Background and Purpose Stroke is a common cause of mortality and morbidity in Europe and a major consumer of healthcare costs. However, outcomes from stroke vary significantly across Europe, raising the issue of the extent to which the delivery of care varies across Europe.
Methods A multicenter, multinational study collected data on hospital admissions for stroke. These included patient baseline characteristics, clinical status, and use of inpatient services.
Results Initial results examined inpatient services in nine hospitals in six countries. Statistically significant differences existed between hospitals in key processes of care, most notably in the areas of (1) mean length of stay (11 to 39 days) and median length of stay (8 to 21 days), (2) percentage of admissions receiving brain imaging (30% to 98%) and neurosurgery (0% to 31%), and (3) percentage of admissions with an identified “need” who received physiotherapy (44% to 90%) and occupational (0% to 65%) and speech (0% to 59%) therapy. Although there were significant hospital differences (P<.001) in the case mix of admissions in terms of level of consciousness, presence of incontinence, prestroke Rankin Scale score, and age, these did not explain the differences in care across sites.
Conclusions There is geographic inequity across Europe in the care a stroke patient can expect to receive. Ongoing analysis will examine the link between European variations in service use and outcomes, in terms of mortality, handicap, and functional ability, as a means of indicating which patterns of care are the most effective.
Throughout Europe stroke represents a major health problem, both in terms of its impact on mortality and morbidity and its consumption of healthcare resources and costs.1 2 3 4 It is estimated that around 1 million ischemic strokes occur in Europe each year.2 Overall, stroke is the third most common cause of death in Europe, with an estimated 400 000 subjects of European member states dying each year because of the consequences of stroke.1 It has also been estimated that stroke care accounts for 3.9% and 4.6% of National Health Service expenditure in England and Wales and in Scotland, respectively.4
However, within this global picture there is considerable variation in standardized mortality ratios for stroke among countries. With a European average of 100, indirect standardized mortality ratios vary from 78 in France to 199 in Portugal, with direct age-standardized mortality rates being 33.53 per 100 000 in France and 85.42 per 100 000 in Portugal.5 The limited evidence available before the start of the study described here also suggested variation both across and within countries in the way stroke care is delivered. Admission to the hospital is a common form of care, but studies of defined populations have exposed variations in hospitalization rates, ie, the proportion of stroke cases that are admitted to the hospital. A study in Oxford, England, estimated a hospitalization rate of 55%,6 a study of three health authorities in southern England an overall rate of 71%,7 and a study in Umbria, Italy, a rate of 85%.8 The MONICA study of a range of populations throughout Europe also found that the proportion of nonfatal stroke cases diagnosed and treated out of the hospital varied from 0% to 16%, being 5% or less in 13 populations.9
Statistically significant differences have been found among three hospitals in southern England in the percentage of patients receiving inpatient physiotherapy, ranging from 59% in one hospital to 91% in another.7 There were also nonsignificant differences in the percentages of patients receiving inpatient occupational and speech therapy, ranging from 46% to 62% and from 41% to 58%, respectively. As a preliminary exercise to the research described in this article, 13 European centers were surveyed to identify the types of care received by stroke patients.10 Results based on routinely available data and the judgments of key staff involved in the care of stroke patients indicated variations across centers in length of stay (LOS), in the type of bed used by patients, and in the use of community rehabilitation. Clearly, such variations in practice will lead to variations in costs among hospitals. A study of the direct costs of stroke care in two hospitals in southern England estimated costs per case as £2650 in one hospital and £4450 in another.11
Such variations among hospitals in the use of health services and in the outcomes of care raise questions about the extent to which the variations in the deployment of healthcare resources are leading to the observed variations in outcomes and, as a consequence, whether the resources devoted to stroke care could be used in a more cost-effective manner. A detailed study of care in a sample of centers throughout Europe allows these questions to be addressed. Such a study is timely given that there is currently a lack of information on the most effective and cost-effective methods of deploying stroke services. Indeed, a recent survey in England demonstrated the poor organization of stroke services.12 In addition, all European states exist within a climate of healthcare cost containment expressed in attempts to stop costs rising in real terms, real costs per head increasing, or costs rising faster than national resources.13 Such pressures are leading to moves to alter the pattern of healthcare provision, with stroke care being no exception.
This article presents results from a study that is analyzing the care provided, and the subsequent outcomes of that care, for patients who have been admitted to a sample of hospitals throughout Europe for a first-ever stroke. These results describe the European variation in the quantity and types of services received by stroke patients and allow analysis of the extent to which any variations in care are due to variations in case mix across study sites. The final section of the article discusses the implications of these findings and the long-term goals of this study.
Materials and Methods
The long-term goal of the study is to develop protocols for improving the effectiveness and cost-effectiveness of European stroke care. The objectives that need to be achieved to meet this goal have been described elsewhere.10 This article concerns the objective of maintaining hospital-based stroke registers to describe the services received by stroke patients in individual study hospitals.
The study currently covers 22 hospitals in seven European states: England, France, Germany, Hungary, Italy, Portugal, and Spain. The hospitals were chosen because they contained staff with an interest in stroke research and the ability to provide the resources for collecting the data needed by the study. The hospitals are not necessarily “typical” of care in their country, but the vast majority provide general acute care and some are also teaching hospitals.
Patient-based data collection began in the majority of hospitals in September 1993. Data collection relates to all first-ever stroke admissions during the subsequent year as defined according to the World Health Organization (WHO) definition of stroke.14 The Appendix lists the data collected on core patient characteristics and inpatient service use. These data were agreed on at an initial workshop of study participants that drew on their previous involvement in stroke research and the establishment of stroke registers.15 16 The clinical variables were largely identical to those that have been used in previous stroke research (eg, in the MONICA and WHO studies).14
The variables relevant to this article include the baseline characteristics of patients (eg, age, sex, prestroke level of handicap as defined by the Rankin Scale17 ); clinical state at time of maximum impairment (eg, level of consciousness as defined by the Glasgow Coma Scale18 ); site of paralysis; existence of speech or swallowing problems; use of inpatient services (eg, type of bed occupied and LOS in each [medical, neurological, intensive care, etc]); use of major diagnostic (brain imaging, etc) or therapeutic interventions (neurosurgery, etc); and use of rehabilitation services (physiotherapy and occupational and speech therapy).
To improve interobserver agreement, the Rankin Scale categories were merged into “no symptoms/minor symptoms,” “minor handicap and moderate handicap,” and “severe handicap” as described in the Oxfordshire Community Stroke Project.6 In the Glasgow Coma Scale, the somnolent and semiconscious categories were merged. Outcome data are also being collected at baseline and at 3 months and 12 months after the onset of stroke. These include mortality, ability to perform activities of daily living as defined by the Barthel Index,19 and handicap according to the Rankin Scale.17
These data are extracted prospectively from case notes and other routine hospital records. In addition to the modification of some scales, to standardize data collection across centers and between observers, a manual was produced containing the definitions of each data item and how it could be collected.
To explore existing European variations in stroke care for a similar mix of patients, the results in this article focus on nine hospitals where data collection was largely complete at the time of the analysis. Further criteria for the selection of hospitals were as follows. With the exception of Hungary, from each country a hospital was selected that admitted a relatively large number of patients relative to other study hospitals in that country. In Germany and England, further hospitals from different parts of the country were also included. Finally, one of the hospitals (hospital 9) was deliberately selected because it differs from the others in both its case mix and in the services it provides for stroke care. This was to help illustrate the way in which a multicenter study such as this can allow the impact of variations in practice on patient outcome to be explored.
For each hospital, this study presents descriptive statistics of the case mix and services received by stroke patients. For key process of care variables, regression analysis was used to identify the extent to which variations in practice across sites were due to variations in case mix. In each case, an initial model covering all case-mix variables was fitted followed by backward stepwise elimination to identify case-mix variables significant at the 5% level.
Table 1⇓ details key case-mix variables for patients admitted to the nine hospitals during the 1-year period of case ascertainment. Hospital 9 admitted a younger case mix and a higher percentage of patients diagnosed as having a subarachnoid hemorrhage. The percentage of patients with no or only minor symptoms of handicap before stroke ranged from 29.2% in hospital 8 to 86.4% in hospital 5, and of those with severe handicap before stroke, from 1.4% in hospital 7 to 5.9% in hospital 9. The percentage of fully conscious patients at the time of maximum impairment ranged from 53.9% in hospital 2 to 86.2% in hospital 7. The percentage of patients incontinent during the 7 days following their stroke ranged from 17.4% in hospital 7 to 69.5% in hospital 9. Overall, there were significant differences between hospitals in the number of admissions with the following characteristics: level of consciousness (P<.001), presence of incontinence (P<.001), prestroke Rankin score (P<.001), and age (P<.001).
Table 2⇓ lists for each hospital the type of bed that patients were first admitted to and their mean and median LOS to death in the hospital or to discharge, inclusive of any subsequent bed transfers within the study hospital. Admission to an acute medical bed was the most common first bed type for stroke care, although in hospitals 3, 7, and 9 a substantial minority of cases were admitted to neurology beds and in hospital 6 this was the dominant form of care. Hospital 9 also admitted just under half of its patients to intensive care beds. In hospitals 2, 4, 5, 6, and 8, upwards of 80% of patients remained in their bed type of admission. In hospital 1, 87.1% of transferred patients moved to a rehabilitation bed on a stroke unit. Corresponding statistics for the main bed type after transfer for hospital 3 were 75.3% of patients transferred to a neurology bed; for hospital 7, 82.0% to a neurology bed; and for hospital 9, 79.3% to a neurosurgery bed.
Mean LOS inclusive of deaths and discharges alive ranged from 11.0 days in hospital 5 to 38.9 days in hospital 1, and median LOS from 8 days in hospital 5 to 21 days in hospitals 2 and 7. Results also given in Table 2⇑ indicate the percentage of stroke patients dying in the hospital and the mean and median LOS statistics by hospital for those patients who were discharged alive. In all of the hospitals, mean LOS statistics are longer when hospital deaths are excluded. However, even allowing for variations in the number of patients dying in the hospital, large variations remain in the mean and median LOS of stroke cases.
Backward stepwise linear regression revealed that the factors significantly relating to hospital LOS (including all subjects) were prestroke Rankin score, incontinence, and level of consciousness. After these factors had been taken into account, all hospitals except hospital 2 and hospital 7 had significantly shorter LOS than hospital 1. The same factors were significantly related to LOS in the hospital when the model was restricted to those subjects alive on discharge from the hospital. A similar pattern of results emerged, but hospital 2 had a significantly longer LOS than hospital 1. The results in Table 3⇓ estimate what the mean LOS differences between hospitals would be if they were all treating an equivalent case mix. Differences are given relative to the estimated LOS of hospital 1. In the model that includes all patients, a difference of 29.6 days between the hospitals with the shortest and longest LOS is predicted, and in the model that excludes deaths in the hospital, there is a prediction of a difference of 40.0 days. Comparison of these results and the unadjusted mean LOS figures given in Table 2⇑ demonstrates that factors other than case mix are the main determinants of the LOS differences across study sites.
Previous research20 has demonstrated that for elderly patients their LOS in an acute-care hospital can be influenced by the proximity of institutions, such as nursing homes, that provide lower levels of care. The final row of Table 2⇑ indicates the percentage of patients who were discharged alive to a destination classified as “home alone” or “home with caregiver.” The residual percentage covers transfer to an institution providing a lower level of care. Transfer to another institution does not seem to be an underlying cause of the variations in LOS.
Table 4⇓ indicates, by hospital, the percentage of patients who had major diagnostic and therapeutic interventions for their stroke care. Brain imaging, usually in the form of CT scanning, was common in all but one of the hospitals. Hospital 2 is equipped with a CT scan machine, but stroke patients were not given a high priority.
Variations in the percentage of patients receiving angiography (0% to 30.0%), Doppler sonography (0% to 97.1%), and echocardiography (0.4% to 55.9%) were revealed. The use of neurosurgery, carotid surgery, and other forms of vascular surgery is uncommon except in hospital 9. Here a relatively large percentage of patients underwent neurosurgery, explaining its high use of intensive care beds (Table 2⇑). The factors influencing the use of neurosurgery were investigated using logistic regression. Overall, 63 people underwent neurosurgery, with the number per hospital ranging from 0 (0%) to 37 (31%). The case-mix variables that were significantly related to use of neurosurgery were the subject's age, diagnosis, and presence of incontinence. Once these had been controlled for, there remained a significant difference between hospital 9 and the other hospitals. The odds ratio for having neurosurgery was 48.5 (95% confidence interval, 6.07 to 387.56) for hospital 9 compared with all other hospitals. In terms of the specific diagnosis of the patients undergoing neurosurgery in hospital 9, 11 patients with a subarachnoid hemorrhage had neurosurgery (6 did not) and 26 patients without a subarachnoid hemorrhage had neurosurgery.
The results in Table 5⇓ focus on the provision of physiotherapy and occupational and speech therapy services for stroke care. The data refer only to formal therapy sessions provided by trained therapists. There is variation across sites in the overall percentage of patients receiving therapy. However, it is more revealing to compare hospitals in terms of the number of patients receiving therapy as a percentage of those with an identified “need” for therapy. For the results in Table 4⇑, a patient having an identified site of paralysis or deficit was regarded as a proxy indicator of need for physiotherapy or occupational therapy. A patient having an identified speech or swallowing problem was regarded as a proxy indicator of need for speech therapy. The percentage of patients with an identified need who actually received therapy ranged from 44.0% to 90.3% for physiotherapy, from 0% to 59.5% for speech therapy, and from 0% to 64.9% for occupational therapy. When case-mix variables had been taken into account, there were significant differences among the hospitals in the proportion of those in need who received therapy.
For a sample of European hospitals, the results have exposed wide variations in the patterns of care for stroke. Furthermore, it has been demonstrated that case mix, as assessed in this large population-based study using only major clinical predictors of outcome, does not account for all of this variation. Residual factors will include variations in the availability of resources among hospitals and in local clinical preferences. The findings are in line with previous evidence that has studied within-country variation of services.7 21 22 However, this is probably the first study to expose the nature of the variation in service use among countries based on analysis of data collected using standard questionnaires covering a wide range of clinical and service use variables. Perhaps the current variation in services between hospitals is inevitable given that there is currently a lack of consensus about which patterns of care are the most effective and cost-effective.
The mixed pattern of care that is indicated does have two major implications. First, there appears to be geographic inequity across Europe in services for stroke. Even though the efficacy of many interventions may be unclear, the type of care a stroke patient can expect depends on their hospital of admission. Second, the results suggest that the current overall pattern of care cannot be either effective or cost-effective. There is likely to be overprovision of ineffective services in some centers and underprovision of effective services in others.
Using a sample of European hospitals, this study has exposed the current baseline in terms of variations in stroke care. To some extent, though, the more interesting questions regarding the care they provide are still to be addressed. The first of these is whether the variations in care are leading to variations in outcome. By linking data on the process of care with data on outcome at 3 and 12 months, results will be generated indicating which patterns of care appear to be the most successful. Although findings may need to be tested in more detailed trials, this study will have provided an important function in generating the hypotheses. Indeed, a strength of European multicenter studies is that they provide an efficient means of studying the implications of a range of alternative options for delivering care.
Ongoing analysis will also examine the variations in the resources devoted to stroke care within the study hospitals in terms of the types and numbers of staff, for example. The organization of these resources will also be analyzed. Hence it will be possible to explore the factors that are leading to the relatively short LOS of hospital 5 and assess whether those factors could be replicated in hospital 1. To supplement this aim, a grant for an extension to the existing study has recently been secured from the European Union. This extended study will examine European variations in the levels of support available from family members and the impact this may have on the demands patients place on health services.
Finally, further analysis will compare the costs of stroke care among centers. By using data on service use, resource use, and the local cost of those resource items (eg, the cost of employing a nurse in different countries), it will be possible to assess the extent to which hospital variations in the overall costs of stroke care are due to variations in the amount of services patients receive, variations in the number of individuals providing that care, or variations in the local cost of the staff, materials, and equipment.
In conclusion, the study currently holds data relating to over 4000 admissions for stroke across 22 European hospitals. Both of these figures will continue to grow during the follow-up study of stroke. This article presents early results that expose European variations in the types of care provided for stroke patients. These variations are not explained by variations in case mix, and hence the results have major implications for the equity and cost-effectiveness of European stroke care. Ongoing analysis will explore the links between service use, resource use, costs, and outcomes, and in so doing it will provide valuable evidence for improving the quality, effectiveness, and cost-effectiveness of European stroke care.
Prof D.H. Barer, Dr Y. Ellul, Department of Medicine for the Elderly, Newcastle General Hospital, England; Prof S. Ebrahim, M. Ayana, Dr P. Gompertz, Dr R. Harwood, Dr P. Pound, Department of Primary Care and Population Sciences, Royal Free Hospital School of Medicine, London, England; Dr H. Rogers, Centre for Health Service Research, University of Newcastle, England; Prof M. Giroud, Dr M. Menassa, Service de Neurologie, Centre Hospitalier Regional et Universitaire de Dijon, France; Prof K. Kunze, Neurologischen Universitatsklinik, Hamburg-Eppendorf, Germany; Dr J. Berger, Institute of Mathematics and Computer Science in Medicine, University Hospital Eppendorf, Germany; Dr B. Hausler, Dr W. Mall, Dr H. Nolting, Institute fur Gesundheits und Sozialforschung GmbH (IGES), Berlin, Germany; Prof Z. Nagy, Dr C. Ovary, Dr Z. Vokoq, National Stroke Centre, Budapest, Hungary; Prof D. Inzitari, Dr A. Di Carlo, Dr M. Lamassa, Dr P. Vanni, Dipartimento di Scienze Neurologiche & Psichiatriche, Ospedale Careggi, Italy; Prof M. Carrageta, Prof Bigote de Almeida, Dr J. Namora, Dr I. Remedios, Hospital Garcia de Orta, Servico de Cardiologia, Almada, Portugal; Dr J. Dias, Divisao de Epidemiologa, Direccao Geral da Saude, Lisboa, Portugal; Dr A. Arias, Dr P. Casquero, Dr S. Montserrat, Dr M. Torrent, Direccion Provincial INSALUD, Gabinete de Estudios, Palma de Mallorca, Spain.
Key Patient Data Collected
Date of birth: day/month/year
Ethnic group: Caucasian, non-Caucasian
Sex: male, female
Living conditions prior to stroke: alone, not alone, institutionalized, unknown
Employment status: employed, unemployed, retired, housewife, unknown
Rankin scale score prior to stroke: no symptoms, minor symptoms, minor handicap, moderate handicap, moderately severe handicap, severe handicap
Clinical State at Time of Maximum Impairment (Within First 24 Hours)
Coma: fully conscious, somnolent, semicoma, coma, no record
Degree of impairment: no deficit, weakness, paralysis
Site of impairment: upper limb−left/right, lower limb−left/right
Has the patient any speech disturbance as a result of stroke? Y/N
Is there a speech problem? Dysarthria: Y/N; Aphasia: Y/N
Has the patient any swallowing problem as a result of stroke? Y/N
Is the patient incontinent of urine or catheterized in the first 7 days after stroke? Y/N
Was the patient confused in first week after stroke (ie, disoriented, abnormal behavior)? Y/N
Specific stroke diagnosis: cerebral infarction (unspecified), total anterior circulation syndrome, partial anterior circulation syndrome, posterior circulation syndrome, lacunar syndrome, cerebral hemorrhage, subarachnoid hemorrhage, unclassified, not known
Inpatient Care for Current Stroke or for Reoccurrence During the Year After First Stroke
Date of stroke: day/month/year
Date of admission: day/month/year
Type of bed: acute medical, neurology, intensive care, surgery, long stay, rehabilitation ward, private
Date of first transfer: day/month/year
Bed type after first, second, or third transfer: acute medical, neurology, intensive care, surgery, long stay, rehabilitation ward, private
Date of second or third transfer: day/month/year
Diagnostic techniques used: brain imaging (Y/N), angiography (Y/N), Doppler (Y/N), echocardiogram (Y/N)
Surgical interventions: neurosurgery (Y/N), carotid surgery (Y/N), other vascular surgery (Y/N)
Amount of inpatient rehabilitation: number of sessions of formal physiotherapy and occupational and speech therapy
Date of discharge/transfer from acute hospital: day/month/year
Destination from acute hospital: home alone, home with companion, institution, rehabilitation hospital, other
The authors thank the European Union for providing the funds for the project. They also thank their fellow participants for their help, advice, and encouragement throughout this project. In particular, Professor M Giroud, Professor D. Inzitari, and Professor K. Kunze are thanked for their comments on this paper.
- Received April 11, 1996.
- Revision received July 1, 1996.
- Accepted July 19, 1996.
- Copyright © 1996 by American Heart Association
Annual World Health Statistics. Geneva, Switzerland: World Health Organization; 1990, 1991, 1992.
Office of Health Economics Stroke. London, UK: Office of Health Economics; 1988.
Carstairs V. Resource consumption and the cost to the community. In: Gillingham FJ, Mawdsley C, Williams AE, eds. Stroke. New York, NY: Churchill Livingstone Inc; 1976:516-528.
Holland WW (project director). European Community Atlas of Avoidable Death. Oxford, UK: Oxford University Press; 1991.
Oxfordshire Community Stroke Project. Incidence of stroke in Oxfordshire: first year's experience of a community stroke register. BMJ. 1983;287:712-717.
Wolfe CDA, Taub NA, Woodrow J, Richardson E, Warbuton FG, Burney PGJ. Patterns of acute stroke care in three districts of southern England. J Epidemiol Community Health. 1993;47:144-148.
Ricci S, Celani MG, La Rosa F, Vitali R, Duca E, Ferraguzzi R, Paolotti M, Seppoloni D, Caputo N, Chiurulla C. SERPIVAC: a community based study of stroke incidence in Umbria, Italy. J Neurol Neurosurg Psychiatry. 1991;54:695-698.
Asplund K, Bonita R, Kuulasmaa K, Rajakangas AM, Feigin V, Schaedlich H, Suzuki K, Thorvaldsen P, Tuomilehto J. Multinational comparisons of stroke epidemiology. Stroke. 1995;26:355-360.
Wolfe CDA, Taub N, Bryan S, Beech R, Warburton F, Burney PGJ. Variations in the incidence, management and outcome of stroke in residents under the age of 75 in two health districts of southern England. J Pub Health Med. 1995;17:411-418.
Lindley RI, Amayo EO, Marshall J, Sandercock PAG, Dennis M, Warlow CP. Hospital services for patients with acute stroke in the United Kingdom: the Stroke Association survey of consultant opinion. Age Ageing. 1995;24:525-532.
Abel Smith B. Cost Containment and New Priorities in Health Care. Aldershot, UK: Avebury; 1992.
Wolfe CDA, Taub NA, Woodrow J, Richardson E, Warburton FG, Burney PGJ. Does incidence, severity, or case fatality of stroke vary in southern England. J Epidemiol Community Health. 1993;47:139-143.
Giroud M, Milan C, Beuriat P, Gras P, Essayagh E, Arveux P, Durnas R. Incidence and survival rates during a two year period of intracerebral and subarachnoid haemorrhages, cortical infarcts, lacunes and transient ischaemic attacks. The stroke registry of Dijon: 1985-89. Int J Epidemiol. 1991;20:892-899.
Beech R, Withey C, Morris R. Understanding variations in lengths of stay between hospitals for fractured neck of femur patients and the potential consequences of reduced stay targets. J Public Health Med. 1995;17:77-84.
King's Fund Consensus Statement. The treatment of stroke. BMJ. 1988;297:126-128.
Wade DT. Stroke (acute cerebrovascular disease). In: Stevens A, Raftery J, eds. Health Care Needs Assessment: The Epidemiology Based Needs Assessment Reviews. Oxford, UK: Radcliffe Medical Press; 1994;1:111-255.