Abstract
Background and Objective Stroke reperfusion therapy is time critical. Improving prehospital diagnostic accuracy including the likelihood of large vessel occlusion can aid with efficient and appropriate diversion decisions to optimize onset-to-treatment time. In this study, we investigated whether prehospital telestroke improves diagnostic accuracy when compared with paramedic assessments and assessed feasibility.
Methods We conducted a pragmatic, community-based, cluster randomized controlled trial comparing the diagnostic accuracy of telestroke assessments inside the ambulance with a modified Los Angeles Motor Scale (PASTA score). The primary outcome was the accuracy of predicting reperfusion candidates; secondary outcomes were accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of either approach to identify IV thrombolysis (IVT) and endovascular thrombectomy (EVT) candidates and true stroke patients by study group. The accuracy of telestroke and PASTA assessments was compared against in-person assessment in the emergency department and with the final diagnosis/intervention for the patient. We also monitored for technical challenges.
Results We recruited 76 patients (35 telestroke and 41 PASTA) between August 2019 and September 2020. The mean age was 72.2 (±14.6) years. Telestroke was 100% (95% CI 90%–100%) and PASTA 70.7% (54.5%–83.9%) accurate in predicting reperfusion candidates compared with preimaging emergency department neurologist assessment (p < 0.001). When compared with actual reperfusion therapy administered, the predictive accuracy was 80% (63.1%–91.6%) and 60.1% (44.5%–75.8%) for telestroke and PASTA, respectively (p < 0.001). In predicting the administration of IVT, telestroke was 80% (63.1–91.6) and PASTA was 56.1% (39.8–71.5) accurate (p < 0.001). In predicting intervention with EVT, telestroke was 88.6% (73.3–96.8) and PASTA 56.1% (39.8–71.5) accurate (p = 0.005). The service model proved technically feasible and was acceptable to neurologists.
Discussion Prehospital telestroke assessment is feasible, accurate, and superior to the PASTA score in predicting acute reperfusion therapies, presenting an effective option to guide prehospital diversion decisions.
Classification of Evidence This study provides Class I evidence that intra-ambulance telestroke evaluation has a greater diagnostic accuracy compared with the PASTA score performed by paramedics in distinguishing hyperacute stroke patients who are candidates for reperfusion therapy.
Glossary Information
- EVT=
- endovascular thrombectomy;
- ED=
- emergency department;
- IVT=
- IV thrombolysis;
- LVO=
- large vessel occlusion
Stroke is a medical emergency and one of the leading causes of death and disability worldwide.1 In the acute setting, the morbidity and mortality of ischemic stroke can be reduced by reperfusion therapies with IV thrombolysis (IVT), endovascular thrombectomy (EVT), or both. These interventions are time critical, with delays in treatment as short as 15 minutes being associated with a lower probability of independent ambulation, a higher risk of mortality, and a greater risk of symptomatic intracranial hemorrhage.2,3 However, not all hospitals have the capacity to perform EVT, which often results in further time lost in transferring patients between hospitals to an EVT-capable center. Such transfers are associated with worse clinical outcomes.4 There have been a number of different approaches taken to try to minimize transfer time, while also avoiding unnecessary diversions: the use of mobile stroke units with a CT scanner and using scoring systems in the ambulance to triage patients are 2 such examples.5 However, it remains uncertain which approach is the most effective in ensuring that those patients requiring intervention have timely access to it, while still optimizing health resource utilization and minimizing unnecessary transfers or diversions that can unduly burden both patients/families and the health sector.
In the area of acute stroke care, telestroke is the use of videoconferencing to facilitate remote assessment by an expert stroke physician to provide guidance on diagnosis and management of patients. This infrastructure has been used in the Central Region of New Zealand since 20166 and has been demonstrated in several regions to increase IVT rates and reduce door-to-treatment times.7,8 In this study, we sought to use the existing telestroke expertise of neurologists working at Wellington Hospital to assess whether telestroke assessment in the ambulance, before the arrival of the patient in the hospital, was feasible and able to achieve high diagnostic accuracy when compared with assessment by paramedics using scoring tools. If superior, it could offer a more reliable service model to make appropriate diversion decisions and call off nonstroke/nonreperfusion prehospital alerts, directing patients faster to suitable services. It could also build on prior reductions in door-to-imaging times demonstrated for paramedic prenotification to hospital services.9
The primary research question addressed by this study is whether intra-ambulance telestroke evaluation has a greater diagnostic accuracy compared with the PASTA score performed by paramedics in distinguishing hyperacute stroke patients who are candidates for reperfusion therapy.
Methods
Design
This trial was conducted as a prospective, single-blind, community-based, cluster randomized controlled trial at a single center in Wellington, New Zealand, from August 2019 to September 2020. It was designed to assess the diagnostic accuracy of the tested tools without affecting clinical patient management, interventions, or diversion throughout the study period. The primary outcome was the accuracy of prehospital telestroke assessment (teleambulance) in correctly identifying stroke patients suitable for reperfusion therapies, in comparison with standard practice including the use of paramedic screening tools. The secondary outcomes included the sensitivity, specificity, and positive and negative predictive values of the teleambulance assessment in correctly identifying stroke patients suitable for IVT and those suitable for EVT. We also monitored for technical challenges and sought neurologist feedback to gauge the acceptability of the service.
Participants
Patients were those older than 18 years presenting to Wellington Hospital through the Wellington Free Ambulance service with potential stroke, between the hours of 0800-2200. Patients were those identified by paramedics using the FAST assessment (Face, Arm, Speech, and Time; Figure 1) along with other criteria (e.g., no seizure and normal blood glucose) and FAST-negative patients with clear posterior circulation symptoms such as ataxia as per the paramedics’ Clinical Procedures and Guidelines manual; hereafter referred to as FAST (eFigure 1). The time frame for prehospital alerts was symptom onset within the past 6 hours or wake-up stroke. An additional screening tool used in Auckland, New Zealand, called the PASTA score (Prehospital Acute Stroke Triage and Assessment [Figure 1]) was recorded for those patients not receiving a teleambulance assessment. The PASTA score is based on the Los Angeles Motor Scale and aims to identify patients with large vessel occlusion who are candidates for EVT.
BGL = blood glucose level; GCS = Glasgow Coma Scale.
Patient exclusion criteria were insufficient time to complete the telestroke assessment (≤5 minutes to the hospital), transport in a nonrandomized ambulance, failed or ineffective notification of the on-call neurologist, and technical issues precluding completion of the telestroke assessment.
In this study, there was no inadvertent cross-over between randomized groups. An intention-to-treat or, more accurately, an intention-to-assess analysis, including patients who were intended for teleambulance assessment but then could not complete this for various reasons, was not possible because of the lack of the main predictor variable (teleambulance assessment outcome) when intended telestroke assessments were not completed.
Test Methods
In the Wellington Region, when a patient is thought to have a stroke, bystanders will call “111,” and the call will be attended by paramedics from Wellington Free Ambulance. They will assess the patient as per their Clinical Procedures and Guidelines Manual (eFigure 1, links.lww.com/WNL/C310), which includes FAST assessment. If stroke continues to be suspected in accordance with these criteria, paramedics call Wellington Hospital Emergency Department (ED) staff who then call the hospital switchboard to activate an emergency call to the acute stroke team. In this trial, we maintained this flow of information, which informs all the relevant parties.
For the trial, 12 ambulances (with each ambulance being a single cluster) were randomized 1:1 to either receive an iPad tablet device with telestroke capability (6 ambulances) or complete a prehospital PASTA score (6 ambulances). This arrangement captured most of the ambulances in the Greater Wellington region (86%). Teleambulance or PASTA assessments were then performed while the patient was en route to the hospital, in addition to standard care (Figure 2), without affecting patient transport time. Ambulance teams shift between ambulances so that clustering was not linked to a particular set of paramedics.
ADLs = activities of daily living; BSL = blood sugar level; CPG = Clinical Procedures and Guidelines; ED = emergency department; FAST = Face, Arm, Speech, Time score for the identification of stroke patients; GCS = Glasgow Coma Scale; LAMS = Los Angeles Motor Scale; NHI = National Health Institute patient identification number.
The stroke team was alerted for both patient groups through the standard notification system, except that the additional information of whether the patient was in a telestroke-capable ambulance was also provided in the alert text. The neurologists included 2 attending stroke neurologists, 6 attending general neurologists, 2 stroke fellows, and 1 senior neurology resident working at Wellington Hospital.
For the teleambulance assessment, both the neurologist and the paramedics would connect through 4G to a virtual meeting room, using Polycom RealPresence video conferencing software. The neurologist would have a view of the patient and the inside of the ambulance (Figure 3). During the assessment, the neurologist would obtain a brief history of the presentation and time of onset, obtain available history on medications and existing medical comorbidities, and determine usual functional status. They would also perform an NIHSS assessment, with support from the paramedic when examining visual fields, limb ataxia, sensation, and inattention. After this, they would assign the patient to 1 of 5 categories: 1) stroke: IVT eligible, 2) stroke: EVT eligible, 3) stroke: eligible for both IVT and EVT, 4) stroke: not eligible for intervention, and 5) not stroke. The patient would arrive at the emergency department and follow the usual procedure for acute stroke assessment.
(A) The back of a Wellington Free Ambulance, (B) the tablet for telestroke assessment at the foot of the patient, (C) and the tablet showing a telestroke assessment in progress. (D) The paramedic and the patient sit beside each other to look at the tablet during assessment, and (E) the neurologist’s view of the patient from their office. Please note that these images are of a simulation and that usual patient positioning would be more reclined.
The neurologist conducting the ED assessment was not routinely the same individual conducting the prehospital assessment, and they were not informed of the prehospital results. To achieve this, for the purposes of this trial, an additional team member was rostered specifically to conduct teleambulance assessment, who was otherwise not rostered on for hyperacute stroke care. This individual would receive the prehospital alerts and only conduct the prehospital assessment.
The outcome from teleambulance or PASTA assessments was recorded by paramedics and neurologists into a central database before hospital arrival. Eventual treatments received, final clinical diagnosis, imaging results, and baseline clinical data were captured from medical notes by researchers blinded to the treatment arm (C.M. and I.M.S.).
Paramedics received training on how to mount and operate the tablet from Wellington Free Ambulance staff before the study. In addition, the study team provided written education materials on the use of the PASTA score and a video clip demonstrating visual field, heel-to-shin, finger-to-nose, and sensory/inattention testing (A.R.) to paramedics to enhance their ability to assist the neurologist during the NIHSS assessment inside the ambulance. All educational materials were made available to all participating paramedics.
Sample Size and Data Analysis
This study was designed to detect a 20% difference in diagnostic test accuracy with 5% error and a power of 90%. Adjusting for the cluster effect, this required 55 patients in each arm. Because of slower-than anticipated recruitment resulting in a much longer than anticipated study duration, a concurrent change in ambulance provider executive leadership, and the impacts of the COVID-19 pandemic on their team’s priorities, the ambulance provider requested trial termination after 76 of the planned 110 patients had been recruited. Patient enrollment stopped before final study results were available.
Prehospital assessments were compared with both the preimaging neurologist clinical diagnosis in the emergency department (ED) (prespecified primary comparator) and the final discharge diagnosis/reperfusion interventions received by the patient. The ED comparison provides information on discrepancies between a virtual ambulance–based neurologist assessment and an ED in-person neurologist assessment. It also serves as a gold standard for the detection of patients in the field who are candidates for direct transfer to an EVT-capable center without the benefit of imaging. The final diagnosis/intervention received served as a gold standard by which to evaluate the sensitivity, specificity, and accuracy of the 2 diagnostic tests incorporating imaging findings into the decision making. Both comparisons are reported in the Results section.
Outcome variables were compared using the χ2 test to calculate p values and CIs and logistic regression to determine cluster-adjusted odds ratios and intracluster correlation coefficients using Stata IC 16.0.
Standard Protocol Approvals, Registrations, and Patient Consents
The trial was registered with the Australian New Zealand Clinical Trials Registry (ACTRN12619001678189) and received Capital and Coast District Health Board (CCDHB) institutional ethics approval, which included exemption from informed consent. The study protocol and statistical analysis plan are available in eSAP1.
Data Availability
The data from this study can be made available on reasonable request from qualified investigators to the corresponding author and acquisition of appropriate ethics approvals.
Results
Patients who were transported by a Wellington Free Ambulance between August 2019 and September 2020 and screened positive for potential stroke were considered for inclusion in the study (n = 294). Of these, 155 met study inclusion criteria (see study flow diagram, Figure 4). Of these, 35 completed a teleambulance assessment, and 41 had a PASTA assessment. All patients who underwent an assessment were included in the analysis.
*Undocumented = patients picked up by the ambulance and screened as possible stroke, however did not complete teleambulance/PASTA assessment, with reasons often unclear from documentation.
The demographics of the study participants are shown in Table 1. The mean (SD) age of participants was 73.2 (13.6) years in the teleambulance group and 71.4 (15.3) years in the PASTA group, with more females than males in both groups (54.3% and 68.3%). There were more New Zealand Indigenous Māori in the PASTA arm (12.2% vs 0%, p = 0.03). There were no other statistically significant differences between the 2 study arms.
Demographic Characteristics of Study Participants
Accuracy of Teleambulance vs PASTA in Predicting Reperfusion Therapy
The complete set of results for the accuracy of teleambulance vs PASTA assessment can be found in Table 2. Teleambulance had an accuracy (95% CI) of 100% (90%–100%) and PASTA 70.7% (54.5%–83.9%) in predicting the preimaging ED neurologist assessment of reperfusion candidacy (p < 0.001). Teleambulance was 80% accurate (63.1%–91.6%) and the PASTA score 61.0% accurate (44.5%–75.8%; p < 0.001) in predicting actual reperfusion treatments administered. Teleambulance accurately predicted IVT administration in 80% (63.1%–91.6%) and PASTA in 56.1% (39.8%–71.5%) of patients (p = 0.03). For EVT, teleambulance accurately predicted eventual treatment in 88.6% (73.3%–96.8%) and PASTA in 56.1% (39.8%–71.5%) (p = 0.002).
Diagnostic Accuracy of Teleambulance vs PASTA Assessment
Sensitivity and Specificity of Teleambulance vs PASTA
The sensitivity (Sn), specificity (Sp), positive and negative predictive values (PPV and NPV), likelihood ratios (LRs), and C-statistics for teleambulance and PASTA can be found in Table 2.
Teleambulance assessment was highly sensitive and specific in detecting stroke patients eligible for reperfusion therapy when compared with ED assessment (Sn 100%, Sp 100%). When compared with the eventual reperfusion treatment received, it remained highly sensitive (100%; 29.2%–100%), but specificity had reduced (78.1%; 60%–90.7%). In comparison, PASTA was only modestly sensitive and specific, with a sensitivity of 76.5% (50.1%–93.2%) and a specificity of 66.7% (44.7%–84.4%) in detecting stroke patients eligible for reperfusion when compared with ED assessment as well as sensitivity 77.8% (40%–97.2%) and specificity 56.3% (37.7%–73.7%) when evaluated against the final diagnosis/reperfusion intervention received.
Of note, PASTA assessment was highly sensitive in detecting stroke patients who eventually underwent EVT (100%; 29.2%–100%), although was far less specific (52.6%; 35.8%–69.0%). Teleambulance by comparison was highly sensitive (100%; 2.5%–100%) and also specific (88.2%; 72.6%–96.7%) in predicting patients who received EVT. PASTA was less sensitive and specific in predicting eventual IVT treatment (Sn 71.4% [29.0%–96.3%], Sp 52.9% [35.1%–70.2%]).
Odds Ratios and Cluster Adjustment
The odds ratios of teleambulance assessment predicting a particular outcome compared with FAST and PASTA can be found in Table 3. These are presented with and without adjustment for cluster, with patients in the same cluster being those who were transported by the same ambulance. As seen in Table 3, the cluster-adjusted odds ratios were almost identical to the unadjusted odds ratios. The intracluster correlation coefficient (ICC) was well below 0.5, ranging between <0.001–0.18. The results provide evidence that ambulance was not an independent predictor of outcome.
Odds Ratios (ORs) of Predictions Comparing Teleambulance With Ambulance Scores (FAST + PASTA or FAST Alone)
For diagnosis of reperfusion candidacy in the ED, the odds ratios could not be calculated because teleambulance prediction correlated perfectly with the ED diagnosis. The ICC for reperfusion accuracy could similarly not be calculated. However, the cluster-adjusted odds ratio (95% CI) of accurately predicting reperfusion therapy was 14.2 (7.0–29.0) using a teleambulance approach vs the FAST score and 2.6 (1.4–4.6) using teleambulance vs the PASTA score. The cluster-adjusted odds ratio of accurately predicting EVT was 98.2 (21.5–448.3) using a teleambulance approach vs the FAST score and 6.1 (1.4–25.6) using teleambulance vs the PASTA score. For IVT, the cluster-adjusted odds ratio was 5.1 (2.6–10.2) for teleambulance vs FAST and 3.1 (1.6–6.3) for teleambulance vs PASTA.
Logistical and Technical Considerations
Telestroke assessments were sometimes not completed for the following reasons: the neurologist was engaged in conflicting patient assessments, the notification of the stroke team did not include the information that the patient was in a telestroke-equipped ambulance, or there were technical difficulties. Technical difficulties precluding assessment occurred in 6.6% of cases and included 2 instances of unacceptably poor sound quality limiting assessment, 2 calls being affected when the Polycom RealPresence app was down for 24 hours because of IT provider issues, and 1 occasion when the paramedics were not able to set up the equipment. The ambient noise issue was addressed through the use of headphones inside the ambulance. Wi-Fi issues were reported on 3 occasions affecting video quality, but this did not preclude completion of the assessment. Consultation durations were all under 10 minutes, and overall, the neurology team reported low impact on routine clinical workflow. On 4 occasions (11.4%), the ED neurologist was also the teleambulance neurologist because of staffing issues. A sensitivity analysis excluding these 4 patients did not significantly alter the results (eTables 1–2, links.lww.com/WNL/C310).
There were also some challenges in the PASTA arm where eligible patients did not undergo PASTA assessment, presumed to be due to the lack of paramedic familiarity with the study and scoring process and/or competing clinical priorities.
After completion of the trial, the neurology department collectively decided to continue the service as part of usual care to improve prehospital triage and in-hospital stroke team workflow. The ability to call-off code stroke for nonstrokes, plan ED assessments before the arrival of the patient, and enable faster transition from door to CT in patients fully assessed prehospital were features highlighted as particularly beneficial. The risk of additional burden on the neurologist was viewed as minimal because the physician would be assessing the patient regardless once they arrive in the ED. This service model simply shifts this assessment to occur earlier in the patient journey. Technical challenges were all addressed by the end of the trial, and no further issues were reported over the final 6 months.
Classification of Evidence
This study provides Class I evidence that intra-ambulance telestroke evaluation has a greater diagnostic accuracy compared with the PASTA score performed by paramedics in distinguishing hyperacute stroke patients who are candidates for reperfusion therapy.
Discussion
Our study found that teleambulance is highly accurate at predicting candidates for reperfusion therapy, with little to no difference to the assessment performed by neurologists in the emergency department. It remained highly sensitive at predicting final treatment decisions after imaging, and its accuracy was significantly greater than the PASTA score across all patient subgroups. We also found that, despite early remediable challenges, implementing a teleambulance approach was technically feasible. Furthermore, we found that this service model could be integrated into routine workflow and was acceptable to neurologists, although we cannot comment on the potential impact of an overnight service. Our study provides evidence that teleambulance assessment can be successfully used to make accurate diversion decisions to hospitals equipped for EVT, without missing many patients who would benefit from EVT or those who would instead benefit more from faster IVT access at a closer center. Despite the fact that some aspects of neurologic examination may be difficult to perform in a virtual manner (e.g., visual field, sensory, and ataxia assessments), it appears that the teleambulance assessment was not overly hindered by these limitations with neurologists’ prehospital impressions around reperfusion candidacy almost exactly aligning with preimaging neurologist diagnoses in the ED. Providing education materials to paramedics on these specific facets of clinical examination may have been an important component to enabling high-quality teleambulance assessments, although reviewing these materials was not mandatory.
This study also illustrated the limitations of the PASTA tool. PASTA was less accurate than teleambulance in identifying stroke patients for reperfusion, and in particular, it was poor at accurately identifying patients for IVT. This is perhaps to be expected, given that the PASTA score is based on the Los Angeles Motor Scale (LAMS) that was specifically validated in the detection of large vessel occlusion.10 The accuracy of PASTA does appear to be similar in our population to that found in earlier studies using the LAMS,10,11 which is reassuring and indicates that it was likely used accurately by the paramedics in our cohort. However, in some regions in New Zealand, the PASTA score has been viewed as a suitable and sufficiently sensitive alternative to the FAST screening tool, and such a use cannot be supported by our findings.
We were also able to assess the FAST score and found that although it appears to be a useful initial screening tool, it is not very specific with the percentage of false positives comparing very poorly with teleambulance assessment. Given that FAST was part of the screening we used to enter patients into the study, we were not able to assess the sensitivity or negative predictive values of the FAST score to determine the rate with which actual reperfusion candidates may have been missed by this initial screen.
The use of telemedicine in acute stroke assessment by emergency medical services has been piloted previously, although information on diagnostic accuracy and outcomes has been limited, with most reports focusing on feasibility and technical challenges including connectivity issues.12,–,14 This may be due to the limitations of connectivity with a 3G approach, with the benefits of faster transmission achievable through 4G only available in the past few years.15 A recent review confirmed prehospital telestroke feasibility but found very limited evidence on diagnostic accuracy.16 One pilot study from Belgium, again primarily assessing the feasibility and effectiveness of transmitting clinical information, also reported an 83% concordance rate between having a prehospital impression and the final diagnosis of stroke.17 Another study from New Jersey looked at door-to-needle times in a series of 89 potential stroke patients and found that it was reduced from 41 to 28 minutes with a prehospital telestroke approach.18 This study reported that in their prehospital telestroke intervention period, no alteplase-treated patients were missed by telestroke, although no further diagnostic accuracy information was reported.
Our study using a 4G approach was able to demonstrate a high level of diagnostic accuracy, while also achieving a relatively low rate of technical issues (6.6%), all of which had been fully addressed over the past 6 months of the trial.
Another approach to prehospital stroke diagnosis is the use of mobile stroke units (MSUs) that have shown effectiveness in reducing door-to-needle and door-to-groin puncture times,19 as well as reducing the number of emergency department presentations.20 The cost of implementation of this strategy is a major barrier. By comparison, intra-ambulance telestroke is cheaper as regards equipment costs and staffing,18,21 with 1 study reporting MSUs being 179 times more expensive than the teleambulance model.21 Cost becomes an even greater factor in low population density countries such as New Zealand where urban populations and stroke volumes are small, making widespread implementation of MSU impractical. In addition, teleambulance effectively simply shifts the specialist assessment to earlier in the patient journey with the ability to abort neurology team ED assessments and thus might further reduce staff resource implications or exert at least a neutral effect. However, it is also possible that this type of assessment adds additional workflow disruption to stroke teams with adverse flow on effects not necessarily captured in prior economic analyses. Our neurology team felt that any negative impacts did not outweigh the positive ones on workflow and patient care, which provides reassurance, but a full economic impact assessment as part of a future study is planned to provide more quantitative data in this area.
There were several limitations to this study. The first was practical barriers. For example, in our effort to deviate as little as possible from the hospital’s usual hyperacute stroke protocol, we used the multistep alerting process from the emergency department to the hospital switchboard, from where an alert was forwarded to the acute stroke neurologist. The multistep process proved challenging for a research project, and occasionally, trial-related information was not effectively communicated. There were also a significant number of patients who could potentially have been included; however, for a number of reasons, the paramedics did not complete either the teleambulance or PASTA assessment in some patients (Figure 4). Based on the documentation by paramedics, it was not always possible to tell the reasons why a teleambulance/PASTA assessment had not occurred, but it was largely attributed to short travel distance to the hospital, unfamiliarity with the study protocol, and/or other competing clinical demands requiring deprioritization of a research study based on reports from the Wellington Free Ambulance team. Although this occurred in both study arms and groups were generally well balanced, we cannot exclude that this may have resulted in a degree of selection bias. In addition, this may signal a potential limitation to either PASTA/complex prehospital paramedic scores or teleambulance implementation, although over time the barriers became much less prominent suggesting this challenge can be addressed. For the study, this also meant overall recruitment was slower than expected, which resulted in early termination and a final sample size 30% lower than the preset target sample. This along with an already relatively small sample size could have biased the outcomes especially around secondary endpoints, and these should be interpreted with a degree of caution. A further challenge was technical issues including ambient noise, which fortunately were resolved early in the study. In a small number of cases (11%), the ED and the teleambulance neurologist were the same individual potentially biasing the initial ED assessment; however, based on our sensitivity analysis, this did not significantly affect our results. Also, given this was a small single-center study, the results may not be broadly generalizable to all geographic regions, especially because multiple different stroke scales are used in different areas around the world. Further studies from other regions with larger sample sizes and using other stroke scales will provide additional evidence to support widespread roll-out of this approach.
In summary, we found that teleambulance is superior to paramedic screening tools in making appropriate prehospital triaging decisions, is acceptable to clinical teams, is feasible to implement, and is appropriate for incorporation into routine clinical care. The service model is now being rolled out to a wider area within New Zealand, and we plan to assess the impact of teleambulance in clinical practice on door-to-needle/groin times, patient outcomes, and service costs using a sequential comparison. In areas such as Wellington, New Zealand, where low population density and resource constraints make mobile stroke units impractical, we believe that teleambulance, or mini-MSU, is a very viable alternative likely applicable to many other regions around the globe.
Study Funding
This study did not receive any external funding. The equipment costs and Wellington Free Ambulance staff costs were funded internally by Wellington Free Ambulance, and the hospital clinician/researcher time was funded through existent hospital/university contracts.
Disclosures
A. Ranta discloses research funding from the New Zealand Health Research Council and the Ministry of Health; neither sources of funding applied to this project. The other authors report no relevant disclosures.
Acknowledgment
The authors acknowledge the contribution of the wider Wellington Hospital Neurology and Wellington Free Ambulance teams without whose contribution this study would not have been possible.
Appendix Authors
Footnotes
-
Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.
-
Submitted and externally peer reviewed. The handling editor was José Merino, MD, MPhil, FAAN.
-
Editorial, page 825
-
Class of Evidence: NPub.org/coe
-
Infographic: links.lww.com/WNL/C451
- Received February 1, 2022.
- Accepted in final form June 28, 2022.
- © 2022 American Academy of Neurology
