Endovascular thrombectomy (EVT) is now the standard of care for eligible patients with acute ischemic stroke (AIS) secondary to emergent large vessel occlusion (ELVO). However, there remains uncertainty in how hospital systems can most efficiently route patients with suspected ELVO for EVT treatment. Given the relative geographic distribution of centers with and without endovascular capabilities, the value of prehospital triage directly to centers with the ability to provide EVT remains debated. While there are no randomized trial data available to date, there is substantial evidence in the literature that may offer guidance on the subject. In this review we examine the available data in the context of improving the existing AIS triage systems and discuss how prehospital triage directly to endovascular-capable centers may confer clinical benefits for patients with suspected ELVO.
Statistics from Altmetric.com
Endovascular thrombectomy (EVT) is the standard of care for the treatment of acute ischemic stroke (AIS) secondary to emergent large vessel occlusion (ELVO).1 2 Previously, when the timely administration of intravenous alteplase was the primary focus of treatment, patients were emergently routed to any hospital equipped to administer alteplase. However, now that treatment emphasis has shifted to EVT, state and regional healthcare systems are striving to adjust the former triage and transfer models to rapidly route patients to medical centers with endovascular capabilities. Incorporating EVT into the treatment algorithm and hospital referral networks is needed to reflect the new standard of care, but the optimal patient triage strategy to employ remains debated.1–3 The debate has generated the following question: Should patients with suspected ELVO (1) undergo initial evaluation at a more proximal non-endovascular-capable center (NECC) prior to potential transfer to an endovascular-capable center (ECC) for EVT, or (2) be triaged in the prehospital setting and routed directly to an ECC for evaluation and treatment?4 The optimal solution remains debated, though several investigators have studied the question in detail.5–7 In this review we examine the available and relevant data, including emergency medical service (EMS) triage models, transportation times, delays to alteplase treatment, in-hospital efficiency delays, and reliable detection of ELVO, assessing how each of these may be used to inform AIS triage systems while definitive evidence continues to be collected.
Best available evidence
Currently, there exists a limited number of studies that have collected real-world data about prehospital AIS triage protocols. The scarcity of real data has driven researchers to produce mathematical models to assist in evaluating the optimal triage strategies for patients with suspected ELVO. A type of mathematical modeling known as conditional probability (CP) modeling has been particularly insightful. Briefly, CP modeling applies real data from the literature to a formula expressing the probability of an event, given the probability that another separate event has already taken place. In the case of ELVO, such events might include favorable functional outcome and successful reperfusion, for instance. Researchers have used these models to consider the relative benefits of the two divergent triage strategies introduced above: initial NECC triage (also known as ‘drip-and-ship’) and prehospital triage directly to an ECC (also known as ‘mothership model’ or ‘NECC bypass’).8 9
In the first study of its kind, Holodinsky and colleagues created a CP model that featured two hospitals, one a NECC and one an ECC.8 The transport time between the two hospitals, the distance from the scene to each center, door-to-needle (DTN) time, door-to-puncture (DTP) time, and alteplase/EVT recanalization efficiency were all included in the model (table 1). Functional independence (modified Rankin Scale (mRS) 0–2) was used as the outcome measure. Using high quality data from the ESCAPE trial (Endovascular Treatment for Small Core and Proximal Occlusion Ischemic Stroke) and time estimates from Get With The Guidelines: Target Stroke Initiative, the authors showed that prehospital triage directly to an ECC leads to superior functional outcomes for all cases in which a NECC is less than 45 min away from an ECC, including when driving past the NECC is necessary to reach the ECC.8 As transport time between the two centers increases beyond 45 min, initial NECC triage becomes favorable for cases in which patients are located near the NECC and if driving past the NECC is necessary to reach the ECC.8 Practically, these results suggest that, when transport time between the two centers is under 45 min, patients should be routed directly from the field to an ECC, but if greater than 45 min, the field location should be used to decide between the two triage strategies. However, the conclusions ought to be interpreted cautiously as the model featured two centers in isolation, a great over-simplification of the complex hospital networks that exist in practice, and hospital performance metrics are variable.
The same group of researchers followed this work with a practical study, applying similar CP modeling and data to all NECCs and ECCs in the state of California and the Canadian provinces of Alberta and Ontario.9 The authors input a similar set of variables, including DTN time, and also used functional independence (mRS 0–2) as the outcome measure. The model produced geographic layouts suggesting that prehospital triage directly to an ECC leads to superior functional outcomes in the majority of cases. An exception arises when the NECC and ECC are exceedingly far apart, although, unlike in the previous study, a driving time threshold was not presented. Another exception arises when NECCs are able to achieve a DTN time under 30 min. Efficient evaluation and administration of alteplase make initial NECC triage a viable strategy, even when in close proximity to an ECC. In practice, however, DTN times under 30 min appear difficult to achieve, even with optimized protocols in place.10 Despite process improvements, recent reports show that even DTN times under 60 min are often unfeasible, and continue to comprise the minority of cases.11 12
More recently, Schlemm and colleagues have introduced a practical CP model that incorporates unknown ELVO status at the time of triage.13 Previous models assume that, on arrival to the scene, EMS are aware of whether or not the patient has an ELVO, a fact that is not clearly apparent under realistic conditions. The Schlemm model, however, accounts for a range of stroke severity, including both ELVO and non-ELVO, that is assessed on-scene using a prehospital stroke scale. The model uses a prehospital stroke scale known as the RACE (Rapid Arterial Occlusion Evaluation) score for detection of ELVO. The RACE score is a 9-point scale that assesses a set of exam findings, including facial asymmetry, arm/leg weakness, and gaze deviation.13 14 The scale has variable sensitivity and specificity for ELVO, depending on the cut-off score that is used. A score of 5 or greater is generally used to signify that ELVO is likely (sensitivity 0.85, specificity 0.68). Using each of the variables, including RACE score, the Schlemm CP model generates a temporospatial map that depicts the triage strategy that is most favorable in each geographic location. As seen with previous CP model studies, the results demonstrate that NECC performance measures (ie, door-in-door-out time and DTN time) and transport time between the NECC and the ECC are crucial factors that have an important role in triage decision-making. In addition, though, the Schlemm model temporospatial map depicts the RACE score cutoffs that are most favorable for prehospital triage directly to an ECC in each geographic location. For instance, a low RACE score cut-off for prehospital triage directly to an ECC is favored when patients are located near an ECC, when the transport time between the NECC and the ECC is short, and when NECC performance measures remain suboptimal. High RACE score cut-off for prehospital triage to an ECC is favored when patients are located close to the NECC, when NECC processes are streamlined and optimized, and when transport times between the NECC and the ECC are long. The addition of the RACE score to the model creates a more situation-specific tool that can be applied to individual patients, rather than a strict, uniform, location-based triage rule.
Altogether, CP modeling results demonstrate that prehospital triage directly to an ECC appears to be favorable in the majority of cases, given the current transport/treatment efficiencies and alteplase/EVT outcomes. However, as hospital networks adjust their transfer and transport processes, mathematical models will likely require adjustment.10 15
Experience in practice
While the CP models appear to support prehospital triage directly to an ECC with up to 45 min of transport time between NECCs and ECCs, high quality evidence from real-world experience (ie, randomized trial data) has yet to confirm these predicted results. The RACECAT (Direct Transfer to an Endovascular Center Compared with Transfer to the Closest Stroke Center in Acute Stroke Patients With Suspected Large Vessel Occlusion, NCT02795962) trial, a multicenter randomized trial based in Spain, is currently enrolling patients and is precisely designed to investigate the effect that prehospital triage directly to ECCs may have on functional outcomes for ELVO. While no randomized trial data are available to date, several published studies have examined the topic (table 2).
In 2016, a group of investigators in northwestern Ohio published a study comparing patients with AIS who had arrived at the regional ECC, after implementation of a protocol permitting prehospital triage directly to an ECC, with historical controls.16 While the study design was primarily intended to test whether the RACE score, a prehospital stroke scale intended to identify ELVO,14 was effective in triaging ELVO patients to ECCs, the results have important implications for AIS triage models. Surprisingly, the median time from EMS dispatch to patient arrival at the emergency department (ED) was nearly the same between the prehospital triage group and the historical control group (31 vs 32 min), indicating that institution of the prehospital triage protocol did not significantly prolong EMS transport times. Among patients with confirmed ELVO on evaluation at the ECC who were subsequently treated with EVT, a greater percentage of patients from the group triaged directly to an ECC had functional independence (mRS 0–2) at follow-up (50% vs 36.4%) and fewer died (14.3% vs 27.2%), although these results did not reach statistical significance.16
In contrast to the data above, gathered from a mixed rural/suburban region of Ohio, a study performed in Paris, France at two hospital centers 3.2 miles apart (25 min by ground transportation) with overlapping catchment areas provides insight into the transfer dynamics that face urban medical centers.17 In the study, one hospital was a NECC and the other an ECC. EMS personnel were instructed to transport patients to the nearest hospital with available bed capacity. Patients first evaluated at the NECC who were subsequently transferred to the ECC for EVT evaluation were compared with those presenting directly to the ECC. Whether treated with alteplase and EVT or with EVT alone, functional independence rates were not significantly different between the two groups although, importantly, the cohort presenting directly to the ECC had higher National Institutes of Health Stroke Scale (NIHSS) scores and lower ASPECTS (Alberta Stroke Programme Early CT Score) at admission, indicating a worse prognosis at the outset. Onset-to-needle (ie, time from onset to alteplase treatment), onset-to-puncture (ie, time from onset to EVT treatment), and needle-to-puncture times were all significantly shorter in the group presenting directly to the ECC, suggesting greater in-hospital protocol efficiencies at the ECC.17
A study in central Denmark examined a similar pair of hospitals, one a NECC and the other an ECC.18 The two hospitals, however, were located much farther apart, approximately 75 miles (120 km). Similar to the study performed in northwestern Ohio, the study intended to test the utility of a prehospital stroke scale in identifying and directing patients with suspected ELVO to the ECC to reduce delay to EVT. Therefore, prehospital triage directly to the ECC was carried out for patients with suspected ELVO. The results of the study revealed that median time from first medical contact to arrival at the ECC ED was significantly reduced with prehospital triage directly to the ECC. Interestingly, the study authors found that institution of the protocol did not significantly prolong the time to alteplase treatment for patients who bypassed a NECC due to suspicion of ELVO. This may have been due to greater in-hospital efficiencies at the ECC.18 Patients treated with EVT after implementation of prehospital triage protocol had significantly higher rates of functional independence (mRS 0–2) at 90 days follow-up and, importantly, independent functional outcomes for patients treated with alteplase alone were not significantly affected. Thus, prehospital triage directly to the ECC, applied in this region, effectively reduced treatment delays to EVT without prolonging delays to alteplase treatment or influencing outcomes for those treated with alteplase alone.
Another study, carried out by investigators of the STRATIS (Systematic Evaluation of Patients Treated With Neurothrombectomy Devices for Acute Ischemic Stroke) registry, examined a large cohort of patients with AIS to assess the outcomes of prehospital triage directly to ECCs.19 20 The STRATIS registry includes 55 ECCs spread across the USA, an advantageous design that effectively addresses regional differences inherent in single-system/region studies. In the analysis, a significantly greater proportion of patients in the prehospital triage group had independent function (mRS 0–2) at follow-up.19 Importantly, however, patients treated by prehospital triage directly to an ECC had significantly lower NIHSS scores and higher ASPECTS scores at admission, suggesting they may have had a better prognosis from the outset. Onset-to-reperfusion time was significantly lower with prehospital triage to ECCs, supporting previous findings.16 17
Transportation time delays
A primary concern with prehospital triage directly to ECCs is that additional transfer time to reach an ECC may render patients who would have been eligible for alteplase at a more proximal NECC ineligible when they reach the ECC, due to treatment time window restrictions. In recognition of this concern, the authors of the aforementioned studies have examined the timing differences that face these two AIS triage models.16–18 21 Based on the studies reviewed here, median transport time to an ECC varies depending on geographic location, from 23 to 56 min (table 3). The median expected time delays to EVT for patients treated through a model without prehospital triage directly to an ECC range from 45 to 124 min (table 3).
Data from the ESCAPE trial suggest that improving on these lengthy time delays is possible, since arrival to the ECC ED was prolonged by just 34 min for patients who were first evaluated at a NECC.22 However, a pooled individual data meta-analysis of five randomized trials found that the time from symptom onset to arrival at the ECC ED is 142 min greater when the patient arrives via transfer from an NECC, suggesting much longer delays that include both NECC evaluation time and transport time.23 Inter-hospital transfer data collected at an urban hub-and-spoke hospital network in Australia offered similar findings suggestive of long NECC transfer delays.24 The authors reported a median NECC door-in-door-out time of 106 min, more than five times the median of 19.5 min needed to transport patients from the NECC to the ECC. Prolonged transfer times from NECCs, despite relatively short transport distances between NECCs and ECCs, may result in fewer attempted EVTs on arrival to the ECC.25
The STRATIS registry analysis presents the largest direct examination of the alteplase eligibility concern to date.21 The authors performed hypothetical calculations for patients transferred to ECCs via ground transportation and found that, had all patients instead been directly routed to the ECC, 4.9% (6/122) would have become ineligible for alteplase due to added transfer time.21 Accordingly, the median onset-to-needle time for these patients would have increased by 17 min, though the median onset-to-puncture time would have decreased by 81 min for EVT-eligible patients. When the authors then restricted the analysis to only those within 20 miles driving distance, the percentage of patients that would have become ineligible was reduced to 2.8% (2/71). The results from the study in central Denmark support the same conclusion, that few patients become ineligible for alteplase as a result of prehospital triage directly to an ECC.18 The authors specifically noted that the prehospital ECC triage protocol instituted in the study did not significantly affect onset-to-needle times for patients who had bypassed a NECC. The prehospital ECC triage protocol did, however, significantly reduce the onset-to-puncture times for patients requiring EVT.18 The STRATIS registry investigators note that some of the time lost to added transportation time may be recouped by the inherent efficiencies of ECCs, a concept that has been confirmed in other studies.16 17 21
Some healthcare systems have attempted to eliminate transportation time altogether by creating mobile interventional stroke teams (MIST) that can be dispatched to provide on-site triage and EVT.26 One institution found that dispatching a MIST in cases of suspected ELVO reduced initial DTP time by 79 min and reduced the time to reperfusion by 67 min.26 While highly efficient for patient care, the MIST model is exceedingly resource intensive. It requires that processes be in place to rapidly assemble all the members of an endovascular team and for neurointerventional providers to have the flexibility to respond to stroke codes in geographically separated locations. The NECCs served by MISTs must also maintain facilities equipped for interventional treatment and have the proper care settings and personnel to manage patients after EVT treatment.
Over-triage of ELVO
AIS can be challenging to identify in the prehospital and ED settings, and only a minority of patients with suspected AIS have an ELVO.14 16 27–29 Diverting patients with stroke mimics to ECCs may burden these hospitals with a greater volume of patients who would have otherwise been appropriately cared for at a NECC.30 Ischemic stroke mimics—including encephalopathy, seizures, and intracerebral hemorrhage (ICH)—are not uncommon in the suspected AIS patient population, comprising 25.5–31% of such cases.28 29 However, some stroke mimics, notably ICH and subarachnoid hemorrhages, would likely benefit from direct admission to an ECC, where continuous neurosurgery coverage may offer additional treatment options for these patients. This was specifically noted in the study by Zaidi et al, in which 61.5% of stroke mimics admitted as a result of prehospital triage directly to an ECC required admission to the intensive care unit, a service typically present at ECCs but not universally available at NECCs.16 Moreover, the authors estimated that, had the study patients instead been triaged at the NECC, 66.1% would have required subsequent transfer to a tertiary care medical facility.16 This finding indicates that prehospital triage in cases of suspected ELVO may improve healthcare system efficiency and optimize EMS resources. False positive ELVO rates can likely be lowered with focused training, improved prehospital stroke scales, and the development of ambulance-based ELVO detection devices.14 16 31 On the other hand, the transfer time delays incurred by under-recognition of ELVO in the prehospital setting reduces the likelihood of favorable outcomes, and the individual and societal costs of such outcomes are devastating and costly, respectively.32
For those patients with suspected ELVO triaged at the NECC who may qualify for interventional treatment at an ECC, vascular imaging prior to transfer to an ECC is typically recommended to prevent unnecessary transfer, so long as it does not unreasonably delay transfer time.33 Appropriate vascular imaging prior to ECC transfer, including perfusion weighted imaging in select cases, has been identified as an effective method to reduce the number of EVT-ineligible and stroke mimic transfers that do not require care at an ECC.33
Much as at NECCs prior to transfer, vascular imaging in the prehospital setting may also reduce over-triage of suspected ELVO to ECCs. Mobile stroke units (MSU), ambulances equipped with CT angiography, and specially-trained personnel qualified to administer alteplase infusion while en route to the hospital can detect the presence of ELVO and provide definitive prehospital triage to the appropriate treatment center (NECC or ECC).34 35 Currently, however, mobile stroke units have a limited role in prehospital stroke care due to remaining questions regarding their clinical benefit and the high financial costs associated with establishing such a program.36
Analogous triage model: ST-elevation myocardial infarction
In devising the optimal solutions for stroke care networks, it may be useful to look at the ways in which systems designed for similar diseases are organized. Despite a few fundamental differences in predisposing pathophysiology and the cardiovascular responses to infarction of cerebral and cardiac tissue, there are remarkable parallels between the treatment of ELVO and the treatment of ST-elevation myocardial infarction (STEMI).37 ELVO and STEMI have analogous ischemic pathophysiology and favorable outcomes for both diseases are highly dependent on the rapidity of reperfusion.22 23 38 39 Both diseases can be treated with fibrinolysis and/or EVT. The distribution of hospitals capable of delivering each type of STEMI treatment are comparable to those for ELVO treatment, with some hospitals able to deliver fibrinolysis only and others capable of both fibrinolysis and percutaneous coronary intervention (PCI).40–42 Similar to EVT in ischemic stroke, PCI is more efficacious than fibrinolysis alone.43 44 Unlike for ELVO, the feasibility and potential benefit of prehospital triage to a PCI hospital in the setting of STEMI has been demonstrated repeatedly in multiple large multicenter registries.45–47 Hospital systems have therefore adjusted to guidelines and the numerous favorable analyses describing the benefits of prehospital triage to PCI hospitals in cases of suspected STEMI.
High quality data have shown that the median estimated driving time between non-PCI hospitals and PCI hospitals is approximately 39 min.48 In one study, the median estimated additional transport time needed to bypass non-PCI centers that are within 20 miles of a PCI center was 19 min.48 Another study found the additional transport time needed was around 16 min.45 While these lengths of time may appear small, in the context of the guidelines recommending first medical contact to balloon time of 90–120 min, they represent a significant proportion. Interestingly, despite these stringent time goals, large hospital networks have embraced the need for prehospital triage directly to PCI hospitals in cases of suspected STEMI.45 The response to similar time metrics for ELVO triage decisions remains to be seen.
The American Heart Association (AHA) recommends that, for cases in which additional transport time is less than 30 min, patients with suspected STEMI in the prehospital setting should bypass non-PCI hospitals in favor of direct transport to a PCI-capable medical center.49 As a thought experiment, one could extrapolate this recommendation to ELVO time goals. With onset to alteplase and EVT treatment intervals of 3–4.5 hours and 6 hours, respectively, the recommendation would allow for additional transport time approaching 1.5–3 times those for STEMI (ie, approximately 45–90 min). This assumes the distribution of NECC and ECC hospitals is comparable to that of non-PCI and PCI hospitals, which data suggest is the case.40–42 Hospital systems have already begun to recognize that the extrapolation of STEMI triage transfer protocols may benefit patients with ELVO.50 A shift in EMS triage and routing of patients with suspected ELVO will likely continue as additional evidence is used to inform evolving hospital networks.7
Despite the lack of Level I evidence on the topic, there exist considerable data to guide potential changes to existing transport schemes. The available evidence suggests that prehospital triage directly to ECCs may hold clinical benefits for patients, but that benefits are limited to geographic areas within reasonable proximity to an ECC. Conditional probability models predict that 45 min driving distance between a NECC and an ECC is the upper limit at which benefit from prehospital triage directly to an ECC is observed. Evidence from practice suggests that even longer transport times may continue to confer benefit, and extrapolation of analogous studies in the STEMI literature suggests that 45–90 min of additional transport time might also be reasonable.
Recently, the AHA released the Mission Lifeline statement that recommended a maximum of 15 min additional transport time when performing prehospital triage directly to an ECC.51 However, based on the available evidence in the literature, this recommendation appears conservative.52 Importantly, though, there is likely no universal rule which will apply to all hospital networks and in all geographic regions. Fifteen minutes of additional transport time may only adequately capture suspected ELVO patients in large metropolitan areas with a high population density and multiple ECCs, while underestimating the potential benefit of extended travel times for those in more rural areas.10 15 53 54 There are many factors that contribute to the complexities that face AIS care, including prehospital triage systems, rapid response multidisciplinary team processes, robust data collection and analysis, and critical feedback mechanisms to ensure continued process improvement. Each is unique to a community’s geographic location. As a result, it is likely that each community will need to individually determine its own optimal transport practices.10 15 50 53 54 Ultimately, the interest of the patients must remain paramount and we should continue to analyze and evolve our systems towards their benefit.
Although definitive high-level evidence from practice has yet to emerge regarding the optimal AIS triage protocols and transport system characteristics, there are substantial data to suggest that prehospital triage directly to ECCs holds valuable clinical benefits for patients with suspected ELVO. Based on the available literature, it appears likely that prehospital triage directly to ECCs may be beneficial for patients with suspected ELVO when the ECC is within 30–45 min of additional travel time.
Contributors All authors contributed to the manuscript through manuscript composition and critical review. All authors provided final approval for publication.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.