Management Patterns and Costs of Acute Ischemic Stroke
An International Study
Background and Purpose—With the ever-increasing pressure on healthcare budgets, we witness a growing demand for evidence of the economic implications of care across many therapeutic areas. Stroke is no exception.
Methods—Detailed information on healthcare use was collected in conjunction with two 12-week international trials designed primarily to assess the safety and efficacy of a new potential neuroprotective agent. The information was gathered prospectively by means of a customized resource use instrument that included both acute and long-term inpatient management as well as community care. In this report, the results pertaining to the 1341 acute ischemic stroke patients are described.
Results—More than 70% of the mean cost ($13 668) was explained by the initial hospitalization, which averaged 24 days. The total cost and its components varied according to patient age, the presence of comorbidities, and several indicators of disease severity. Pronounced country differences could be observed in the management of this fairly homogeneous patient group.
Conclusions—This study provides a comprehensive picture of the healthcare services used for the treatment and rehabilitation of stroke victims, presented with respect to various patient and disease characteristics. It is expected that researchers evaluating the cost-efficiency of specific stroke treatments will benefit from the detailed information presented in this report.
As a result of the healthcare cost escalation spiral, there has been a substantial increase in formal assessments of the economic implications of care in most industrialized countries.1 New pharmacological interventions for the treatment of acute stroke are no exception.2 Information on healthcare utilization and associated costs collected within clinical trials not only contributes to understanding the short-term cost implications of the treatment under evaluation but is also frequently used as the starting-point for modeling the lifetime implications of stroke and its treatment (eg, see References 3 through 53 4 5 ). To this end, detailed information on healthcare utilization was collected in conjunction with 2 large, placebo-controlled studies designed to evaluate a new agent for the acute treatment of ischemic stroke. Although development of the drug was discontinued, these trials provided a unique opportunity to examine the healthcare use and costs in the management of acute stroke in 13 countries. This report focuses solely on univariate analyses. Multivariate regression analyses to determine the predictors of cost and patient outcomes will be the subject of a separate report.
Detailed descriptions of the study procedures and clinical outcomes have been published previously.6 7 In brief, the 2 trials enrolled 1446 stroke patients in 13 countries: Australia, Austria, Belgium, Denmark, Finland, France, Germany, the Netherlands, Norway, Sweden and the United Kingdom in 1 trial and Canada and the United States in the other. The trial population consisted of patients at least 18 years old who presented with substantial neurological deficit within 6 hours after stroke onset. Patients were randomized to receive a continuous infusion of either active treatment or placebo for 5 days or until complete neurological recovery was attained. They were followed-up for a period of 12 weeks after the start of treatment. The main clinical assessments took place before randomization, at the end of the treatment, and after 4 and 12 weeks. Neurological recovery was evaluated on the basis of the European Stroke Scale (ESS) in the Australian-European study and the National Institutes of Health Stroke Scale (NIHSS) in the North American study. Functional recovery was measured by using the Barthel Index and Rankin scale in both studies.
Information on both inpatient and community healthcare resource utilization was collected during the trial period. Inpatient management included the time spent in both acute-care (ie, hospital and rehabilitation center) and long-term-care (ie, nursing home) facilities. The initial hospital stay was subdivided further by type of ward: intensive care, general ward, stroke unit, rehabilitation ward, and surgical ward. The outpatient care included the various types of rehabilitation therapy and other healthcare services that the patients received while they stayed in a residential location, ie, in a retirement home or at home. The specific services included were emergency room visits, doctor visits, neurologist consultations, mental health visits, occupational therapy, physical therapy, speech therapy, social worker visits, home nursing, and home aids. Where applicable, services were further divided into 2 care settings: services provided at the patient’s place of residence and those provided at an office. Finally, modifications of the patient’s home (eg, entrance ramp, handrails, shower) and equipment purchases (eg, wheelchair, walker, leg brace) made to accommodate the disability caused by the stroke were also documented.
The information was collected by means of a customized resource use instrument that consisted of 2 parts: a case record form to be completed by the study personnel during scheduled study visits (baseline, discharge from the hospital, and at weeks 4 and 12) and a patient checklist to be completed by the patient or caregiver on a weekly basis after discharge from the hospital. The case record form documented the patient’s stay in various treatment and residential locations over time. A major concern in multinational studies of this sort is lack of clarity regarding terminology. Namely, similar locations may be termed differently in different countries and vice versa. Moreover, organizational structures and division of responsibilities may vary by geographic region. To minimize this problem, the location categories were defined in fairly broad terms, and a country-specific description of the type of location that would belong in each category was provided for every country. Details of the hospitalizations were validated against the hospital records. The case record form also contained questions related to home modifications and equipment purchases, which were completed on the basis of a brief interview with the patient or caregiver at the end of the study. The patient checklist asked the patients on a weekly basis to record the number and types of outpatient healthcare services that they received while in a retirement home or at home.
Participation in the resource use portion of the studies was voluntary, and patients were assured in writing that refusal to have this information recorded would not affect their care in any way. Also, at any point in time, patients could withdraw from the resource use portion of the study while continuing in the main study. The studies adhered to the International Conference on Harmonisation (ICH)/WHO Good Clinical Practices guidelines.
Unit of Comparison
For analyses that compared overall resource use across various subgroups, a means of aggregating resource use was required. This aggregation was done by using a single set of unit costs from 1 country. This strategy is not intended to imply that these unit costs are applicable across all countries but rather to incorporate a measure of the relative value of the various inpatient and outpatient resources. The unit costs from the United Kingdom (Table 1⇓) were used for this purpose, as these were thought to be the most well founded because there was a high degree of agreement between the unit costs reported in 2 independently conducted surveys. Costs were obtained from publications of the Personal Social Services Research Unit at the University of Kent8 and the Scottish National Health Service9 and are reported in 1996 values. The 1996 UK general inflation rate of 4.5% was used to inflate some 1995 Scottish healthcare unit costs to 1996 values. Costs were also expressed in terms of the equivalent number of days spent in a hospital’s general ward. This procedure allows estimation of the approximate equivalent cost in another country by multiplying the number of days by that country’s per diem cost. This method, which has been applied previously,10 builds on the premise that the local per diem cost provides a reasonable indication of the country-specific cost level. For the base noncomparative analysis, the aggregate results are also reported by using the unit costs from 3 other countries: France, Germany, and Sweden.
To facilitate interpretation of the findings by an international audience, results are reported in US dollars, using an exchange rate–based conversion. Because the point of view taken in this study is that of the healthcare system, costs borne solely by the patient (out-of-pocket expenses) that were not reimbursed by a government insurance plan are not included in the calculations.
Results are presented by patient characteristics and by several disease severity indicators. The latter have been based on well-known, widely used scales and classification systems, which should enhance translation of the findings to clinicians’ and researchers’ own experiences. Five severity levels were defined on the basis of the ESS and NIHSS: very mild (NIHSS 0 to 9, ESS 60 to 100), mild (NIHSS 10 to 12, ESS 46 to 59), moderate (NIHSS 13 to 15, ESS 38 to 45), severe (NIHSS 16 to 19, ESS 28 to 37), and very severe (NIHSS 20 to 34, ESS 0 to 27). The strokes were classified as small-vessel occlusive, large-vessel occlusive, and cardioembolic, based on the TOAST (Trial of Org 10172 in Acute Stroke Treatment) criteria.11 Patients were considered to have a major functional impairment when they scored ≤70 on the Barthel Index and a minor impairment otherwise.
Institutional length of stay (LOS) was defined as the amount of time spent in an acute-care hospital (including the initial hospitalization as well as readmissions), an inpatient rehabilitation facility, or a nursing home. It was decided to focus on institutional LOS, as opposed to initial hospital LOS, to minimize the potential influence that differences in the organization of care across centers and countries might have on the results. For example, some acute-care hospitals have an associated rehabilitation center attached, whereas in other cases, a patient would have to be physically moved, and thus discharged, to a different facility to obtain the same level of rehabilitation. Analyses were conducted for institutional LOS both including and excluding time spent in nursing homes. Because the use patterns were very similar, only 1 analysis is presented here, ie, institutional LOS including nursing home stays.
As is frequently the case with cost data, the distributions were very positively skewed.12 The costs were therefore logarithmically transformed to normalize the distributions for statistical testing. The significance of observed differences was tested by ANOVA. The mean costs reported are the actual, untransformed costs. The χ2 test was used to test for differences in proportions.
Although the resource use portion of the trial was optional, the large majority of patients opted for participation. Indeed, economic information was missing for only 34 patients. Another 71 patients did not have an ischemic stroke, thus reducing the study population included in the analyses to 1341 patients. The United States accounted for ≈44% of the patients, and each of the other countries for <10%.
The characteristics of the study population are summarized in Table 2⇓. The average age of the patients was 70.5 years, ranging from 19 to 96. Males accounted for 55% of the study population. Approximately 63% of the patients were married or lived with a partner. Based on the ESS and NIHSS, the initial severity distribution was evenly balanced in the total population, ie, ≈20% of patients in each of the 5 severity categories considered. Twenty-one percent of the patients died during the course of the 12-week follow-up period, most (79%) during the initial hospital admission. The median survival of these patients was 11 days (interquartile range, 5 to 29 days).
Stroke Management Cost and Inpatient Care
The total management cost per patient over the 12-week period ranged from $298 to $110 032, with a mean of $13 668 (interquartile range, $6047 to $19 264) when UK unit costs were used. This mean cost is equivalent to that of 46 general hospital ward days. The average cost per surviving day was $257, or 0.86 per diem. Although the shape of the cost distribution was similar (ie, skewed to the right), a clear shift could be observed when comparing the distributions for North America and Europe/Australia, resulting in a statistically significant difference of ≈$2000 (P<0.0001) in the average cost ($12 673 versus $14 748). Initial hospitalization was found to be the driving force behind the total stroke costs, accounting for almost 71% of the cost. Although some variation in the mean total cost could be observed when applying the unit costs from other countries, the proportion of this cost represented by the various location categories seemed fairly consistent (Table 3⇓).
The box-and-whisker plots in Figure 1⇓ reveal the skewness of the cost and LOS distributions for each of the subgroups evaluated. Although the per-patient cost and LOS increased with stroke severity, this trend was partly concealed, and even reversed for the most severe group, by the high mortality rate among the more severe patients. Indeed, when considering the cost per surviving day, for the very severe patients it was almost triple that for the mildest group (see also Table 4⇓). In the same manner, patients with a major disability 12 weeks after the stroke had a higher median cost ($18 529) than did those with a minor disability ($10 500) or than those who died during the follow-up period ($5144). The median institutional LOS for surviving patients ranged from 19 days for very mild to 84 days for severe and very severe strokes. Survivors with a major stroke stayed ≈3 times longer in an institutional setting than did their counterparts with a minor stroke.
Table 4⇑ details the 12-week cost and the cost per surviving day by various demographic and disease characteristics. The cost per surviving day did not appear to be influenced by sex (P=0.538) or prestroke living situation (P=0.823). The daily cost increased with age (P<0.0001), but for the total cost this effect was again largely counterbalanced by the higher mortality in the older age groups (P=0.291). Stroke subtype was a strong determinant of cost, with cardioembolism and large-vessel occlusion having a much higher cost than small-vessel occlusion (P<0.0001). This finding confirms the well-known differential prognosis following the various stroke subtypes13 14 from a cost viewpoint.
Comorbid neurological or medical diseases are known to affect recovery from a stroke or to contraindicate specific treatments or rehabilitation. The most common comorbidities in this population were hypertension (56% of patients), ischemic heart disease (35%), atrial fibrillation (24%), and diabetes (20%). Nineteen percent of the patients had suffered a previous myocardial infarction. Not surprisingly, the daily stroke management cost was ≈15% higher for patients who presented with comorbidities at the time of stroke onset (P=0.0007). Refinement of this dichotomous division (the presence or absence) to account for the number of comorbidities observed in an individual patient did not reveal a further relationship with the total or daily cost.
The relation of 12-week cost to the amount of improvement in patient function observed over time was also examined. Improvement was defined as a change in Barthel Index to >70. Patients who had a Barthel Index >70 shortly after stroke onset were not included in this analysis. Patients who improved had a considerably lower cost ($14 219) than did those without improvement ($19 719) (P<0.0001).
Figure 2⇓ displays the amount of time patients spent in the various treatment and residential locations during the trial period. Of the average 71 survival days, patients spent, on average, ≈37% of the time in hospital, 17% in an inpatient rehabilitation facility, 9% in a nursing home, and 37% in a residential location, either at home or in a retirement home. Given the relatively high cost of a day spent in an institutional as opposed to an outpatient setting, the findings on LOS closely mirror the cost findings. That is, patient groups with the highest management cost spent the longest time in an institutional setting (ie, older patients [P=0.0034] and those with more serious neurological deficits [P<0.0001], worse functional impairment [P<0.0001], more severe stroke subtype [P<0.0001], or comorbid conditions [P=0.0185]). Whereas patients spent, on average, 29% of the 12 weeks in the acute-care hospital (readmissions excluded), this initial acute care represented 71% of the cumulative cost (Figure 2⇓). Overall, institutional care represented 92% of the total cost.
Patient Referral and Outpatient Care
Twenty-three percent of the patients in this study were never discharged from the hospital: 16% died during the acute hospitalization and 7% were still in the hospital at the end of follow-up. Of the remainder, 31% were referred to an inpatient rehabilitation facility and another 14% to a nursing home. The remaining 32% returned home or went to a retirement home after discharge from the hospital. The mean hospital stays for these patient groups lasted 17, 25, and 21 days, respectively. The pattern of referral according to age, stroke severity, and stroke subtype is given in Figure 3⇓.
Whether or not patients are able to return to their home environment during the first 3 months after stroke onset seems closely related to indicators of the patients’ frailty and impairment. For example, the proportion of patients who returned home varied between 81% for very mildly impaired patients and 26% for the most severely impaired (Table 5⇓). Of those patients who were living with a partner, a considerably higher proportion were able to return home (54%) than of those who were living alone (43%). This indicates that the availability of an informal caregiver at home positively influences discharge from an institutional setting. Also, patients without comorbidities were more likely to return home (47% versus 66%) and, once at home, fewer of them needed to rely on outpatient care (88%) than did their counterparts (96%). All of these differences were statistically significant at the P<0.0001 level. The daily at-home cost is also provided in Table 5⇓. As can be deduced from the results, the distribution of the daily at-home cost was skewed to the right for a number of subgroups, with a few high-cost patients inflating the mean values. Overall, the patterns with respect to the intensity of use were consistent, however, with what was observed for inpatient care. More details with respect to the resource use underlying this daily at-home cost are provided in Table 6⇓.
The initial hospital LOS by country and its breakdown among the 5 specific wards are illustrated in Figure 4⇓. The general ward was most frequently used in all countries except Norway, Sweden, and Denmark, where the stroke ward accounted for more than half of the initial LOS. Some studies have found that use of a stroke unit may be associated with a shorter hospital LOS and expedient discharge from the hospital to a nursing home or home (eg, see References 15 through 1715 16 17 ). It is believed that this is due to a timely and multidisciplinary integrated treatment. However, this anticipated relationship was not found among the trial countries. Countries with extensive use of a stroke unit still had an above-average LOS. It may be argued, however, that the proportion of patients admitted to a stroke ward is more important than the average LOS in a stroke ward when describing the intensity of use. The line in Figure 4⇓ represents the proportion of patients initially admitted to a stroke ward. The anticipated relationship between these proportions and the hospital LOS was not detected. For example, Denmark had a very large proportion of patients admitted to a stroke ward, but its average hospital LOS was also the longest. Caution should be taken when interpreting these findings: due to the integrated treatment approach in stroke units, discharge from the hospital does not necessarily accurately mark the distinction between acute care and rehabilitation for these patients.
The patients’ place of residence by country at the end of the study period is illustrated in Figure 5⇓. Overall, 59% of the surviving patients returned to a residential setting (home or retirement home) after 3 months, varying between 28% for Norway and 71% for the United States. The remainder were still being cared for in an inpatient setting: 15% in a rehabilitation center, 14% in a nursing home, and 12% in an acute-care hospital. On average, >80% of the outpatients continued to use healthcare services.
This study documents the management patterns and associated costs during the first 3 months after acute ischemic stroke in 2 international trials. The mean total cost was found to be ≈$14 000 (equivalent to the cost of ≈46 hospital days) in this multinational study, with acute hospitalization accounting for most of this cost. Indicators of patient frailty and disease severity were clear determinants of the total cost and its components, a finding that agrees well with the results of single-setting studies on the determinants of stroke outcome or cost (see References 18 through 2018 19 20 ). It should be noted, however, that conclusions about the independent impact of these factors and thus potential causal relationships must be cautious, as only univariate analyses are presented here.
The data for these analyses were collected in the context of 2 large, international studies that evaluated the efficacy and safety of a potential neuroprotective agent. Our experience was that this form of data gathering is not only feasible but is also generally welcomed by patients and their caregivers, judging by the high response rates and informal feedback received by trial coordinators. The major strength of this approach is that detailed data are collected according to a standardized procedure on a patient population that is fairly homogeneous across centers and countries.
This approach also has some drawbacks, however.21 A concern often voiced is that clinical trials include highly selected cases, and results are therefore not readily transferable to an unselected stroke population. Although true to some extent, the inclusion and exclusion criteria of these trials were broad enough for the population to be representative of the target population for neuroprotective treatment. Second, it is believed that clinicians may intentionally or otherwise modify their patient care when participating in a clinical trial, thereby causing the observed management patterns to be less representative of actual daily practice. In this regard, some investigators suggested that in these trials, patients might have been kept in hospital longer to complete the treatment regimen and closely monitor their progress. The acute-hospital LOS reported in our study does appear to be somewhat higher than that observed in other studies. Beech and colleagues22 observed a mean LOS of 36 and 39 days in 2 UK centers, 16 days in France, and between 20 and 26 days in Germany. The corresponding numbers in our study were 32, 24, and 31 days for the respective countries. A median LOS of 6 days (range, 1 to 63 days) has been reported for 1 academic center in the United States,19 compared with 9 days (range, 1 to 84) in the present study. The Copenhagen Stroke Study reported a mean LOS of 37 days, including inpatient rehabilitation.18 Institutional LOS in Denmark, defined similarly, amounted to 50 days in our study. A World Health Organization study, on the other hand, has shown that in 7 European countries the median hospital LOS for stroke survivors was 4 weeks,23 which agrees well with the 26 days observed for the European countries in our study. A third limitation is the relatively brief follow-up in the context of a condition with long-term sequelae. It is unclear whether the patterns observed in this study would hold over the longer term.
The cost of institutional care was derived from the LOS in the various institutions and ward types and the corresponding average per diem cost. As a result, potential differences in inpatient diagnostic and therapeutic services received were not documented. Research has shown, however, that the cost of inpatient care is largely driven by the LOS as opposed to work-up procedures and salaries of physicians and therapists.24 25 This is especially true for longer LOSs, as the contribution of the fixed costs, those related to diagnostic studies, for instance, to the total drops with increasing LOS. In a separate study using information on >100 000 patients from 5 US state discharge databases, we estimated the total hospital costs (including physicians’ fees) for ischemic stroke at $8728. This compares well with the US inpatient cost of $7461 calculated in the present study, based on the per diem cost only.
These per diem costs as well as the unit costs for outpatient care have been retrieved from the United Kingdom. One could argue that for estimation of the costs in each participating country, local unit costs should have been used instead. This would indeed be appropriate if the primary intent were to inform decision makers about the budgetary impact of the observed stroke management in the health jurisdictions concerned. As our aim was rather to report the influence of various patient and disease characteristics on stroke management patterns and costs across a range of countries, use of local unit costs would have artificially created cost differences attributable to variations in reimbursement systems and cost structure, as opposed to true differences in the underlying management patterns, and as such, would have unnecessarily obfuscated the findings.
In contrast to some other studies, we did not observe a relationship between the intensity of use of a stroke ward, defined as either average LOS in the stroke unit or as the proportion of patient admitted, and the initial hospital LOS. This result may be due to the way treatment is pursued and what is counted as “hospital” LOS. Some countries with infrequent use of stroke wards may be shifting a large portion of the burden of care into facilities beyond the hospital. For the same reason, one should be cautious in drawing conclusions based on a cross-country comparison, as it is conceivable that the LOS in countries with intensive use of stroke wards was induced by other external factors and therefore would have been even higher with less reliance on these specialized units.
This study provides a comprehensive picture of the healthcare services used presently for the treatment and rehabilitation of stroke victims. The findings suggest that medical treatment aimed at a reduction of the initial stroke severity can not only decrease human suffering but may also affect the use of healthcare resources. The detailed information presented here should benefit researchers who conduct formal evaluations of the cost-efficiency of specific stroke treatments.
Stroke Economic Analysis Group
John Cai, PhD, Caro Research, Concord, Mass (currently affiliated with the Massachusetts Division of Health Care Financing and Policy, Boston, Mass); J. Jaime Caro, MDCM (Chairman), Caro Research, Concord, Mass; Ingrid Caro, MEd, Caro Research, Montreal, Canada; Inge Duchesne, MS, Janssen Research Foundation, Beerse, Belgium; Krista F. Huybrechts, MS, Caro Research, Concord, Mass; Wendy S. Klittich, BS, Caro Research, Concord, Mass; Judith A. O’Brien, RN, Caro Research, Concord, Mass; and Gabriel Raggio, ScD, Caro Research, Concord, Mass.
This work was supported in part by a grant from the Janssen Research Foundation, Beerse, Belgium.
- Received November 10, 1999.
- Revision received December 15, 1999.
- Accepted December 27, 1999.
- Copyright © 2000 by American Heart Association
Holloway RG, Benesch CG, Rahilly CR, Courtright CE. A systematic review of cost-effectiveness research of stroke evaluation and treatment. Stroke. 1999;30:1340–1349.
Hacke W, Kaste M, Fieschi C, Toni D, Lesaffre E, von Kummer R, Boysen G, Bluhmki E, Höxter G, Mahagne M-H, Hennerici M, for the ECASS Study Group. Intravenous thrombolysis with recombinant tissue plasminogen activator for acute hemispheric stroke: the European Cooperative Acute Stroke Study (ECASS). JAMA. 1995;274:1017–1025.
Caro JJ, Huybrechts KF, for the Stroke Economic Analysis Group. Stroke treatment economic model (STEM): predicting long-term costs from functional status. Stroke. 1999;30:2574–2579.
Fagan SC, Morgenstern LB, Petitta A, Ward RE, Tilley BC, Marler JR, Levine SR, Broderick JP, Kwiatkowski TG, Frankel M, Brott TG, Walker MD, NINDS rt-PA Stroke Study Group. Cost-effectiveness of tissue plasminogen activator for acute ischemic stroke. Neurology. 1998;50:883–890.
Grotta J, for the US and Canadian Lubeluzole Ischemic Stroke Study Group. Lubeluzole treatment of acute ischemic stroke. Stroke. 1997;28:2338–2346.
Netten A, Dennett J. Unit Costs of Health and Social Care. Canterbury, UK: Personal Social Services Research Unit, University of Kent; 1996.
National Health Service. Scottish Health Service Costs. Edinburgh, Scotland: Information and Statistics Division, National Health Service; 1995.
Caro J, Klittich W, McGuire A, Ford I, Pettitt D, Norrie J, Shepherd J, for the WOSCOPS Economic Analysis Committee. International economic analysis of primary prevention of cardiovascular disease with pravastatin in WOSCOPS. Eur Heart J.. 1999;20:263–268.
Adams HP Jr, Bendixen BH, Kappelle LJ, Biller J, Love BB, Gordon DL, Marsh E, and the TOAST Investigators. Classification of subtype of acute ischemic stroke: definitions for use in a multicenter clinical trial. Stroke. 1993;24:35–41.
Kalra L. The influence of stroke unit rehabilitation on functional recovery from stroke. Stroke. 1994;25:821–825.
Strand T, Asplund K, Eriksson S, Hägg E, Lithner F, Wester PO. Stroke unit care: who benefits? comparisons with general medical care in relation to prognostic indicators on admission. Stroke. 1986;17:377–381.
Stroke Units Trialists’ Collaboration. Collaborative systematic review of the randomized trials of organized inpatient (stroke unit) care after stroke. BMJ. 1997;314:1151–1159.
Jørgenson HS, Nakayama H, Raaschou HO, Olsen TS. Acute stroke care and rehabilitation: an analysis of the direct cost and its clinical and social determinants: the Copenhagen Stroke Study. Stroke. 1997;28:1138–1141.
Diringer MN, Edwards DF, Mattson DT, Akins PT, Sheedy CW, Hsu CY, Dromerick AW. Predictors of acute hospital costs for treatment of ischemic stroke in an academic center. Stroke. 1999;30:724–728.
Alexander MP. Stroke rehabilitation outcome: a potential use of predictive variables to establish levels of care. Stroke. 1994;25:128–134.
Beech R, Ratcliffe M, Tilling K, Wolfe C, on behalf of the participants of the European Study of Stroke Care. Hospital services for stroke care: a European perspective. Stroke. 1996;27:1958–1964.
Aho K, Harmsen P, Hatono S, Marquardsen J, Smirnov VE, Strasser T, on behalf of the participants in the WHO Collaborative Study on the Control of Stroke in the Community. Cerebrovascular disease in the community: a result of a WHO collaborative study. Bull World Health Organ. 1980;58:113–130.
Dennis M, Langhorne P. So stroke units save lives: where do we go from here? BMJ. 1994;309:1273–1277.