Objectives

The purpose of this report is to systematically review the available literature in the field of acute stroke evaluation and treatment, in order to provide organized evidence relating to a number of objectives put forth by the AHRQ. The findings of the report are intended to assist the American Association of Health Plans (AAHP) and its member health plans in their development and translation of the research findings into practical information for healthcare providers and consumers, where applicable. The University of Ottawa Evidence-based Practice Center (UO-EPC) task involves the following three objectives:

(1)

To determine what interventions for acute stroke (delivered within the first 24 hours from onset of symptoms) are effective in reducing stroke-related morbidity or mortality. (Note: some studies that initiated interventions beyond 24 hours were also included if they were deemed relevant to the topic area being studied). As this question was very broad it was refined into 12 key questions.

(2)

To review how the safety and effectiveness of these interventions vary with the timing of intervention in relation to onset of symptoms.

(3)

To determine what is the evidence that specific systems of care (i.e., dedicated stroke programs) improve outcomes of acute stroke.

Background

Stroke has been defined as “rapidly developing clinical signs of focal or global disturbance of cerebral function with symptoms lasting 24 hours or longer or leading to death with no apparent cause other than of vascular origin.”¹ Mechanistically, stroke may be organized into ischemic and hemorrhagic forms. In ischemic stroke the primary pathology is one of occlusion of arteries carrying blood to the brain. Hemorrhagic stroke includes intracerebral hemorrhage (ICH), which is the consequence of vessel rupture and cerebral damage due to pressure and toxic effects of blood within the closed space of the skull, and sub-arachnoid hemorrhage (SAH), which is bleeding into the space outside the brain but within the subarachnoid membrane. While SAH is often included in the definition of stroke for the purposes of surveillance and descriptive epidemiology, its presentation and management are sufficiently unique that it will not be considered for this analysis.²

The incidence of stroke is strongly age-linked.³ The increasing median age of national populations occurring worldwide has led the World Health Organization (WHO) to target this disease for surveillance with the goal of reducing associated morbidity and mortality.⁴ In developed countries such as the U.S., the burden of disease is expected to be higher due to the age structure of these populations and has led to the call for the development of organized systems of stroke care, with particular emphasis on thrombolytic therapy.⁵ Public education, which teaches the symptoms of acute stroke and emphasizes the need to seek urgent care, has been seen as a means of increasing acute stroke treatment rates and, therefore, improving outcomes.⁶

Stroke Epidemiology

Cerebrovascular disease is the third leading cause of death on a global basis, with rates for males ranging from 340.3 cases per 100,000 population in the Russian Federation to 58.7 cases per 100,000 in the U.S.⁷ An estimated 700,000 strokes occurred in the U.S. in 2002, with approximately 500,000 cases being first events,^{7, 8} with mortality rates ranging from 39 cases per 100,000 in New York State to 80.8 per 100,000 in South Carolina. The global burden of mortality was estimated at 4.7 million in 1995.⁹

Mortality rates due to stroke have declined for a number of populations in the twentieth century.^10–13 In the U.S., the rate of decline was approximately 0.5% per year between 1900 and 1920, and approximately 1.5% per year from 1950 to1970.^{14, 15} All age groups were affected, suggesting that this was a period effect rather than a cohort effect. Interestingly, this decrease began long before effective therapies for stroke prevention treatment were available, and thus cannot be ascribed to medical care. This trend has not continued in the latter part of the century. In Rochester Minnesota, the incidence of stroke stabilized at approximately 150 per 100,000 population and remained so from the mid-1970's to the mid-1990's.^{16, 17} While the initial period of this stabilization coincided with the introduction of CT scanners, and thus, may be explained by an increased probability of diagnosis, such an effect is less likely to explain a persistent stabilization through the next two decades.

While most states have shown a decrease in the mortality rate due to stroke between 1990 and 2000, Alaska, Maryland and Oregon have shown increases over this time period. Between 1979 and 2001, stroke discharges from short stay hospitals in the U.S. increased by 25%.⁷

Stroke incidence in males is 1.25 times greater than that for women. Age-adjusted stroke incidence rates per 100,000 population are 167 for White males, 138 for White females, 323 for Black males, and 260 for Black females.¹⁸ The risk for Black Americans is greater than that for White Americans with a risk ratio of 5 for the 35 to 44 year age group in the greater Cincinnati/northern Kentucky study.^{19, 20} The risk ratio is greater than 1 for all age groups, although it approaches unity in the over 85 age group. It is unclear if this risk ratio reflects biologic or social factors or some combination thereof. In spite of the increased risks, fewer blacks in the U.S. are candidates for thrombolytic therapy.²¹

Hispanic Americans may have an age-specific stroke incidence that differs from non-Hispanic whites. Stroke surveillance in this population is challenging and the databases are limited. The National Longitudinal Mortality Study (NLMS) suggests that for the 45 to 59 year age strata, the relative risk for Hispanic men and women is 1.0 and 1.17, respectively. For the 60 to 74 year strata, it is 0.53 to 0.76.²² These figures should be interpreted with caution due to small numbers. It is unclear if the reduced incidence in the older age strata reflects a true effect or incomplete ascertainment.

A link between incidence, outcome, and socioeconomic status has also been documented in the U.K., where unskilled manual workers have a 60% higher risk of stroke than professionals, along with a 50% higher age-adjusted mortality rate.²³

While age-specific death rates for stroke have fallen over the last century,^24–26 the number of such deaths is currently, and is expected to remain, greater for women than men.⁹ The projected increase in absolute numbers in the U.S. is from approximately 700,000 in 2002 to 1,136,000 in 2025.⁸

Both the pending increase in stroke mortality and the gender differences may be explained by the strong relationship between stroke and age. It has been over 30 years since the logarithmic relationship between stroke mortality and age was described by Kurtzke.^{27, 28} Consequently, it is expected that the burden of disease will increase in absolute terms due to the increase in population median age and the alteration of the population age structure, despite stable or falling age-adjusted incidence rates.

The risk factors for stroke overlap significantly with those for ischemic heart disease. After age, the most important risk factor is blood pressure. The risk of stroke increases across the measured pressures for both systolic and diastolic pressure.²⁹ For each 10 mm Hg increase in systolic blood pressure or 5 mm Hg increase in diastolic pressure, the relative risk of stroke increases by a factor of 2.3.³⁰ Anti-hypertensive treatments result in significant reductions of risk for first and subsequent stroke.^{31–33, 33, 33, 33, 34} Diabetes and smoking each increase the relative risk by a factor of 2.^35–39 Hyper-homocysteinemia is an emerging risk factor with a relative risk of 5 to 7 between the highest and lowest quartiles of serum homocysteine concentration.⁴⁰ Atrial fibrillation likewise carries a relative risk of 5 for stroke. The link to serum cholesterol is somewhat more complex for stroke than for coronary artery disease. Observational studies do not show an increased risk with elevated cholesterol levels.⁴¹ Low cholesterol may result in an increased risk of hemorrhagic stroke.⁴¹ In spite of this, a major therapeutic trial of lipid lowering therapy has demonstrated a reduction in ischemic stroke incidence without an increase in hemorrhagic stroke.⁴² However, in a sub-group within this trial, patients with prior ischemic stroke randomized to statin therapy did not show a reduced incidence of stroke; the benefit among this cohort was in reducing subsequent coronary artery disease.(HPS analysis, LANCET)

Stroke Costs

Stroke currently consumes significant resources through healthcare costs and disability. Twenty-eight percent of total stroke incidence occurs in individuals under the age of 65,⁴³ and accounts for 20% of all acute care beds, and 25% of all chronic care beds.⁴⁴ The acute cost per stroke in Ontario was estimated at C$27,500 in 1996. The absolute number of hospitalizations for stroke has been increasing for the past 20 years with a projected increase in hospitalizations of 10% to 15% between 1996 and 2016.⁹ Hospitalization makes up 87% of the total direct cost of stroke care, which was estimated by the Heart and Stroke Foundation of Canada to be 2.8 billion dollars in 1996.⁹ However, this cost does not include costs related to either short- or long-term disability. Such costs may be considerable since, in the case of ischemic stroke, only 25% of people make a full recovery.⁴⁵ More recent costs are available for the U.S., where the estimated direct cost for stroke in 2004 is 33 billion dollars—41% of this is due to hospital costs reflecting the expense of acute care. Indirect costs due to loss of productivity are estimated at 53.6 billion dollars.⁷

Given the trends observed in stroke over the last three decades and the associated costs, stroke is, and will remain for the foreseeable future, a significant problem for North American and other societies.

Integration of Thrombolytic Therapy into Current Practice

Thrombolysis for acute stroke has received widespread, though not universal, support. It is a complex intervention making intensive use of resources and personnel, with a narrow therapeutic window. Current clinical protocols limit use to a 3-hour window from symptom onset. The Brain Attack Coalition (BAC) in the U.S. and the Heart and Stroke Foundation of Canada(HSFC) advocate for multilevel system changes to increase the number of patients eligible for, and receiving, thrombolytic therapy.^{46, 47} Such advocacy has influenced public policy.

The province of Ontario, Canada is in the process of a major system change in stroke care. This process is unique among large jurisdictions and forms an important framework for initiating and evaluating such strategies. In May 1997, the HSFO of Ontario, a nonprofit group, proposed the creation of a coordinated system of stroke care for the province of Ontario. In 1998, a pilot program for regional coordination of care was launched at four sites. This was shared by the HSFO, and a joint stroke strategy-working group was established in partnership with the Ontario Ministry of Health.⁴⁸ In 2000, based on the results of the initial experience, the Ministry of Health funded a coordinated province-wide stroke strategy. Over the next 3 years, a total of nine regional stroke centers were designated. The development of integrated acute stroke care was seen as a key role of these centers.

The process of delivering acute stroke therapy involves a pre-hospital phase and an ED phase. The latter requires a rapid, intense, and at times, parallel application of clinical, radiological, and biochemical analysis of the potential candidate. These evaluations serve to interpret compliance with eligibility criteria, thereby maximizing the probability of outcomes that parallel results published by the National Institutes of Neurological Disorders and Stroke (NINDS).⁴⁹ The main determinant of eligibility is time since onset.^{50 51} Further, within the eligible group, regression analysis suggests that earlier treatment is associated with better outcome.⁵¹ Attention has, therefore, been focused on the pre-hospital phase. Shortening delay time in this phase requires recognition of symptoms, a decision to seek care, and transportation to a facility capable of delivering care. Multiple mass media strategies were evaluated by the HSFO.⁵² A positive effect was noted in the ability to name two or more warning signs of stroke after the mass media campaign and, therefore, a subsequent television campaign was launched in Ontario in October 2003. It is not clear, however, that such an education strategy increases the number of potentially treatable patients or that it improves outcomes. Furthermore, such an intervention runs a risk of increased number of non-stroke patients presenting to the ED. Evaluation and treatment of such patients might be expected to increase resource utilization and possibly worsen outcomes by exposing individuals to the risk of treatment who do not have any possibility of benefit.

Patient Selection for Thrombolysis

Current treatment protocols for thrombolysis rely on the entry criteria used for the NINDS Trial.⁴⁹ The NINDS trial published in 1995 was the pivotal evidence leading to regulatory approval of tPA in North America.⁴⁹ Clinical protocols currently used for thrombolysis use trial criteria regarding patient selection and blood pressure management. This trial has received much attention and has been reanalyzed independently on at least two separate occasions.^{53, 54}

NINDS criteria require a diagnosis of acute ischemic stroke of less than 3 hours' duration, absence of hemorrhage on CT scan, good blood pressure control, a NIH Stroke Scale Score of greater than 4 along with exclusion of those at high risk for bleeding with therapy. Reanalysis of the results from the original trial suggest a better outcome if treated within 0 to 90 minutes compared to 91 to 180 minutes.⁵⁵ The stringent time constraints decrease the absolute number and percentage of individuals receiving treatment. This review will investigate the data regarding the relationship between onset to treatment time and outcome.

The Cerebral Ischemic Penumbra denotes that a portion of the brain which has been rendered inoperative by ischemia but has not yet died.⁵⁶ Time since onset forms a surrogate for tissue viability.⁵⁷ Imaging of acute stroke holds the possibility of identifying the ischemic penumbra on a physiologic basis and thus increasing safety, efficacy, and through an extension of the time window applicability of revascularization therapy. We will review the evidence that pretreatment imaging improves stroke outcomes.

Community Education Programs for Stroke

The rates of treatment of acute stroke with the thrombolytic, tPA, are low—2% in the U.S.⁵⁸ and 1.4% in Canada.⁵⁹ The narrow eligibility criteria, the lack of comfort with acute neurology by many physicians and the perception of a high risk of adverse events contribute to the low treatment rates.⁵⁰ The most significant limiting criterion is time since onset of symptoms. Englestein and colleagues retrospectively examined the records of patients admitted to a New York hospital with a diagnosis of stroke, for the presence of exclusion criteria for tPA.⁶⁰ Of 201 patients identified by ICD-9 codes, 94% were excluded based solely on time of presentation to the ED.⁶⁰ Reports from other centers have documented rates of exclusion by delay time criteria of 44%,⁶¹ with speculation that differences arise due to variations in public awareness of symptoms in different communities.⁶⁰ In Calgary, 1,168 patients with ischemic stroke were prospectively identified and evaluated for reasons for exclusion for tPA therapy. Of these, 73.1% presented beyond 3 hours after symptom onset and thus could not be considered for treatment.⁵¹ This was the most common reason for exclusion from treatment. An education effort leading to better awareness of symptoms may, therefore, improve treatment rates.

The Chain of Recovery Writing Group has identified a sequence of events that must take place in order to access time-dependent therapies in emergency situations.⁶² By analogy with processes successfully deployed for acute myocardial infarction and trauma, the chain consists of: identification of symptoms by patient or bystander; activation of the EMS; alerting treating center; and, diagnosis and treatment.

Initiation of this entire sequence of events is contingent upon the correct identification of symptoms along with the appreciation of their gravity. In contrast to major trauma, the seriousness of stroke symptoms may not be obvious.⁶²

Gaps have been identified in the public knowledge of stroke symptoms. A population-based random-digit telephone survey was conducted in the Cincinnati region.⁶³ While 70% of 2,173 respondents correctly identified at least one symptom of stroke, groups at highest risk of stroke, including people over 75 years of age, men, and blacks, were the least knowledgeable. In a national U.S. phone survey of 750 adults over age 50, 42% could not identify limb numbness or weakness as stroke symptoms.⁶ Forty percent of respondents were unaware that stroke occurs in the brain. Of patients with stroke, 39% were unable to identify any symptoms of stroke. This proportion was worst for those over the age of 65 than for those under 65 (47% vs. 28%, p=0.016).

Williams interviewed consecutive admitted stroke patients within 72 hours of stroke onset regarding knowledge of, and attitude to, stroke symptoms.⁶⁴ While 38% of 67 individuals purported to know stroke symptoms, only 25% correctly interpreted their symptoms as being due to stroke. Eighty-six per cent of those arriving after 3 hours felt that the symptoms were not serious. Interestingly, patients with prior stroke (46% vs. 16% p=0.03) were more likely to ascribe their symptoms to stroke but were no likelier to seek early attention (19% vs. 39% p=0.35). Ambulance transport was independently associated with early arrival (OR 5.55, 95% CI 1.37 to 22.6). Instructions to use an ambulance in the setting of acute stroke may, therefore, result in earlier intervention and improved outcomes.

Wein reported on the individual activating the EMS in 429 validated admitted stroke patients, as part of the TLL Temple Foundation Stroke Project.⁶⁵ Of these, 38% (163) of patients arrived by ambulance. In these cases, the person activating the system was: self, 4.3%; family member, 60.1%; paid caregiver, 18.4%; and, coworker or other, 12.9%. It was concluded that educational efforts directed exclusively at patients themselves were likely to be of low yield and wider educational efforts were required.

Consequently, a number of calls have been made for public education to increase knowledge of stroke symptoms.^{6, 66–68} Such community interventions are expected to decrease delay time and consequently improve the thrombolysis treatment rates. However, several concerns about this approach remain. First, the experience in acute myocardial infarction raises concern about the effectiveness of media campaigns for public education in a similar disease state.⁶⁹ Second, knowledge may not translate into action. Patients involved in the Asymptomatic Carotid Atherosclerosis Study⁷⁰ received targeted education on the warning signs of stroke. In spite of these efforts in a group motivated enough to participate in the study, only 40% of all first events were reported within 3 days of occurrence.⁷⁰ Participants in an advertised stroke-screening program were assessed before and after completion of a detailed evaluation of individual risk factors and counseling on risk factors and symptoms.⁷¹ After 3 months, 77% of participants could name warning signs, compared with 59% prior to the intervention; however, 73% reported no change in lifestyle in spite of an individualized written plan of action provided at the screening. Finally, the symptoms transmitted to the public in education campaigns are non-specific.⁵² Proper identification and immediate action by the public may, therefore, result in an increase in the number of non-stroke patients reporting to the ED and thus have no impact on outcomes.

At this stage, while community education programs appear to be attractive, it is clear that further review of their effectiveness is warranted.

Acute Stroke Centers

Community education programs do not occur in isolation. It is clear that the activities carried out at the receiving ED may have a significant impact on treatment rates and subsequent outcomes. The initial experience in the Cincinnati program suggested rates of hemorrhage with thrombolysis as high as 16% associated with an almost 50% rate of violation of protocols.⁷² This experience was followed by a Stroke Quality Improvement Program in nine hospitals and a subsequent report from the same group suggested the effectiveness of this intervention, reporting a protocol deviation rate of 19% with symptomatic intracranial hemorrhage rate of 6.4%, a subsequent reduction over the previously reported adverse event rate.⁷³ The Brain Attack Coalition has suggested criteria for the designation of stroke centers.⁵ Key elements of stroke centers include acute stroke teams, stroke units, written care protocols and an integrated emergency response system. Support services felt to be important include the availability and interpretation of CT scans and rapid laboratory testing. The criteria for such centers were developed on the basis of a literature review of English language articles published between 1966 and 2000, with recommendations issued after review by an expert panel. Establishment of stroke centers of this type has been promoted within North America. An estimate of their effectiveness would be helpful in terms of planning and justification of resource allocation in the wider context of healthcare. Thus, it is our objective to systematically review literature pertaining to the use of designated stroke centers to determine their effectiveness in the for acute stroke treatment.

Summary

Stroke is common, lethal, debilitating, and costly. Treatment in the acute phase is effective, although available to only a small number of individuals with stroke, resource intensive, and potentially hazardous. Increasing the probability of treatment is beneficial on the individual level and is expected to also be beneficial at the societal level. The number of potential candidates accessing acute stroke treatment is significantly limited by the narrow time window between the onset of symptoms and the initiation of treatment. It is widely believed that strategies to increase public awareness of stroke and EMS diagnosis are likely to increase the number and proportion of patients who are able to receive effective but time-dependent interventions. Prior to advocating wide implementation of pre-hospital programs in a complex system, it is important to evaluate the likelihood of success of these strategies within plausible variations of the effects of these strategies and the risks and benefits of thrombolytic treatment of acute stroke.