Abstract
Prehospital stroke care is the first link in the stroke chain of survival and includes symptom recognition, engagement of the Emergency Medical Services (EMS) system, timely and effective dispatcher response, and emergency medical response. Prehospital stroke screening tools are an important component in guiding the EMS response to stroke and proper triage of patients. Additionally, there is a growing body of research focused on applications for telemedicine, mobile stroke units, and diagnostic testing in the prehospital arena. Prehospital stroke care is integral to the organization of regionalized stroke systems. Implementation of stroke systems of care can lead to improved access to specialty services and improved patient health outcomes. In addition to increasing access to acute stroke care, telestroke shows great potential for integrating stroke systems of care and facilitating interactions between centres.
Prehospital stroke care is the first link in the stroke chain of survival and includes symptom recognition, engagement of the Emergency Medical Services (EMS) system, timely and effective dispatcher response, and emergency medical response. Prehospital stroke screening tools are an important component in guiding the EMS response to stroke and proper triage of patients. Additionally, there is a growing body of research focused on applications for telemedicine, mobile stroke units, and diagnostic testing in the prehospital arena. Prehospital stroke care is integral to the organization of regionalized stroke systems. Implementation of stroke systems of care can lead to improved access to specialty services and improved patient health outcomes. In addition to increasing access to acute stroke care, telestroke shows great potential for integrating stroke systems of care and facilitating interactions between centres.
Prehospital Stroke Care
Achieving optimal patient health outcomes after stroke depends on effective systems of care and begins in the prehospital setting. There are many important components of prehospital stroke care, including symptom recognition, engagement of dispatch, deployment of EMS, appropriate prehospital triage, and pre-arrival notification to the receiving hospital.
Symptom Recognition
Background
Patient or bystander recognition of stroke symptoms is necessarily the first step in the stroke chain of survival (Powers et al., 2018). Once symptoms are recognized, stroke patients or bystanders must also be aware of the importance of immediately engaging the EMS system. Yet public awareness of stroke warning signs is poor (Jurkowski et al., 2010; Madsen et al., 2015); less than 60% of stroke patients use EMS (Mochari-Greenberger et al., 2015); and nearly half of calls to EMS systems for stroke events are made more than 1 hour after symptom onset (Mosley et al., 2007). For this reason, prolonged onset-to-arrival time is the greatest source of delays in care and a frequent cause of ineligibility for reperfusion therapy (Madsen et al., 2016; Matsuo, et al. 2017; Centers for Disease Control and Prevention, 2007). Improving timeliness of patient presentation depends on public awareness and symptom recognition.
Evidence
Public education programmes may be effective in improving public awareness, highlighting stroke signs and symptoms, the urgency of immediate care, and the need to call the EMS system for suspected stroke symptoms. In two studies, participants were administered structured questionnaires assessing knowledge of stroke incidence, symptoms, consequences, and behaviour in case of a stroke. The group completing the questionnaire after a widespread informational campaign demonstrated better knowledge about stroke than did the pre-initiative participants (Chiti et al., 2007; Nishikawa et al., 2016). Another study assessed the effect of a multimedia stroke education campaign and found that individuals in the intervention region demonstrated increased awareness of stroke symptoms and the importance of immediately calling 911 in case of stroke relative to individuals in a control region (Jurkowski et al. 2010). A randomized controlled trial (RCT) in Texas found that classroom instruction significantly improved middle school children’s knowledge of stroke symptoms and intention to call 911 (Morgenstern et al., 2007). The FAST mnemonic (Facial droop, Arm weakness, Speech disturbance, Time to call 911) and its language-independent Stroke 112 version may also be effective (Zhao et al., 2018). A pre-post study of elementary school students found that students’ knowledge of stroke symptoms substantially improved with a hip-hop curriculum incorporating the FAST components (Williams and Noble, 2008); and a FAST educational campaign directed towards adults found that nearly all participants recalled stroke warning signs at 3-month follow-up (Wall et al., 2008). Yet public education programmes must be sustained to maintain their benefit. A study using a television advertising campaign in Ontario found that public awareness of stroke warning signs increased during the period of the campaign and Emergency Department (ED) stroke presentations increased, but decreased during an advertising blackout (Hodgson et al., 2007, 2009). Public education programmes should also provide instructions to reach diverse populations. For example, the Massachusetts Department of Public Health provides stroke education written materials and videos in multiple languages including Spanish, Portuguese, and Khmer Cambodian (see Figure 3.1) (Massachusetts Health Promotions Clearinghouse, 2018). In a systematic review, among 15 studies of mass media campaigns and community initiatives aiming to reduce patient delays by promoting the signs and symptoms of a stroke, the majority showed positive intervention effects, although methodological rigour was variable (Mellon et al., 2015).
Figure 3.1 Stroke Heroes Act FAST Poster in English (a), Khmer Cambodian (b), Portuguese (c), and Spanish (d).
By improving public awareness and symptom recognition, such campaigns may lead to earlier ED presentation and increase the proportion of patients eligible for acute therapies. A multilevel intervention conducted in East Texas targeted public stroke identification, outcome expectations, and social norms in addition to provider norms and behaviour. In this study, rate of tissue plasminogen activator (tPA) treatment increased significantly in the intervention community and was unchanged in the comparison community (Morgenstern et al., 2002).
Comment
The available limited evidence from randomized controlled trials and prospective time series studies, and expert consensus, supports the importance of public education for stroke symptom recognition, the importance of seeking care urgently, and of engaging EMS for suspected stroke (Powers et al., 2018; Higashida et al., 2013). Such programmes should be designed to maintain efficacy and be culturally competent to reach diverse populations, especially targeting those populations at highest risk.
Dispatcher Response
Background
Once stroke symptoms are recognized and patients or bystanders make the decision to call the EMS system, dispatchers must be able to recognize stroke signs and symptoms and dispatch the highest level of care available with maximal efficiency.
Evidence
The appropriate EMS response to a potential stroke call begins with dispatcher recognition. Studies of consecutive stroke dispatches found that a Medical Priority Dispatch System structured caller interview algorithm enabled dispatchers to identify stroke patients with 41–83% sensitivity (Ramanujam et al., 2008; Buck et al., 2009); and in an intervention study, the Emergency Stroke Calls: Obtaining Rapid Telephone Triage (ESCORTT) training package improved dispatcher recognition of stroke from 63% to 80–88% of cases (Watkins et al., 2014). In reviews of 911 calls and EMS run sheets of patients with calls for potential stroke, dispatchers were able to identify stroke with high reliability if the caller described facial droop, weakness/fall, or impaired communication or used the word stroke (Reginella et al., 2006; Richards et al., 2017). Another analysis of consecutive patients found that emergency dispatcher stroke recognition was associated with higher rates of on-scene stroke recognition and stroke scale performance by paramedics and emergency medical technicians (EMTs) (Oostema et al., 2018).
Comment
While research devoted to the role of medical dispatch is limited, available evidence and expert consensus underscore the important role that dispatchers play in the rapid recognition of potential stroke and expedient mobilization of EMS (Acker et al., 2007; Higashida et al., 2013). Suspected stroke calls should be given the highest priority dispatch and predetermined plans should guide dispatcher response with algorithms taking into account stroke severity, patient last known well time, distances to various hospitals, and capabilities of regional hospitals (Acker et al., 2007; Higashida et al., 2013; Mission: Lifeline Stroke, 2017). In order to ensure appropriate response, dispatcher guidelines should be promoted that prioritize stroke response, and call centres will require adequate funding for resources and continued training.
Emergency Medical Response
Background
The emergency medical response is a crucial component of prehospital stroke care and includes stabilization in the field, ground or air transport, and hospital pre-arrival notification. Despite significant fragmentation and regionalization in the organization of EMS systems in the USA and worldwide (Evenson et al., 2009; Higashida et al., 2013), the primary goals of emergency medical response are consistent throughout: rapid evaluation, early stabilization, and rapid transport and triage to an appropriate level facility.
Evidence
Despite the importance of timely presentation, many stroke patients still do not use EMS transportation to the hospital. In analyses of patients presenting to hospitals participating in Get with the Guidelines – Stroke, fewer than two-thirds of patients arrived by EMS (Ekundayo et al., 2013; Mochari‐Greenberger et al., 2015). EMS activation was more likely among older patients, patients with Medicaid and Medicare insurance, and patients with more severe stroke; it was less likely among patients of minority race and ethnicity and those living in rural communities (Ekundayo et al., 2013; Mochari‐Greenberger et al., 2015). The reasons for this observed disparity remain poorly understood.
There is a paucity of evidence to support requiring a specific level of prehospital provider response in standard ambulances for suspected stroke patients, among the options of EMTs who are basic life support (BLS) providers, paramedics who are advanced life support (ALS) providers, and, in Europe and some other regions of the world, physicians. One single-centre study of 203 patients with altered level of consciousness found no significant difference in admission rate, mortality, or disposition for patients transported by BLS versus ALS prehospital providers; however, only about 20% of the study population were stroke patients and the study did not assess other important outcomes such as time benchmarks and tPA delivery (Adams et al., 1996). In EMS systems in which BLS providers greatly outnumber ALS providers, restricting first response or transport to ALS providers would likely increase prehospital delays.
Some observational studies have considered the role for helicopter transport of patients in rural locations and in interfacility transfers. Helicopter transport has important potential to improve access to primary and comprehensive stroke centres. One study found that, in 2014, one-third of the US population did not have 60-minute drive time access to a primary stroke centre (PSC) by ground transport, but 91% of the population would have access to a PSC by air (Adeoye et al., 2014). In a large observational study of data from 32 stroke units between 2003 and 2009, patients transported by helicopter had shorter arrival times and higher thrombolysis rates when compared with patients transported by ambulance (Reiner-Deitemyer et al., 2011). Smaller studies have noted similar findings, illustrating the potential for helicopter transport to shorten transport times and extend access to thrombolytic therapy, stroke unit care, and study enrolment (Silliman et al., 2003; Hawk et al., 2016). For interhospital transfers, observational studies examining transport times and safety have found that helicopter transport was faster than ground and not associated with significant differences in complications, morbidity, or mortality (Svenson et al., 2006; Olson and Rabinstein, 2012; Hesselfeldt et al., 2014; Hutton et al., 2015). For endovascular thrombectomy, observational studies have found that helicopter transport of likely large vessel occlusion (LVO) patients directly from the field or by interfacility transfer can increase availability and timeliness of intervention to remotely located patients (Gupta et al., 2016; Regenhardt et al., 2018). Furthermore, helicopter transfer of ischaemic stroke patients for potential thrombolysis has been reported to be cost-effective ($3700 per quality-adjusted life-year [QALY]) (Silbergleit et al., 2003) and has potential to extend recruitment into acute stroke trials for patients in remote areas (Leira et al., 2006, 2009).
Several screening tools have been developed to enable accurate identification of stroke patients in the prehospital setting. The following scale and screening tools were the first to be developed and are now the most widely deployed:
Cincinnati Prehospital Stroke Scale (CPSS): The CPSS is a three-item examination of facial weakness, arm strength, and speech disturbance. In a study comparing physician and paramedic scoring of 171 patients, there was high agreement for both total score (correlation: 0.89, 95% confidence interval [CI]: 0.87–0.92) as well as for each scale item (Kothari et al., 1999). Another study, examining EMS stroke recognition and accuracy, reviewed 441 EMS-transported stroke patient cases. Identification of stroke in the prehospital setting was higher among cases with CPSS documentation than in cases without (sensitivity 84.7% versus 30.9%, positive predictive value 56.2% versus 30.4%) (Oostema et al., 2015).
Los Angeles Prehospital Stroke Screen (LAPSS): The LAPSS uses medical history and fingerstick glucose items to exclude acute stroke mimics (history of seizures, hypoglycaemia or hyperglycaemia, baseline ambulatory status, onset >24 hours prior), and examination for unilateral face, arm, and/or hand weakness to identify the most common types of stroke. One study retrospectively applied the LAPSS to 48 acute stroke patients and found that the LAPSS had a sensitivity of 92% (Kidwell et al., 1998). Another study assessed paramedic use of the LAPSS after a training and certification process; among 206 patients on whom the LAPSS was completed, sensitivity was 91%, specificity was 97%, positive predictive value was 86%, and negative predictive value was 98% (Kidwell et al., 2000).
Many additional stroke recognition instruments have subsequently been developed and used in the prehospital setting, generally similar in character, including the Melbourne Ambulance Stroke Screen (MASS), the Ontario Prehospital Stroke Screen (OPSS), the Medic Prehospital Assessment for Stroke Code (Med PACS), the Face Arm Speech Test (FAST), and the Recognition of Stroke in the Emergency Room (ROSIER) (Brandler et al., 2014; Rudd et al., 2016).
Systematic reviews of the performance of these stroke screens have found that most perform fairly well, with relative advantages and disadvantages to each. In one systematic review of eight studies, the LAPSS and OPSS had higher point estimates for accuracy than the CPSS (44–95%), although confidence intervals overlapped, while the MASS, Med PACS, ROSIER, and FAST had less favourable operating characteristics (Brandler et al., 2014). In another systematic analysis that included 21 studies, some not fully published, higher sensitivity point estimates were seen for the CPSS and FAST, while the LAPSS and similar instruments had higher specificity point estimates (Rudd, 2016). For the CPSS, sensitivity point estimates ranged from 44–95% and specificity from 24–79%, while for the LAPSS, sensitivity point estimates ranged from 59–91% and specificity from 48–97%.
In addition to stroke screening tools, prehospital instruments have been developed to assess stroke severity and patterns of stroke deficits in order to identify: (1) patients with LVOs among all acute cerebral ischaemia (ACI) patients (thrombectomy-capable stroke centre-appropriate patients), and (2) patients with LVO-ACI or with intracerebral haemorrhage from among all acute focal cerebrovascular disease patients (comprehensive stroke centre [CSC]–appropriate patients). The following three stroke severity scales were among the earliest developed and have undergone at least partial validation in actual prehospital settings:
Los Angeles Motor Scale (LAMS): The LAMS assigns points to each of the LAPSS exam items of face, grip, and arm weakness, yielding a total score of 0–5 when there is unilateral weakness and 0–10 when there is bilateral weakness (Nazliel et al., 2008). In a field validation study, when applied by paramedics to consecutive patients enrolled at the prehospital scene in a clinical trial, a LAMS score ≥4 identified LVOs among ACI patients with 76% sensitivity and 65% specificity, and CSC-appropriate patients among all suspected focal stroke patients with 73% sensitivity and 71% specificity (Noorian et al., 2018).
Rapid Arterial oCclusion Evaluation (RACE) Scale: The RACE assigns points for face, arm, and leg weakness; gaze deviation; inability to follow verbal commands; aphasia; and unawareness of limb weakness, for a total 0- to 9-point scale. In a mixed field and interfacility transfer validation study, when applied by paramedics to patients arising from these site types, a RACE score ≥5 identified LVOs among ACI patients with sensitivity of 85% and specificity of 65% (Perez de la Ossa et al., 2014).
Cincinnati Stroke Assessment Tool (C-STAT): The C-STAT assigns points for arm weakness, gaze deviation, and inability to state age or month or follow verbal commands, for a total 0- to 4-point scale (Katz et al., 2015; McMullan et al., 2017). In a field validation study, when applied by prehospital personnel to a subset of transported stroke patients, a C-STAT score ≥2 identified LVOs among ACI patients with 71% sensitivity and 70% specificity, and CSC-appropriate patients among all suspected focal stroke patients with 57% sensitivity and 79% specificity (McMullan et al., 2017).
Many additional stroke severity instruments have been developed for prehospital use; they are generally similar in character to the LAMS, RACE, and C-STAT, but not yet validated in actual prehospital studies. They include the Field Assessment Stroke Triage for Emergency Destination (FAST-ED), Prehospital Acute Stroke Severity (PASS) scale, Vision-Aphasia-Neglect (VAN) scale, 3-item Stroke Scale (3i-SS), and Emergency Medical Stroke Assessment (EMSA) (Noorian et al., 2018; Gropen et al., 2018). In studies comparing simultaneous application of 5 (Zhao et al., 2017) and 7 (Noorian et al., 2018) of these scales at the same time to consecutive patients in the ED, all generally performed comparably in identifying LVO and CSC-appropriate patients.
Mobile telemedicine permitting remote stroke physicians to evaluate potential stroke patients at the scene or in standard ambulances has great potential. Studies evaluating feasibility have had mixed results; technical challenges have been frequent, but agreement on National Institutes of Health Stroke Scale (NIHSS) scores between remote and in-person examiners range from moderate to excellent (Liman et al., 2012; Van Hooff et al., 2013; Wu et al., 2014; Chapman Smith et al., 2016). A systematic review of telemedicine in prehospital care concluded that neurovascular disease was the most common focus and telemedicine improved the prehospital diagnosis of stroke (Amadi-Obi et al., 2014).
A technologically advanced approach to prehospital stroke care has been the development of mobile stroke units (MSUs), ambulances equipped with a computed tomography (CT) scanner, a CT technologist, a mobile blood laboratory, a critical care nurse, and a stroke physician either in person or via telemedicine, in addition to a paramedic or EMT (Figure 3.2). Deployment of mobile stroke units has been shown to shorten onset to thrombolysis treatment times in two cluster-controlled trials. In a single centre, 100-patient trial in Homburg, Germany, median time from dispatcher calling a stroke alarm to decision to administer intravenous thrombolysis was reduced in weeks that the MSU was active, from median 76 to 35 minutes, without differences in safety endpoints (Walter et al., 2012). In the larger Pre-Hospital Acute Neurological Treatment and Optimization of Medical care in Stroke (PHANTOM-S) trial in Berlin, comparing 3213 patients treated during MSU-active weeks and 2969 during control weeks, alarm-to-thrombolysis treatment time was reduced from 76 to 52 minutes for patients treated by a Stroke Emergency Mobile (STEMO) unit, with no increased risk for intracerebral haemorrhage (Ebinger et al., 2014). The proportion of patients receiving thrombolysis within the golden hour of 60 minutes from onset was 31% for MSU-treated patients versus 4.9% in control patients, with no difference in mortality but increased likelihood of discharge home (Ebinger et al., 2015).
Figure 3.2 A Stroke Emergency Mobile (STEMO) unit.
All CT scanners in mobile stroke units can perform CT angiography, and some can perform perfusion CT, so that in the prehospital setting MSUs can definitively identify patients with LVOs causing ACI appropriate for direct routing to thrombectomy-capable receiving hospitals (John et al., 2016; Fassbender et al., 2017). In the PHANTOM-S trial, MSU deployment significantly shortened median alarm-to-imaging time, from 50 to 35 minutes (Ebinger et al., 2014).
An additional future potential treatment approach is paramedic delivery in standard ambulances of neuroprotective therapies that do not require brain imaging to distinguish ischaemic and haemorrhagic stroke prior to initiation. Several prehospital randomized trials have demonstrated the feasibility of prehospital start of potentially neuroprotective agents (Ankolekar et al., 2013; Hougaard et al., 2014; Saver et al., 2015). The largest was the Field Administration of Stroke Therapy-Magnesium (FAST-MAG) phase 3 trial, a multicentre, double-blind, randomized controlled study in which patients with suspected stroke in the prehospital setting were randomized to receive intravenous magnesium sulfate or placebo within 2 hours of symptom onset. Among the 1700 patients enrolled, there were no significant differences in disability at 90 days, mortality, or all serious adverse events (Saver et al. 2015). The study did, however, establish the feasibility of rapidly achieving informed consent and delivering a study agent to suspected hyperacute stroke patients in the prehospital setting, with median time from last known wellness state to start of neuroprotective agent of 45 minutes.
Arrival at the hospital by ambulance has also been associated with improved emergency treatment compared with private transport. In an analysis of 35,936 patients from Get with the Guidelines-Stroke (GWTG-S), patients arriving by ambulance were more likely to have brain imaging completed within the guideline-recommended 25-minute window (Kelly et al., 2012). This finding has been shown in other state-level analyses as well (Patel et al., 2011; Sauser et al., 2014a). In an analysis of 204,591 patients from 1563 GWTG-S hospitals, arrival by ambulance was associated with increased likelihood of tPA administration among eligible patients (odds ratio [OR] 1.47 versus non-ambulance arrival), and with increased likelihood of tPA treatment within 60 minutes (OR: 1.44) (Ekundayo et al., 2013). Arrival by ambulance has also been associated with shorter door-to-needle times for thrombolysis in Michigan’s Coverdell state stroke registry data (Sauser et al., 2014b).
While ambulance arrival in itself has been associated with improved care delivery and treatment times, pre-arrival notification by prehospital providers has been shown to lead to further improvements in stroke care by enabling earlier resource mobilization. In a state-level analysis of 13,894 patients with suspected stroke, time to brain imaging and to imaging interpretation was faster in patients arriving by ambulance versus those who did not, but was fastest among those arriving by ambulance with hospital pre-arrival notification (Patel et al., 2011). In a single-centre review of 229 stroke patients, prehospital notification by EMS was associated with improved time to stroke team arrival and to CT scan completion and interpretation, and was associated with increased rates of intravenous thrombolysis (McKinney et al., 2013). Another single-centre review of 102 suspected stroke patients found that prehospital notification was associated with significantly shorter door-to-imaging and door-to-needle times (Bae et al., 2010). A third single-centre review of 118 acute stroke patients found improved door-to-imaging times and increased rates of thrombolysis in the patients for whom the hospital had pre-arrival notification (Abdullah et al., 2008). A review of 371,988 patients presenting to GWTG-S hospitals found that pre-arrival notification was associated with increased likelihood of tPA administration among eligible patients and shorter median door-to-imaging times, door-to-needle times for thrombolysis, and meeting national benchmarks (door-to-imaging within 25 minutes, door-to-needle within 60 minutes, onset-to-needle within 120 minutes, and tPA use within 3 hours) (Lin et al., 2012).
Comment
Prehospital care has significant downstream effects on stroke patients’ care delivery and outcomes, yet many stroke patients still do not activate EMS transportation to the hospital. Stroke education campaigns need to highlight the importance of early use of EMS, especially among minorities. For patients in rural and remote settings, there is a role for helicopter transport when ground transportation would take more than 1 hour. For all potential stroke patients transported by ambulance, consistent use of a validated prehospital stroke identification tool (e.g. CPSS, LAPSS) enables early and reasonably accurate identification of patients with acute cerebrovascular disease, enabling improved hospital pre-arrival notification and timely treatment upon hospital arrival. In addition, consistent use of a validated prehospital instrument for assessing stroke severity (e.g. LAMS, C-STAT, RACE) allows reasonably accurate identification of patients with LVOs and with intracerebral haemorrhage for potential direct routing to more advanced stroke centres. Stroke severity assessments should be deployed when two-tier routing of select patients directly to thrombectomy-capable and CSCs is advantageous, and a single scale should be used consistently within regional EMS systems. More advanced diagnostics and therapeutics in the prehospital setting, including prehospital telemedicine for stroke recognition, paramedic delivery of neuroprotective therapies, and mobile stroke unit provision of CT scanning and thrombolytic therapy in the field, show substantial promise, but further research is necessary to evaluate their utility and cost-effectiveness.
Figure 3.3 Field triage principles.
Regionalized Stroke Systems
Stroke Systems of Care
Background
The American Stroke Association has highlighted the need for organized and coordinated stroke systems of care in order to promote access to evidence-based stroke care. Barriers to access are substantial, particularly in rural and neurologically underserved areas, and fragmentation in care leads to suboptimal care delivery and treatment, as well as safety concerns and inefficiencies (Schwamm et al., 2005; Acker et al., 2007). A stroke system of care is a framework that integrates regional stroke facilities, EMS, and public and governmental agencies and resources and coordinates access to the full range of stroke prevention, treatment, and rehabilitation. This may include implementation of telemedicine and aeromedical transport in order to increase access in neurologically underserved areas (Schwamm et al., 2009a).
Potential elements in a hierarchical, regional stroke system of care include hospitals of five capacity levels: non-stroke centre hospitals, Acute Stroke Ready Hospitals (ASRHs), PSCs, Thrombectomy Stroke Centres (TSCs), and CSCs (Figures 3.3 and 3.4 and Table 3.1). In the USA, ASRHs, PSCs, TSCs, and CSCs are certified by the Joint Commission, by other national accrediting bodies, or by state departments of health (Powers et al., 2018).
A non-stroke centre hospital has not made an institutional commitment to maintain a capable stroke care service 24/7/365. These hospitals should be bypassed by ambulances carrying stroke patients, and should have protocols for rapidly transferring patients who self-present with stroke or develop stroke as inpatients to facilities with stroke care capability.
An ASRH has made an institutional commitment to provide initial emergent intravenous therapies to stroke patients and has written protocols for emergency stroke care, access to emergency brain imaging and laboratory testing at all times, the ability to administer intravenous tPA and coagulopathy reversal drugs, and written transfer agreements to efficiently send patients after initial stabilization to more advanced stroke centres. The ASRH does not have fully organized inpatient stroke care or a stroke unit, and some resources may be via telemedicine.
A PSC has all of the components of an ASRH; in addition, a PSC has a fully organized inpatient stroke care system or stroke unit. PSCs can provide both emergency and full inpatient care for the preponderance of stroke patients, except for those who require endovascular, neurosurgical, or neurological intensive care unit care (Higashida et al., 2013).
A TSC has all of the components of a PSC and in addition provides around-the-clock access to endovascular mechanical thrombectomy for acute ischaemic strokes due to LVOs.
A CSC has all of the components of a PSC and in addition has around-the-clock access to state-of-the-art care including endovascular techniques, the presence of a neurological intensive care unit, and neurosurgical services in order to provide care for the most complex stroke patients (Higashida et al., 2013; Powers et al., 2018).
Figure 3.4 The Stroke Care Pyramid in the United States.
ASRH | PSC | Comment | |
---|---|---|---|
EMS training provided | + | + | ≥2 h annually |
ED stroke education (annual hours) | + (4 h) | + (8 h) | |
Acute stroke team with at least 2 members | + | + | 15 min response time |
Stroke protocols for treatment and stabilization | + | + | Annual revisions |
Laboratory testing, ECG, CXR results available within 45 min of ordering | + | + | 24/7 availability |
Brain imaging completed and read within 45 min of ordering | + | + |
|
IV tPA for eligible patients, with goal DTN ≤60 min | + | + | 24/7 availability |
Stroke unit, including protocols and telemetry | * | + | |
Neurosurgical services available within: | 3 h | 2 h | Onsite or by transfer |
Telestroke | Initiated within 20 min when indicated | If hub site, response within 20 min of request | |
Transfer of patients if indicated | To PSC or CSC | To CSC | Within 2 hours of ED arrival |
ASRH: Acute Stroke Ready Hospital. PSC: Primary Stroke Centre. EMS: Emergency Medical Services. h: hour/s. ED: Emergency Department. min: minutes. ECG: electrocardiogram. CXR: chest x-ray. tPA: tissue plasminogen activator. DTN: door-to-needle time. CSC: Comprehensive Stroke Centre.
* Only required if patients are admitted.
Evidence
The evidence for the hospital elements of regional stroke systems of care focuses on the distinctive advantages of PSCs and CSCs. The remaining levels are gap-fillers: ASRHs provide some of the services of PSCs in regions without a PSC, and TSCs provide some of the services of CSCs in regions without a CSC.
Primary Stroke Centres
A wide range of studies indicate that outcomes for all patients with acute ischaemic and haemorrhagic stroke are improved if care is delivered at facilities with organized inpatient stroke care. An analysis of 6223 ischaemic stroke patients in the Registry of the Canadian Stroke Network used an organized care index (OCI) to quantify the amount of hospitals’ organized stroke care (namely, the presence of physical and occupational therapies, a stroke team, and a stroke unit). Patients treated at hospitals with higher OCIs had lower odds of 30-day mortality, and the number needed to treat to prevent 1 death at 30 days ranged from 4 to 9 across ischaemic stroke subtypes (Smith et al., 2010). In a German study, five network community hospitals with specialized stroke services and telemedical support from academic hospitals were matched with five community hospitals without specialized stroke care. Care delivery and outcomes of 3122 stroke patients presenting to these hospitals were studied; network hospitals were more likely to meet quality of care indicators, and patients treated in network hospitals were less likely to have poor outcomes (death, institutional care, or disability) (Audebert et al., 2006). An observational study of 30,947 ischaemic stroke patients treated at hospitals in New York State studied the relationship between admission to a stroke centre and use of thrombolytic therapy and patient mortality, finding that patients admitted to stroke centres were more frequently treated with tPA and had lower mortality (Xian, 2011). Other observational studies have also found that PSCs have a shorter time to imaging and to tPA treatment and higher tPA treatment rates (Rose et al., 2008; Sung et al., 2010).
Transitions of hospitals and regional care systems to organized inpatient care have been associated with improved patient outcomes. In the USA, a matched-hospital analysis examining 732 hospitals and 173,985 ischaemic stroke admissions found that hospital adoption of the Get with the Guidelines-Stroke hospital-based quality improvement programme was associated with reduced patient mortality and increased patient discharge to home rather than institutional care (Song et al., 2016). In London, a citywide hub-and-spoke model was implemented centred on eight hyperacute stroke units designed to treat suspected acute stroke patients during the first 72 hours of symptom duration, with faster response times and 24/7 access to specialist care. A pre–post analysis found a 12% relative reduction in deaths and a total 90-day cost savings of £5.2 million per year (Hunter et al., 2013).
Controlled clinical trials also support the benefits of organized inpatient care. A meta-analysis of 28 trials and 5855 participants found that stroke unit care reduced 1-year mortality (OR: 0.81), reduced odds of death or institutionalized care (OR: 0.78), and reduced the odds of death or dependency (OR: 0.79) (Figure 3.5) (Stroke Unit Trialists’ Collaboration, 2013).
Figure 3.5 Organized stroke unit care versus alternative source.
Outcome: death or dependency by the end of scheduled follow-up.
Structural aspects of stroke care (i.e. stroke centre certification) or performance recognition programmes best able to identify high-performing hospitals need consideration. In a hospital-level examination of 1356 hospitals participating in GWTG-S from 2010–2012, hospitals were classified by PSC status and by whether they had received recognition for performance with a GWTG-S achievement award. When hospitals were examined for conformity to quality of care measures for ischaemic stroke, PAA recognition was a stronger predictor of high performance than was PSC certification (Fonarow et al., 2013).
Comprehensive Stroke Centres
Care for select patients at CSCs is supported by the simple fact that many effective endovascular and neurosurgical therapies are available only at CSCs or TSCs, mandating patient access to advanced care sites. Endovascular thrombectomy for acute ischaemic stroke due to LVOs (Goyal et al., 2016), coiling or clipping of ruptured saccular aneurysms (Falk Delgado et al., 2017), minimally invasive haematoma evacuation for intracerebral haemorrhage (Tang et al., 2018), decompressive hemicraniectomy for large hemispheric infarcts (Alexander et al., 2016), and neurointensivist care of critically ill stroke patients (Knopf et al., 2012) are all distinctively or only available at CSCs.
In addition, multiple studies indicate that patients with complex neurovascular conditions have better outcomes when cared for at advanced care facilities with high-volume experience. For example, a relation between case volume and outcome is well-established for subarachnoid haemorrhage. In a meta-analysis of four observational studies with a total of 36,600 subarachnoid haemorrhage patients, mortality was decreased with care in high-volume hospitals: OR: 0.77, 95% CI: 0.60–0.97, p < 0.01 (Boogaarts et al., 2014). In an analysis of 32,336 subarachnoid haemorrhage patients in the United States Nationwide Inpatient Sample, mortality rates at hospitals with 100, 80, 60, 40, and 20 cases per year increased from 18.7% to 19.8%, 21.7%, 24.5%, and 28.4%, respectively (Pandey et al., 2015).
The relationship between high-volume centres and improved outcomes among patients with subarachnoid haemorrhage suggests that preferentially transporting certain types of suspected stroke patients to stroke centres may optimize their outcomes. An observational analysis examined 61,685 ischaemic stroke patients treated in 333 hospitals in Finland. Hospitals were classified as CSCs, PSCs, or general hospitals according to Brain Attack Coalition criteria, and patients treated at stroke centre hospitals had lower 1-year mortality and reduced institutional care compared with those treated at general hospitals, and over the 9-year follow-up median survival was increased by 1 year for patients treated in stroke centres (Meretoja et al., 2010).
Concentrating care for complex neurovascular patients at advanced stroke centres has been shown to be cost-effective from a societal perspective. One cost-utility analysis compared two scenarios for subarachnoid haemorrhage: regionalization of care in which patients at low-volume hospitals (<20 cases annually) would be transferred to hospitals with high volume (≥20 cases annually) versus a scenario in which every patient is treated at the geographically closest hospital. The study found that transferring a patient with subarachnoid haemorrhage (SAH) from a low- to high-volume hospital would gain 1.6 QALYs at a cost of $10,548/QALY (well within the generally accepted range for cost-effectiveness from the societal perspective) (Bardach et al., 2004). A similar cost-effectiveness analysis examined the role for transferring patients with intracerebral haemorrhage from hospitals without specialized neurological intensive care units to centres with specialized units. As there are limited high-quality data on the effect of neurological intensive care units on functional outcomes, the analysis explored various favourability assumptions and found a cost of $47,000/QALY and $91,000/QALY in the favourable and moderately favourable scenarios – straddling the societal willing-to-pay threshold (Fletcher et al., 2015). Among patients with ischaemic stroke, treatment with tPA and treatment with endovascular thrombectomy have been shown to be highly cost-effective (Joo et al., 2017; Shireman et al., 2017). An analysis of endovascular thrombectomy found that catheter treatment increased QALYs by 1.74 years and decreased lifetime medical costs by $23,000 (Shireman et al., 2017). Therefore, treatment of large vessel ischaemic stroke patients at centres with higher rates of endovascular thrombectomy delivery is likely to be cost-effective.
Comment
Implementation of stroke systems of care can lead to improved access to specialty services and improved patient outcomes. By preferentially transporting all stroke patients to centres where organized inpatient care and thrombolytic therapy are more likely and more timely, and select stroke patients to centres where advanced endovascular and surgical therapies are expertly and efficiently available, regionalization of stroke care is desirable and cost-effective from both the patient and the societal perspectives. Patients should be escalated up the stroke pyramid to centres of increasing capability when needed; when appropriate, patients should also flow down the pyramid to lower-cost community centres when specialty services are no longer required. Stroke systems should also establish written protocols for interhospital patient transfers, using the trauma system as a model (Acker et al., 2007). Given the substantial international and intranational variations in rules and regulations, geography, and resources (Higashida et al., 2013; Lindsay et al., 2016), implementation of stroke systems of care will necessarily be distinct and customized for each region or locality and may encounter significant obstacles, namely substantial cost and resource constraints (Schwamm et al., 2005). While such constraints may necessitate an incremental approach to implementation of a system of care, it is important to note that such costs may be offset by the potential cost savings realized by individual hospitals and facilities (Schwamm et al., 2005). Additionally, telemedicine resources may offer a cost-effective solution for access to specialty care and keeping patients at community hospitals. Finally, PSCs, CSCs, and stroke systems of care must provide ongoing education to the public and to key personnel and continually evaluate the effectiveness of their care delivery for continuous quality improvement (Powers et al., 2018; Higashida et al., 2013).
Telemedicine
Background
Telemedicine utilizes telecommunication technologies to provide medical information and services. Telemedicine for stroke care (i.e. telestroke) can connect hospitals without full-time neurological or radiological services to around-the-clock acute stroke expertise. Telestroke has the potential to fill the gap that exists in many parts of the USA that otherwise lack access to acute stroke services (Schwamm et al., 2009a, 2009b; Demaerschalk et al., 2017), may allow more hospitals to become ASRHs or PSCs (Powers et al., 2018), and, in providing emergent neurological expertise, may overcome a major barrier to tPA utilization (Brown et al., 2005; Schwamm et al., 2009b; Gadhia et al., 2018). The current structure of telestroke arrangements ranges from small partnerships to large multi-hospital affiliations (hub-and-spoke model) to for-profit transactional relationships (see Figure 3.7) (Schwamm et al., 2009a).
Evidence
The utility of telestroke in the acute hospital-based evaluation of stroke patients has been demonstrated in reports from multiple countries and regions, including Ontario (Waite et al., 2006), Germany (Audebert et al., 2006), Swabia (Wiborg et al., 2003), Texas (Choi et al., 2006), Massachusetts (Schwamm et al., 2004), and rural community hospitals in Georgia (Hess et al., 2005). The most studied utilization of telestroke is with respect to thrombolysis eligibility. Telestroke has been given a Class I recommendation based on Level A evidence in support of its use for National Institutes of Health Stroke Scale (NIHSS) assessment, timely review of brain imaging, and determination of tPA eligibility (Higashida et al., 2013).
The feasibility and utility of assessing NIHSS via telestroke using high-quality, dedicated videoconferencing software has been established. Inter-rater agreement for the NIHSS between bedside examination by a stroke neurologist and remote examination through telestroke by a stroke neurologist has been found to be very high both for console telemedicine equipment, with correlation coefficients of 0.97 and 0.96 (Shafqat et al., 1999; Wang et al., 2003), and for smartphone equipment, with correlation coefficient of 0.95 and kappa of 0.98 (Demaerschalk et al., 2012; Anderson et al., 2013).
Engagement of telestroke has also been shown to increase the use of tPA in eligible patients. A review of 655 stroke patients treated at two community hospitals east of Houston found that, after implementation of a telemedicine project, the rate of tPA treatment more than quadrupled (Choi et al., 2006). Implementation of telestroke led to administration of tPA in rural or island settings where thrombolytics were not previously regularly used (Schwamm et al., 2004; Wang et al., 2004). In a controlled trial, among 234 stroke patients at four spoke sites randomly assigned to use of televideo versus telephone to determine suitability for tPA treatment, correct treatment decisions were more often made in the telemedicine group (OR: 10.9, 95% CI: 2.7–44.6) (Meyer et al., 2008).
In addition to increasing tPA utilization, telestroke may decrease treatment times for tPA administration. An analysis of 50 tPA-treated patients in the REACH telestroke network found that mean onset to treatment time was shorter than other stroke care delivery systems (127.6 minutes vs 145.9 minutes) (Switzer et al., 2009) and improved over time as the system became more efficient (Hess et al., 2005). While onset to treatment times were shorter, it is not clear whether this was also true for door-to-needle times, as this was not reported.
Telestroke is associated with similar long-term functional outcomes and mortality to traditional tPA delivery at the bedside. The STRokE DOC study found no difference in rates of intracerebral haemorrhage after tPA or in 90-day functional outcomes or mortality in telemedicine-treated patients compared with that to be expected from bedside evaluation (Meyer et al., 2008). A telestroke pilot project in Germany examined all patients receiving telemedicine-authorized thrombolysis in 12 regional centres in a telestroke network and found no significant difference in rates of symptomatic haemorrhage after thrombolysis, or in 1-week or in-hospital mortality compared with in person-authorized thrombolysis at two stroke centre hubs (Audebert et al., 2006). Long-term outcomes of 3- and 6-month mortality and functional outcomes after thrombolytic treatment were similar between patients treated at telemedicine-linked community hospitals and stroke centre hospitals (Schwab et al., 2007). In a comparison of stroke patients treated by telestroke in community hospitals and those treated at the Helsinki University Central Hospital hub, there was no significant difference in the proportion of patients with favourable outcomes (Sairanen et al., 2011).
Finally, telestroke in the acute treatment of ischaemic stroke is likely to be cost-effective. A cost-effectiveness analysis compared hub-spoke telestroke network care with treatment in remote EDs without telestroke consultation or stroke experts. The analysis found that telestroke results in a cost of $2449 per QALY over a lifetime horizon (Nelson et al., 2011).
Telemedicine also has important applications in radiological evaluation of acute stroke patients. A pilot study comparing stroke neurologists’ use of telemedicine to read 72 head CT scans versus gold standard readings of hard copies found excellent reliability of stroke neurologists’ reads (kappa statistic = 1.0), sensitivity of 100%, and specificity of 100% (Johnston and Worrall, 2003). A review of telestroke cases in the Partners Telestroke Network similarly found that radiology reads by telestroke were effective for identifying tPA exclusions and making decisions regarding thrombolysis administration (Schwamm et al., 2004).
In a review of 100 head CT images from patients with suspected stroke, remote image review by telestroke showed 100% agreement for images of acute ischaemic stroke, intracerebral haemorrhage, metastasis, and normal scans, but only 88% agreement (7 of 8) for cases of subarachnoid haemorrhage (Phabphal and Hirunpatch, 2008). As expected, teleradiology is subject to equipment and reader variability. In a study of 582 possible stroke patients in the Stroke Eastern Saxony Network (SOS-NET), stroke neurologists’ CT reads during telemedical consultation had discrepancies from neuroradiologist reads in 43 patients (8%), 9 of which were clinically relevant (1.7%) (Puetz et al., 2013).
Studies have also indicated the feasibility of teleradiology image review via smartphones for timely and accurate interpretation of CT scans in possible cases of acute stroke. One study of 120 noncontrast head CT images and 70 CT angiography images from the Calgary Stroke Program database compared neuroradiologist reads via workstation versus smartphone and found that the sensitivity, specificity, and accuracy of detecting haemorrhage were 100% with perfect agreement (kappa = 1) (Mitchell et al., 2011). Diagnosis of acute parenchymal ischaemic change, dense vessel sign, and vessel occlusion on CT angiography were good (parenchymal ischaemic change: kappa 0.8; dense vessel sign: kappa 0.69, vessel occlusion: kappa 1.0) with no significant difference in interpretation time between devices. In another study of 74 acute ischaemic stroke patients, neuroradiologists accurately diagnosed presence or absence of LVOs on CT angiography on smartphones, with 100% agreement versus workstation diagnosis (Hidlay et al., 2018).
The use of mobile telestroke in ambulances has also been shown to be feasible. In 27 ambulance runs of standardized patients, the correlation on prehospital NIHSS scores assessed by a remote rater using tablet-based telemedicine and an in-person rater was 0.96 (Chapman Smith et al., 2016). Remote evaluation permits physicians to contribute to the assessment and routing decisions for stroke versus nonstroke and potential LVO versus non-LVO ischaemic stroke. In mobile stroke units, mobile telestroke assessment enables physicians to make decisions regarding prehospital start of intravenous thrombolysis. Among 50 consecutive mobile stroke unit patients, a remote telemedicine neurologist rendered comparable thrombolytic decisions to an in-person neurologist (Bowry et al., 2018) (see Figure 3.6).