beyond their local organisations to develop services that would work for ambulance and stroke clinicians alike and to gain ongoing support across the system.

Acceptability and usability

Ambulance and stroke clinicians overall found prehospital video triage acceptable and usable. The technology was seen as straightforward to use and generally reliable. A potentially important factor was the nature of training offered in the two areas, with more active approaches (as employed in NC London) preferred by ambulance clinicians. Stroke clinicians reported concerns about delivering prehospital video triage alongside their other duties, suggesting that addressing this issue would be important to ensuring sustainability of services.

Impact on safety and quality

Almost all stroke patients’ ambulance journeys to HASUs remained within recommended conveyance time thresholds. Publicly reported national stroke audit data on times from symptom onset to stroke patients’ arrival at hospital and a HASU suggested that, despite additional time potentially spent on scene, prehospital video triage can be delivered while still supporting timely patient conveyance to hospital and stroke services. In terms of stroke care delivery, we found several significant increases in delivery of key clinical interventions following the introduction of prehospital video triage (above and beyond what was seen elsewhere in England), although other concurrent changes to service organisation may have played influential roles. Our qualitative data – in terms of both interviews and observations of meetings where safety issues were analysed – suggested that safety was a key priority of the clinicians delivering these services; governance processes gave assurance that the services were indeed delivering safe care as well as wider service and system benefits.