The impact of telemedicine enabled pre-hospital triage in acute stroke – a protocol for a mixed methods systematic review

Objective

To summarise and synthesise existing qualitative and quantitative evidence of the impact of telemedicine on key patient and service level outcomes in acute stroke, when used to facilitate patient triage in the pre-hospital setting, and to understand how human factors impact on these complex health system interventions in the acute stroke pathway.

Review questions

Quantitative: What is the impact of pre-hospital telemedicine enabled triage on functional outcomes of patients experiencing acute stroke, and on key time metrics reflective of quality in hyperacute stroke care?

Qualitative: How do the experiences and perspectives of end users of telemedicine-based interventions help explain the effectiveness, acceptability and feasibility of such interventions?

Study eligibility

Study design: This mixed method systematic review will assess telemedicine enabled communication strategies or networks that facilitate and support pre-hospital decision making on the most appropriate patient disposition or routing (primary stroke centre or comprehensive stroke centre) of patients experiencing acute stroke when compared to usual care. Usual care is defined as undifferentiated transportation of patients to their local hospital.

Primary quantitative studies will include randomised controlled trials (RCTs), non-randomised controlled trails (NRCTs), controlled before-after studies (CBA) and interrupted time series and repeated measures studies (ITS). We will also include retrospective or prospective observational cohort studies.

Qualitative studies will include any primary research study that uses recognised methodologies for data collection and analysis. Data collection techniques may include interviews, focus group discussions or surveys, while standard accepted qualitative analysis and synthesis methods such as narrative, thematic or framework approaches may be included. Feasibility studies or process evaluations that include formal qualitative methodology for data collection and analysis will also be included.

Primary mixed methods studies which permit disaggregation and analysis of quantitative and qualitative data separately will be included. If the quantitative or qualitative component of any primary mixed methods study does not meet the pre-established criteria above then only the study component (quantitative or qualitative) meeting the criteria will be included with the second study component data excluded at the full text screening step.

Quantitative inclusion criteria

Using the following population, intervention, comparison and outcome (PICO) elements for quantitative studies or quantitative components of mixed methods studies we will identify studies meeting our inclusion criteria:

Population: The target study population will be adults ≥ 18yrs with suspected acute stroke <24hrs since symptom onset, unknown time of symptom onset or wake-up stroke (WUS). (WUS is defined here as the scenario where a person wakes up with symptoms of stroke having gone to bed previously symptom free, and where their last known well time >4.5 hours prior¹²).

Intervention: The quantitative component of this review will consider studies that examine the use of telemedicine-enabled interventions for patient triage in the pre-hospital setting of acute stroke care. Such interventions should aim to facilitate rapid communication, decision making and triage of patients to the most appropriate destination stroke centre (PSC or CSC) potentially bypassing the geographically nearest hospital, in order to take a person directly to a stroke centre deemed most suitable to their clinical presentation.

Telemedicine in this context may comprise: telephone communications between the pre-hospital and hospital setting that also include a formal objective clinical assessment tool; televisual or tele-audio-visual communications including smartphone applications or other remote Wi-Fi-enabled communication links to hardware on board an ambulance to the hospital-based team; Mobile Stroke Units (MSU). (While MSUs may be deemed an interventional service for the provision of rapid thrombolysis, it may also be considered a rapid triage tool for those who could benefit from thrombolysis and/or thrombectomy).

Included studies may focus on IVT, EVT or both. Any transportation mode will be considered. Pilot studies that meet the inclusion criteria will be included.

This review is focused on pre-hospital tele-triage and so will not include studies that use telemedicine to facilitate remote hospital-based clinical assessment (hub-and-spoke models) or to facilitate inter-hospital transfer decisions between a PSC and CSC (drip-and-ship models).

It is acknowledged that some studies will include results for all patients evaluated by a triage intervention while other studies will report results only of patients receiving treatment for acute ischaemic stroke (eg. onset to needle time, door to needle time, door to puncture time), some of whom will include stroke mimics. If a study does not report its rate of treatment of mimics, authors will be contacted for more specific treatment metrics related to mimics. If additional information on the treatment rate of mimics is not available this will be highlighted and the study excluded where deemed appropriate.

Comparison: The quantitative component of the review will consider studies that compare the intervention to the immediate transportation of a patient with stroke or stroke-like symptoms to the nearest hospital without pre-hospital differentiation or triage (usual care). Historical controls will be accepted. A control group must be present for any quantitative study to be included.

Outcomes: The quantitative component of this review will consider studies that include the following outcome measures: the primary patient-centred outcome will be assessment of functional ability at 90 days using the Modified Rankin Scale (mRS). Safety will be assessed using the 90-day mortality rate.

Secondary quantitative outcomes will be considered at a health service level, where available, to include the key time metrics, proportions of people receiving IVT and additional clinical impacts. The following outcomes will be considered: symptom onset to hospital (door) time; onset to decision time thrombolysis (IVT); onset to decision time endovascular thrombectomy (EVT); onset to treatment time (“onset to needle” for (IVT) and “onset to puncture” for (EVT)); alarm to treatment times; door to first imaging time; overall rates of interventions (IVT and EVT); rates of interventions <60 minutes from symptom onset; time to first contact with CSC; NIHSS at 24 hours; rate of symptomatic intracranial haemorrhage at <1 week; proportion of stroke mimics treated with IVT; hospital length of stay; ambulance usage time. Additionally, depression, cognition and quality of life at 90 days will be assessed where recorded.

Qualitative inclusion criteria

Using the following Population, Phenomenon, Context (PICo) for qualitative studies, or the qualitative components of mixed methods studies, we will identify studies meeting the following inclusion criteria:

Population: Any adult ≥18 years involved in the provision or receipt of an intervention as described above. This may include patients, family members or healthcare staff.

Phenomenon of Interest: The qualitative component of this review will consider studies that investigate the experiences, perspectives and perceptions of those involved in providing or receiving a telemedicine-based intervention.

Context: The setting of the qualitative studies will primarily be in the pre-hospital phase of the acute stroke pathway may cross both primary and secondary care.

Exclusion criteria

We will exclude studies involving children or young people ≤17yr. Studies that involve inter-hospital transfers without pre-hospital triage or studies that used patient simulations will be excluded. Studies that use ICT for pre-notification of ambulance arrival purposes only will not be sufficient for inclusion. We will exclude systematic reviews or any trial without a control.

Search strategy: A systematic search of electronic databases MEDLINE (Ovid), EMBASE, The Cochrane Library, CINAHL and Web of Science, will be undertaken using a search strategy based on the key concepts under review developed in collaboration with a data information specialist(PM). A preliminary scoping search identified qualitative and quantitative studies of interest in the area. An expanded search using relevant terms including stroke, telemedicine, pre-hospital and triage will be incorporated into medical subject headings, and key words will be searched and combined systematically in an iterative approach to maximise information retrieval. No language or date limits will be applied. A sample search strategy for MEDLINE (Ovid) is provided in Table 1 and will be adapted for all databases as appropriate. Published studies in any language available in full text that can be reliably translated into English will be included. Reference lists of included full text articles will be further screened for relevant studies.

Study selection: Two researchers (DMcC and SL) will independently screen by title and abstract against the documented eligibility criteria using Covidence software¹³. A solution to any disagreements will initially be through conferral and discussion between the two reviewers. Unresolved disagreements will be discussed with the project supervisors (DW and AH) in order to reach consensus.

Included studies will be retrieved in full text and will again be screened independently by the two reviewers. Disagreement on inclusion of any study at this stage will be resolved by discussion, or again with the project supervisors, if necessary to gain consensus. Reasons will be recorded for studies excluded at full-text stage. A PRISMA flow diagram will record all steps for accuracy and transparency.

Data management: A record will be maintained of all final database searches. References will be managed using EndNote20.1. Screening software will permit efficient screening of titles and abstracts, and of full texts, independently by both reviewers. Review Manager (RevMan) version 5.4 will be used for meta-analysis and qualitative synthesis if results permit through its open-access link to GRADE-Pro software.

Data extraction: A data collection form will be designed to ensure that complete and consistent data extraction from quantitative, qualitative and mixed methods studies is achieved. Collected data will include year of publication, study design, study setting, population or perspective, participant numbers and demographics, intervention, control or comparator, primary outcome, secondary outcomes, measures of intervention effect and findings or themes.

Reviewers will be trained on the utilization of the data collection form and it will be piloted prior to use to ensure utility and consistent use by reviewers. Data extraction will be undertaken independently by the same reviewers involved in screening. Disagreement on inclusion of any data will be resolved by discussion as described for screening steps. Efforts will be made to find missing data including by contacting study authors.

Assessment of methodological quality: The Mixed Methods Appraisal Tool (MMAT) will used at study level to assess methodological quality of empirical quantitative (including RCTs, non-randomised studies and quantitative descriptive studies), qualitative and mixed methods studies, using five domains for each design assessed¹⁴. Judgements of quality will be by two reviewers undertaking independent appraisals. Sensitivity analyses will be undertaken where possible and appropriate, excluding any studies of low methodological quality if they exert any undue influence on estimates. Unpublished data will be included in this review if it made available by authors. Should none be made available, this limitation will be acknowledged when interpreting any funnel plot assessing potential publication bias.

The TIDIER (Template for Intervention Description and Replication) checklist will be used as a guide to reporting of details of included interventions.

Overall assessment of certainty: An assessment of the overall certainty in the accumulated evidence will be made using the GRADE and Grade-CerQual tools for quantitative and qualitative studies respectively that permits an evidence to decision framework. GRADE narrative statements will further clarify the authors’ confidence in the evidence extracted in the review.

Data synthesis and integration: The convergent segregated mixed methods approach to this systematic review involves separate but parallel synthesis of quantitative results and qualitative findings. Data will not be transformed from quantitative to qualitative or vice versa and a clear distinction will be maintained between quantitative results (including the quantitative results of MM studies) and qualitative results (including the qualitative results of MM studies) until the integration step. Synthesis of each single method type will be complete before moving to integration. Quantitative results will be synthesised using meta-analysis. Subgroup analyses will be undertaken where appropriate for example where geographic location of studies varies, or study methodology may be significant to the quality of any primary analysis. Synthesis of qualitative findings will categorise themes extracted from primary studies based on the research question and will be described in narrative and schematic form. NVIVO (QSR International) software will assist inductive thematic synthesis.

Integration will follow a parallel results-based approach and will be configured via juxtaposition of synthesized quantitative results and synthesized qualitative findings framed around the JBI methodology. This approach will permit formal consideration of how or if the findings of one single method complements the other or can otherwise explore, contextualise or explain findings in the other results set.

Statistics: Statistical analysis will be undertaken using Stata (StataCorp) software with meta-analysis of primary and secondary quantitative outcomes, and subgroup analysis by intervention type, if data permits. Heterogeneity in the study methodologies is likely and so a random effects statistical model will be used for analysis. Statistical heterogeneity across study results will be estimated using the I² statistic and Chi² test. If meta-analysis is not possible, a narrative meta-synthesis will proceed.

PPI: The members of a Public and Patient Involvement (PPI) panel with experience in stroke will be involved in an advisory role.