Introduction
Stroke is a leading cause of death both in the United States and world-wide.Reference Benjamin, Virani and Callaway1,Reference Heron2 Outcomes for acute ischemic stroke (AIS) patients are correlated with the time the brain tissue remains ischemic.Reference Saver, Fonarow and Smith3 Therefore to optimize outcomes, care relies on prompt recognition of suspected stroke and urgent neuroimaging studies such as computed tomography (CT) scans or magnetic resonance imaging (MRI), followed by intravenous (IV) recombinant tissue plasminogen activator (IV tPA) and/or endovascular stroke treatment for eligible patients.Reference Adams, del Zoppo and Alberts4–Reference Powers, Derdeyn and Biller11
There have been numerous efforts to reduce the time to CT scan and to IV tPA treatment for patients upon arrival to hospital by streamlining and standardizing stroke response protocols.Reference Tilley, Lyden, Brott, Lu, Levine and Welch12–Reference Schwamm, Fonarow and Reeves14 Another approach brings diagnostic and treatment capabilities to the prehospital setting via a Mobile Stroke Unit (MSU). A MSU employs a specially trained team onboard an ambulance equipped with telemedicine capabilities and a CT scanner. This allows Emergency Medical Services (EMS) teams to evaluate for AIS and initiate thrombolysis in the field, before transport to a hospital.
The first MSU was implemented in Germany in 2010.Reference Walter, Kostpopoulos and Haass15 In 2014, the first MSU in the United States began operation in Houston, Texas,Reference Parker, Bowry and Wu16 followed by others in Cleveland, Ohio; Denver, Colorado; and the first MSU to operate 24 hours/day, seven days/week in Toledo, Ohio.Reference Calderon, Kasturiarachi, Lin, Bansal and Zaidat17 The number of operational MSUs in the United States continues to rise. These MSUs have the potential benefit of avoiding emergency department (ED) delays by moving stroke treatment from the ED to the ambulance.Reference Parker, Bowry and Wu16
The MSU program in Berlin, Germany has yielded the largest study on MSU treatment times to date. Findings from Berlin included a decrease in time from EMS activation to IV tPAReference Ebinger, Winter and Wendt18 and an increase in the proportion of patients receiving IV tPA within one hour of symptom onsetReference Ebinger, Kunz and Wendt19 when compared to standard hospital-based treatment. Studies in the United States have been smaller in scale but yielded similarly promising results for MSUs.Reference Bowry, Parker and Rajan20
However, implementation of an MSU program is costly in terms of equipment and personnel. An alternative method to expediting stroke care was implemented at Zuckerberg San Francisco General Hospital (ZSFG; San Francisco, California USA) in July of 2017. The “Mission Protocol” (MP) is an acute stroke quality improvement intervention that utilizes prehospital activation by EMS and a direct-to-CT transfer of suspected stroke patients upon hospital arrival. The MP provides a step-by-step guide for EMS and ED personnel that defines specific roles for each provider in order to streamline and expedite every aspect of emergency stroke care.
San Francisco has nine stroke centers within a 47 square mile area with an average EMS transport time for the highest acuity emergencies of 12 minutes and 31 seconds in 2019. The goal was to compare stroke evaluation and treatment times under the MP to those achieved in previously published MSU studies to evaluate whether there is a theoretical advantage to one system over another in a relatively compact city with short EMS transport times, such as San Francisco.
Methods
Study Design
This was a retrospective cohort study of all AIS patients that were evaluated in the two-year period following MP implementation (July 10, 2017 through June 30, 2019). Study endpoints in the cohort were tested against pooled measures from published reports.
Mission Protocol – Study Setting and Population
Located in San Francisco, California, ZSFG serves 106,000 patients each year as the only public, safety net hospital in the city.Reference Ehrlich21 It is an academic, tertiary care institution affiliated with the University of California San Francisco (San Francisco, California USA), with Level 1 trauma center and advanced primary stroke center designations.
All patients in the study period were included in the analysis if they: (1) were brought by ambulance to the ED with a MP activation; (2) had confirmed AIS (by either head CT or MRI) or were treated as a stroke with thrombolysis but ultimately determined to be a stroke mimic; and (3) had all variables of interest available in the clinical documentation. Study approval was obtained from the University of California San Francisco Institutional Review Board [IRB # 17-22932] with a waiver of informed consent.
Mission Protocol – Intervention
The MP is a team-based, direct ambulance-to-CT transfer protocol for suspected stroke patients arriving by ambulance to the ED. Under the current EMS Stroke Protocol, EMS providers notify ZSFG from the field with a “stroke alert” if a patient has a positive Cincinnati Prehospital Stroke Scale (CPSS). An ED nurse fielding the call confirms patient eligibility, and if “Last Seen Well” time is less than six hours, a MP is activated, prompting an ED nurse, Emergency Medicine physician (EMP), and neurologist to assemble prior to patient arrival at Mission Drop-Off, a designated location near the ambulance bay. This team receives a structured handoff communication from EMS, including vital signs and fingerstick blood glucose level.
The MP outlines a well-defined sequence of events via a flow chart and specific roles for each member of the care team. The EMP screens for potential life threats, rules out readily apparent stroke mimics, and decides whether the patient should proceed to CT scanner or go to a resuscitation room either for stabilization or further work-up of non-stroke causes of symptoms. The ED nurse draws blood from an established prehospital IV line. The neurologist further evaluates for stroke and assesses eligibility for IV tPA and other stroke treatments.
If the patient is deemed ready for a CT scan, the patient is transferred from ambulance gurney directly to CT scanner table. Based on the non-contrast CT scan, IV tPA is administered to the patient directly on the CT scanner table while a second IV line is established by the ED nurse for CT angiogram and CT perfusion studies which are obtained during IV tPA infusion.
Outcome Measures
Primary study endpoints were two common stroke treatment metrics: (1) time from EMS activation to CT (EMS-CT), also called “Alarm to Imaging;” and (2) time from EMS activation to thrombolysis (EMS-TPA), also called “Alarm to Treatment.” Both are time variables, measured in minutes, that have been published in numerous studies, both in Germany and the United States.Reference Ebinger, Kunz and Wendt19–Reference Lin, Calderon, Goins-Whitmore, Bansal and Zaidat30
Data Analysis
In the study cohort, performance data for MP patients were abstracted from the electronic health record and ambulance records, including the time and date of EMS activation, the time that the non-contrast head CT acquisition began, and the time that IV tPA was initiated, if applicable. The EMS-CT and EMS-TPA times were then calculated for each patient. In the MP cohort, geometric means with 95% confidence intervals (CI) were calculated to summarize EMS-CT and EMS-TPA with no outliers excluded. To compare with MSU stroke treatment performance, these measures were abstracted from data included in 12 published studies on MSUs,Reference Calderon, Kasturiarachi, Lin, Bansal and Zaidat17 including the largest current MSU study in Berlin, Germany.Reference Ebinger, Winter and Wendt18 Pooled means and 95% CIs for MSUs were estimated from the abstracted data and compared with the cohort means using non-linear regression. Statistical analysis was performed using Stata 15 (StataCorp; College Station, Texas USA). An a priori sample size calculation was not considered. For all comparisons, P values <.05 were considered statistically significant.
Results
From July 2017 through June 2019, there were a total of 423 patients with suspected stroke evaluated under MP, and 166 of these patients were either ultimately diagnosed with ischemic stroke or were initially treated as such but later diagnosed as having a stroke mimic. The EMS and treatment time data were available for 134 of these patients. A total of 61 patients (45.5%) received IV tPA. The geometric mean EMS-CT and EMS-TPA times were 41 minutes (95% CI, 39-43) and 63 minutes (95% CI, 57-70), respectively.
Pooled means for performance data from MSUs were calculated based on previously published MSU performance data including a review of 12 MSU studies.Reference Calderon, Kasturiarachi, Lin, Bansal and Zaidat17 In total, 2,086 MSU patients who received CT scans and 592 MSU patients who received IV tPA were included. A significant proportion of the currently available performance data on MSUs were from Berlin, Germany, which accounted for 1,804 of the 2,086 patients (86.4%) who received CT scans and 192 of the 592 patients (32.4%) who received IV tPA.Reference Ebinger, Winter and Wendt18 The pooled mean EMS-CT time for MSUs was 35 minutes (95% CI, 27-45) and the pooled mean EMS-TPA time was 48 minutes (95% CI, 39-60).
The MSUs achieved faster EMS-CT and EMS-TPA times (P <.0001 for each) than MP. Table 1 shows a comparison of EMS-CT and EMS-TPA times based on published MSU and MP data.
Results: EMS-CT and EMS-TPA Times for Mission Protocol and MSUs
Abbreviations: EMS-CT, EMS activation to CT; EMS-TPA, EMS activation to thrombolysis; MSU, Mobile Stroke Unit.
Discussion
Published studies indicate significant improvements in evaluation and treatment times for suspected stroke patients treated via MSUReference Calderon, Kasturiarachi, Lin, Bansal and Zaidat17 when compared to traditional, hospital-based treatment. The largest MSU study thus far showed significant treatment time improvements for patients treated on MSUs.Reference Ebinger, Winter and Wendt18 The aforementioned study was conducted in Berlin, Germany and utilized MSUs in approximately 56% of stroke dispatches.
The MSU programs include significant resource investment in terms of both equipment and personnel. Some programs, such as in Berlin, utilize a neurologist onboard the MSU, while others communicate with hospital-based physicians via telemedicine, making cost and complexity potential barriers to these approaches. At this time, paramedics in California are unable to give IV tPA in their scope of practice. Also, additional questions remain, such as how many MSUs would be required to optimally cover a specific geographic region in order to completely shift all IV tPA treatment to MSUs and obviate the need for hospital-based protocols such as the MP.
The MP, like MSU programs, seeks to expedite stroke care starting in the prehospital setting. The MP does so by utilizing prehospital activation coupled with streamlined hospital-based evaluation and treatment. For dense, urban locations with short EMS transport times and distances, such as San Francisco, it was hypothesized that treatment time metrics would be comparable between MP and MSU patients. An improvement in door-to-needle times from a median of 35 to 29 minutes following implementation of the MP had been previously demonstrated.Reference Lin, Calderon, Goins-Whitmore, Bansal and Zaidat30 However, while EMS-CT and EMS-TPA times were ultimately faster with MSUs than with MP for both metrics (P <.0001), the absolute difference in EMS-TPA times was approximately 15 minutes.
These results suggest that from purely a treatment time perspective, MSUs may provide a faster approach to stroke care than a hospital-based initiative. Although EMS-CT and EMS-TPA times were faster utilizing an MSU strategy, the mean EMS-CT times and EMS-TPA times for both MP and MSU (41 minutes and 35 minutes, respectively; 63 minutes and 48 minutes, respectively) achieved tremendous metrics. A hospital-based program involving EMS may not have achieved stroke treatment times as fast as an MSU strategy; nonetheless, the potential benefits of an MSU may be highly dependent on the local geography and density of the population, transport times, and available stroke centers. Regardless of the strategy to expedite stroke care, both of these approaches highlight that incorporating the prehospital setting in a unified stroke response is beneficial.
Limitations
This study has important limitations. Data on MSUs were based on multiple studies with unavoidable differences in study setting, including population density and catchment area. The majority of MSU data currently published were from Germany, with likely differences in EMS and hospital systems. The absence of an MSU program in San Francisco precluded a more direct comparison. Additionally, MP is confined to a single hospital with specific characteristics (an urban setting, access to 24-hour neurology consultation, and safety net status) making generalizability an issue. There are currently seven stroke centers sponsored by six health care systems in San Francisco’s 42 square mile area, and the stroke system currently does not have standardized data collection and analysis. It is unknown if other stroke centers are able to achieve faster EMS-CT and EMS-TPA time intervals based on geographic distribution of stroke patients compared with their hospital locations. Without these data, the entire city’s relevant stroke time intervals cannot be determined. With density comes traffic flow issues, and without geographic and traffic data, it would be difficult to predict positioning of one (or more) MSU’s effect on reducing these intervals. Furthermore, funding of MSUs may have impact on their performance, as many United States units are funded by individual hospital systems (as opposed to government/public health entities) and this may have effects on their geographic deployment. Since MP activations and MSU dispatches are determined by the suspicion of stroke by EMS providers, including dispatchers, the potential for bias exists with EMS providers (for MP) and dispatchers (for MSUs) when choosing which patients warranted prehospital notification and activation for stroke. San Francisco does not currently have a policy for prehospital bypass of the nearby non-endovascular therapy (EVT) stroke centers in favor of direct transport to EVT capable centers for patients with high suspicion for large vessel occlusion stroke. The ZSFG is one of the few EVT capable centers in San Francisco and this analysis does not account for the possibility of longer transportation times that could occur in a subset of patients if bypass were implemented. A planned revision of San Francisco’s stroke system in the next year should improve city-wide data collection for both prehospital and hospital treatment of stroke patients. Lastly, outcomes data were not analyzed, so the potential effects of differences in these process metrics on clinical outcomes, or whether more strokes were ultimately diagnosed as stroke mimics in one setting or another, were not evaluated.
Conclusions
In a moderate-sized, urban setting with high population density, a streamlined, step-by-step approach to emergency stroke care which includes prehospital providers was able to achieve EMS activation to treatment times that were slightly slower than published performance data from MSUs. Although there is statistical significance in the compared time segments, the clinically significant difference is unclear. Further comparisons of MSU performance on process metrics, as well as on clinical outcomes, that take into account the suitability of population density and local geography should be used to inform future stroke quality improvement initiatives.
Conflicts of interest/funding
ASK receives research grant support (significant) from the NIH/National Institute of Neurological Disorders and Stroke and NIH/National Center for Advancing Translational Sciences; receives research grant support (significant) from SanBio; received support (modest) from Neuravi as a member of the DSMB for the ARISE-2 trial. All authors have no conflicts of interest to declare.
