ACCESS: Acute Cerebrovascular Care in Emergency Stroke Systems

Objectives  Our primary objective was to determine the proportion of the population able to achieve acute cerebrovascular care in emergency stroke systems (ACCESS) in the United States. In addition, we examined how policy changes, including allowing ground ambulances to cross state lines and allowing air ambulances to transport patients from the prehospital setting to primary stroke centers (PSCs), would affect population access to stroke care.

Design  Data were obtained via the US Census Bureau, The Joint Commission, and the Atlas and Database of Air Medical Services. Driving distances, ambulance driving speeds, and prehospital times were estimated using validated models and adjusted for population density. Access was determined by summing the population that could reach a PSC within the specified time intervals.

Setting/Participants  US population.

Main Outcome Measures  Thirty-, 45-, and 60-minute access by ground and air ambulance to PSCs.

Results  Fewer than 1 in 4 Americans (22.3%) have access to a PSC within 30 minutes, less than half (43.2%) have access within 45 minutes, and just over half (55.4%) have access within 60 minutes. The use of air ambulances to deliver patients to PSCs would increase access from 22.3% to 26.0% for 30 minutes, 43.2% to 65.5% for 45 minutes, and from 55.4% to 79.3% for 60 minutes. The combination of prehospital regionalization and air ambulance transport of patients with acute stroke would reduce the 135.7 million Americans without 60-minute access to a PSC by half, to 62.9 million.

Conclusions  About half of the US population has timely access to a PSC. The use of air ambulances to triage patients with ischemic stroke to a PSC would increase the percentage of the US population with prompt access to stroke care. These data have implications for the ongoing design of the US stroke system.

Stroke is a leading cause of serious, long-term disability and the third leading cause of death in the United States.^1-3 In the United States, every 40 seconds someone experiences a stroke, and every 3 to 4 minutes someone dies of a stroke.⁴ With direct and indirect costs totaling $68.9 billion, stroke is a major public health priority.⁵

The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group demonstrated that administering intravenous (IV) recombinant tissue plasminogen activator (tPA) within 3 hours of symptom onset was associated with a 30% greater likelihood of decreased disability compared with placebo.⁶ Further trials have shown that IV recombinant tPA may be used safely up to 4.5 hours in selected patients.⁷ Stroke is thus a highly time-sensitive disease, similar to trauma or ST elevation myocardial infarction—an estimated 1.9 million neurons die each minute during which a stroke is untreated.⁸ Given the ability for recombinant tPA to reverse or ameliorate deficits, there is increased recognition among system planners and the public that “time is brain.”^9,10

Despite its clinical efficacy^6,11 and cost-effectiveness,^12,13 only 3% to 8.5% of patients with stroke receive recombinant tPA.¹⁴ Although some limitations to treatment are well known (eg, late presentation or recombinant tPA contraindications), others, including timely access to stroke care, have not been fully explored. Creating dedicated stroke teams increases recombinant tPA use and has been associated with a trend toward fewer in-hospital deaths.^15-21 Similarly, implementing “Get With the Guidelines—Stroke,” a national quality-improvement program, has increased adherence over time to 7 important performance measures.²² In addition, access to referral stroke care has been associated with lower stroke mortality rates at 7 days, 30 days, and 1 year.²³ Although these interventions may improve care for patients who arrive promptly at the hospital, they do little to improve access to stroke care from the perspective of the population.

In 2000, the Brain Attack Coalition recommended establishing primary stroke centers (PSCs)¹⁵ and described the establishment of acute stroke teams, stroke units, written protocols, and integrated emergency response systems. Accordingly, The Joint Commission (TJC) and the American Stroke Association developed the PSC Certification Program.²⁴ Hospitals may be certified as PSCs if they (1) comply with consensus-based national standards, (2) use PSC recommendations and practice guidelines, and (3) participate in performance improvement activities.²⁵

Disparities in stroke care have been described for women and minorities,^26,27 as well as by region of the country.^26-29 A 1980 examination of age-adjusted state-level stroke mortality identified 11 states with stroke death rates greater than 10% higher than the US average and dubbed these states the “Stroke Belt.”²⁹ A follow-up analysis 2 decades later again found the highest death rates in these states.¹

Despite disparities in stroke care and differences in mortality, hospital participation in the PSC program is voluntary, and TJC does not influence location of PSCs.³⁰ Although some states (Florida, Massachusetts, New York, and Texas) have independently established protocols directing patients to certified facilities, regionalization of acute stroke care has yet to be addressed in most states or nationally.^31-34

To inform decision making regarding stroke centers and resource allocation, we sought to describe population access to TJC-certified PSCs and describe the proportion of the population able to achieve acute cerebrovascular care in emergency stroke systems (ACCESS) in the United States. We calculated population access to TJC-certified PSCs and examined how emergency medical systems (EMS) policy changes could improve population access to definitive stroke care. The Institute of Medicine recently emphasized the importance of organized, coordinated, and accountable emergency care systems.^35,36 Examining access to specialty stroke care from the population perspective may inform the development of the US stroke system and the larger emergency care system.

Population information was obtained using data from the US Census Bureau and deliverable addresses from the US Postal Service (Claritus Inc, Ithaca, NY).^37,38 Our main geographic units of analysis were block groups. A block group is a geographic unit containing 600 to 3000 people that does not cross state or county boundaries. Each block group's population was assigned a point in space (a centroid) that was nearest to most residents. Population estimates and population-weighted centroids for 208 667 block groups were calculated for 2007 for the entire United States.

The PSC Certification Program was established by TJC in 2003.²⁴ TJC awards PSC certification using criteria based on guidelines from the Brain Attack Coalition and American Stroke Association.¹⁵ Certification is awarded annually with the potential for a 1-year extension. Primary stroke centers receive an on-site review every 2 years. The list of TJC-certified PSCs is publicly available.³⁰

The original National Heart, Lung, and Blood Institute's Stroke Belt list was used to identify the Stroke Belt states (Alabama, Arkansas, Georgia, Indiana, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, and Virginia).³⁹ Corresponding stroke mortality data were obtained from the Centers for Disease Control and Prevention.¹

Air ambulance data were obtained from the January 2008 Atlas and Database of Air Medical Services.⁴⁰ These data included the location, type of air ambulance, and airspeed of all helicopter pad depots operated by air medical service providers that respond to emergency scenes in the United States. Fixed-wing air medical services were excluded.

Geographic access was calculated by summing the population that could reach a PSC by ground ambulance within the specified prehospital period. This was done by calculating the distance to the closest PSC for all US residents, converting distances to ambulance transport times, and summing the population able to arrive at a PSC within 30, 45, and 60 minutes. Subgroup analyses considered only the population 65 years or older because nearly three-quarters of all strokes occur in this group.⁴¹

To estimate the distance from the population to the PSC, we used US Census Bureau block groups. The distance between the population-weighted centroid of the block group and the nearest PSC was calculated using euclidean (straight-line) distance.^42,43 Each block group was linked exclusively to the nearest PSC, and no block group was counted more than once in the summation formula for access. All programming code was written and tested using Compaq Visual Fortran Version 6.6 software (Compaq Computer Corporation, Houston, Texas), translated into C++ code (Microsoft Corporation, Redmond, Washington), and then validated by comparing output from the Fortran code with output from the C++ code.

Once euclidean distance from a population to a PSC was estimated, distance was converted into drive times adjusted for population density. Drive times were classified as urban, suburban, or rural by averaging the population densities of the block group containing the PSC and the block group of origin and then creating tertiles of average population density. We used empirically derived and previously validated ambulance drive speeds of 32.3, 76.4, and 90.8 km/h for urban, suburban, and rural driving, respectively.⁴⁴

To complete the time estimate for prehospital interval, we imputed empirically derived times for other essential prehospital intervals.⁴⁴ The time from the telephone call to a 911 center until ambulance dispatch (activation interval) was imputed as 1.4, 1.4, and 2.9 minutes for urban, suburban, and rural areas, respectively.⁴⁴ The time from ambulance dispatch until scene arrival (response interval) was imputed by multiplying the drive time from scene to hospital derived in the 2 preceding paragraphs by the constants 1.6, 1.5, and 1.4 for urban, suburban, and rural drives, respectively.⁴⁴ Finally, 13.5, 13.5, and 15.1 minutes were added to the model to account for time spent on-scene with the patient as published previously.^45,46

We summed the population able to access a PSC at 30, 45, and 60 minutes. Block group population access calculations were aggregated to compute estimates of access for all 50 states and the District of Columbia and for the United States as a whole. Given the variability in prehospital systems with respect to the ability to cross into neighboring states, we first performed our calculations assuming state lines could not be crossed and then performed a similar analysis in which we allowed the crossing of state lines.

We assumed that helicopters travel in straight-line (euclidean) distance. Thus, to calculate helicopter flying times, we multiplied helicopter depot–specific cruise speeds by the straight-line distance from the helicopter depot to the block group centroid and from the block group centroid to the closest PSC. We then added an empirically derived, previously validated constant for helicopter activation time of 3.5 minutes and for helicopter on-scene time of 21.6 minutes.⁴⁴ In our analysis examining the impact of air ambulance transport, we summed the population of block groups able to reach a PSC within the specified time increment using air or ground ambulance.

Fewer than 1 in 4 Americans (22.3%) have access to a PSC within 30 minutes, less than half (43.2%) have 45-minute access, and just over half (55.4%) have 60-minute access if ground ambulances are not permitted to cross state lines (Table 1 and Figure 1A). For Americans 65 years or older, only 23.7% have 30-minute in-state ground access, 42.6% have 45-minute access, and 53.7% have access to a PSC within 60 minutes, leaving approximately 17.9 million senior citizens without 60-minute access to a PSC (Table 2).

If EMS were to cross state lines, the population with 30-, 45-, and 60-minute access would increase only minimally to 22.6%, 44.2%, and 57.2%, respectively. This translates to 135.7 million Americans without 60-minute access to a PSC. The addition of air ambulances to existing ground ambulances increased access from 22.3% to 26.0% for 30 minutes, 43.2% to 65.5% for 45 minutes, and 55.4% to 79.3% for 60 minutes (Figure 1B and Figure 2). The combination of prehospital regionalization and air ambulance transport of patients with acute stroke would halve the Americans without 60-minute access to a PSC to 62.9 million.

Thirty-minute in-state ground access to a PSC is most available in the Midwest (26.8%), followed by the Northeast (23.0%), the South (22.0%), and the West (19.4%) (Figure 1). Forty-five minute in-state ground access follows a similar pattern, with 49.1% in the Midwest, 48.1% in the Northeast, 40.9% in the South, and 39.3% in the West. The greatest 60-minute in-state ground access is seen in the Northeast (63.7%), followed by the Midwest (61.0%), the South (52.1%), and the West (50.9%). This translates to nearly 35 million people in the West who are not within 60 minutes of a PSC. Although allowing ground ambulances to cross state lines does not considerably alter regional access, combining air ambulances allowed to cross state lines with existing ground ambulances increased 60-minute in-state ground access from 63.7% to 93.1% in the Northeast, 61.0% to 86.2% in the Midwest, 52.1% to 78.3% in the South, and 50.9% to 80.6% in the West (Table 1).

Statewide access to PSCs varies greatly. Sixty-minute in-state ground access to a PSC ranged from 0% (Delaware, New Mexico, North Dakota, Vermont, and Wyoming) to 100% (District of Columbia). When state lines are crossed by EMS, 60-minute PSC access greatly improved in Delaware (28.2%), marginally improved in New Mexico (1.4%), and remained 0% for North Dakota, Vermont, and Wyoming. Assuming that prehospital providers cannot cross state lines, in 25 states less than half of residents have 60-minute ground access to a PSC (Table 1). Although allowing prehospital ground providers to cross state lines would increase access to care in several states, 25 states still would not provide PSC access for half their population. Assuming prehospital triage to a PSC and using the existing complement of air ambulances along with ground ambulances would decrease the number of states providing less than half the population access to a PSC to 9 states (Alabama, Mississippi, Montana, New Mexico, North Dakota, Oklahoma, South Dakota, Vermont, and Wyoming).

Access to in-state PSCs (Figure 3) at respective 30-, 45-, and 60-minute in-state ambulance drive times within the Stroke Belt ranges from 1.9%, 4.5%, and 7.4% (for Mississippi) to 25.7%, 50.6%, and 65.6% (for Georgia). Despite previously described high stroke incidence and mortality,^47-49 the Stroke Belt states make up nearly half (5 of 11) of the states where less than 25% of the population has 60-minute in-state ground access to a PSC.

We present, to our knowledge, the first national estimates of access to TJC-certified PSCs. With the current ground ambulance system, only 55.4% of Americans and 53.7% of senior citizens (aged ≥65 years) have access to a PSC within 60 minutes. We estimate that the 135.7 million Americans without 60-minute ground access to a PSC could be reduced to 62.9 million with the combination of prehospital regionalization and air ambulance transport for patients with acute stroke. Our analysis demonstrates population access to stroke care for all Americans and expands on previous regional findings.⁵⁰

A large number of PSCs have been developed in highly populated areas, and significant gaps remain in underserved rural and urban regions. Recent review of TJC-certified PSCs finds that 636 PSCs have now been certified.³¹ Unfortunately, most of these new PSCs are in close proximity to existing PSCs, creating overlapping service areas and potential redundancies in care. A number of factors may influence the decision to become certified, including competition and market forces. This analysis of access to care for stroke offers promise as a policy planning tool by entering a population health perspective into the many factors influencing the development of stroke systems.

There has been increased attention to the organization of the US emergency care system since the Institute of Medicine's publication of a 3-report series.⁵¹ These reports describe fragmented care and outline the development of a system that is “coordinated, regionalized, and accountable.”^35,36,51 The US trauma system has served as a model for the rapid delivery of specialized care by designating referral centers and organizing the rapid delivery of patients to the appropriate level of care through prehospital regionalization or transfer agreements. Trauma centers were initially developed at tertiary-care facilities, but over time a more comprehensive trauma system was developed using principles of operations research and a population perspective.^52,53 Approximately 83% of the population has access to trauma care within an hour,⁵⁴ and definitive trauma care has been demonstrated to reduce the risk of death by 25%.⁵⁵ Inclusive systems of trauma care in which all facilities have a role in a coordinated system have been demonstrated to improve outcomes compared with exclusive systems in which only referral centers are designated.⁵⁶ As described herein, substantial data suggest that organized stroke care improves outcomes,^16-22 but these findings are not uniform⁵⁷ and most efforts have examined outcomes only from the stroke center perspective. Future efforts should be from the system perspective and elucidate which components of stroke care are associated with improvement. If organized stroke systems of care are demonstrated to improve long-term outcomes, many principles behind the development of the trauma system should be used to improve stroke care.

Stroke care mirrors trauma care because the entire population is at risk and access to resources can improve outcomes.⁵⁸ However, stroke typically does not require complicated surgical intervention but rather requires prompt expert examination and decision making—which may not require transportation to a stroke center. Solutions such as changing EMS policies to permit ground ambulances to cross state lines, bypass facilities not providing stroke care, or allow for air ambulance transport from the field may aid in developing a regionalized stroke system, but other solutions may be equally viable and may be preferable given the cost and safety concerns of aeromedical transport.^59,60 In all likelihood, a comprehensive stroke system will be created from a patchwork of the many candidate solutions. These will include low-tech solutions such as developing explicit interhospital referral networks, using specially trained physician extenders in underserved areas,⁶¹ and incentivizing the development of PSCs in underserved regions. High-tech solutions, including the use of computer models to determine the optimal location for future stroke centers and building telemedicine networks to extend stroke expertise,^62-65 are also likely to optimize future stroke systems.

Our study has several limitations. First, our analysis relies on travel times empirically derived from an analysis of trauma patients. While these times likely serve as reasonable proxies of drive times for stroke care, drive times for trauma care are on average 6 to 11 minutes shorter than transport times for patients with stroke.^66,67 Moreover, trauma triage guidelines have been evolving for decades, and prehospital regionalization of patients with stroke using aeromedical transport is likely to be less efficient. Our assumptions based on the presumed shorter transport times may overestimate population access to PSCs. Second, access estimates are based on where people live, not where they are when stroke symptoms occur; however, the Framingham Study found that most strokes occur at home.⁶⁸ Third, because TJC certification of PSCs is a dynamic process, our inventory of PSCs captures only those centers certified as of November 3, 2008, which may underestimate national access.³⁰ Fourth, our study calculates access to TJC-certified PSCs; it does not include state-certified stroke centers or hospitals participating in national quality-improvement programs.⁵⁰ Fifth, a recent study mapping the locations of all hospitals that treat stroke with IV recombinant tPA at or above the national average suggests that PSCs are not the only centers administering recombinant tPA for stroke.⁶⁹ Our analysis does not attempt to account for treatment strategies such as interhospital transfer after IV administration of recombinant tPA (“drip and ship”)⁷⁰ or telemedicine “hub-and-spoke” agreements,^62-65 and national quality-improvement programs have demonstrated marked improvements in stroke care.²² Sixth, our methods do not account for geographic boundaries such as rivers, mountains, weather, or traffic. Finally, our assumption that EMS policy with respect to transporting patients across state lines is uniform across the United States is clearly not true because many areas permit patients to cross state lines to obtain specialty care including stroke care. We use this assumption to illustrate the importance of planning from the population perspective. Despite these limitations, this report offers an important perspective on how a transparent process of universal verification and systems planning could inform the coordination and optimization of the stroke system.

In conclusion, we present, to our knowledge, the first study to report state, regional, and national access to PSCs. With the current system of EMS ground transport, just over half of Americans (55.4%) and senior citizens (53.7%) have access to a PSC within 60 minutes. Prehospital regionalization of patients with acute stroke and use of existing air ambulances would reduce the number of Americans without 60-minute access to a PSC by half. Given the time-sensitive nature of interventions for ischemic stroke, future efforts to design stroke systems should consider the population perspective and should be integrated into the ongoing development of the US emergency care system as a whole.

Correspondence: Brendan G. Carr, MD, MA, MS, University of Pennsylvania, 929 Blockley Hall, 423 Guardian Dr, Philadelphia, PA 19104 (Carrb@upenn.edu).

Accepted for Publication: April 2, 2010.

Author Contributions:Study concept and design: Albright, Branas, Meyer, Matherne-Meyer, Zivin, Lyden, and Carr. Acquisition of data: Albright, Branas, and Carr. Analysis and interpretation of data: Albright, Branas, and Carr. Drafting of the manuscript: Albright, Branas, Zivin, Lyden, and Carr. Critical revision of the manuscript for important intellectual content: Branas, Meyer, Matherne-Meyer, Zivin, Lyden, and Carr. Statistical analysis: Branas and Carr. Obtained funding: Albright. Administrative, technical, and material support: Albright, Zivin, Lyden, and Carr. Study supervision: Branas, Meyer, and Lyden.

Financial Disclosure: None reported.

Funding/Support: This research was supported in part by a generous gift from Michelle Peck, MPH, FNP. Dr Carr is supported by a career development award from the Agency for Healthcare Research and Quality (K08HS017960).

Disclaimer: This article is solely the responsibility of the authors and does not necessarily represent the official views of the Center for Transportation Injury Research, the Association of Air Medical Services, the US Department of Transportation, or the US Department of Health and Human Services.

Additional Contributions: Justin Williams, PhD, Tara D. Jackson, PhD, Karl A. Dailey, BS, Krista Heinlen, BS, and Vicky Tam, MA, provided programming and mapping support.