Article Text
PDF
Abstract
Background The need for stroke care is escalating with an ageing population, yet methods to estimate the delivery of effective care across countries are not standardised or robust. Associations between quality and intensity of care and stroke outcomes are often assumed but have not been clearly demonstrated.
Objective To examine variations in acute care processes across six European populations and investigate associations between the delivery of care and survival.
Methods Data were obtained from population-based stroke registers of six centres in France, Lithuania, UK, Spain, Poland and Italy between 2004 and 2006 with follow-up for 1 year. Variations in the delivery of care (stroke unit, multidisciplinary team and acute drug treatments) were analysed adjusting for case mix and sociodemographic factors using logistic regression methods. Unadjusted and adjusted survival probabilities were estimated and stratified by levels of Organised Care Index.
Results Of 1918 patients with a first-ever stroke registered, 30.7% spent more than 50% of their hospital stay in a stroke unit (13.9–65.4%) among centres with a stroke unit available. The percentage of patients assessed by a stroke physician varied between 7.1% and 96.6%. There were significant variations after adjustment for confounders, in the organisation of care across populations. Significantly higher probabilities of survival (p<0.01) were associated with increased organisational care.
Conclusions This European study demonstrated associations between delivery of care and stroke outcomes. The implementation of evidence-based interventions is suboptimal and understanding better ways to implement these interventions in different healthcare settings should be a priority for health systems.
- Evidence-Based Neurology
- Cerebrovascular Disease
- Health Policy & Practice
- Stroke
Statistics from Altmetric.com
Introduction
With ageing populations and constrained health budgets, healthcare systems globally will be required to deliver effective interventions and reduce management costs for a range of long-term conditions. Stroke remains one of the most common causes of death and lost disability-adjusted life-years, and the incidence of stroke is predicted to increase.1 ,2
Over the past two decades, wide variations in stroke care and its associated costs in Europe have been reported.3–5 International studies have indicated that stroke unit care reduces mortality by preventing complications,6 and organised and multidisciplinary team (MDT) care are associated with lower mortality.7 A previous European study showed that where there were high costs, associated with stroke unit care, outcomes were improved.5 Leys and colleagues, in 2005, in a postal questionnaire survey, showed that <10% of European hospitals admitting patients with acute stroke had optimal evidence-based facilities and processes and, in 40% even the minimum level was not attained.8
Although the level of evidence for acute stroke management has substantially improved during the past two decades with the resultant emergence of national and European guidelines, wide variations have continued to be reported in acute stroke care processes, costs and outcomes across Europe and worldwide.9–12 In the USA, in the early 2000s, a national multidisciplinary panel developed 44 potential stroke performance measures and identified 21 as suitable for acute care.13 More recently, further performance measures have been developed for stroke and are being implemented by stroke quality improvement programmes, although these measures are focused on process measures identified for acute hospital-based care.14
In the European Registers of Stroke (EROS) collaboration, which included populations from six countries, we have demonstrated both significant variations in the risk of stroke and early death and dependency in a study that overcame the methodological deficiencies of previous European comparisons by including population samples, case-mix adjustment and more detailed estimation of the acute processes of care.15 ,16
In this paper we use these methods to reduce confounding of comparisons between countries and hospitals with the aim of estimating variations in the uptake of acute stroke care interventions and the organisation of stroke care in these six populations. We also assess the relationship between these detailed processes of care, adjusted for sociodemographic and case-mix factors, and survival up to 1 year after stroke.
Methods
EROS was established to study the relationship between costs, resource use and outcomes for patients with stroke across European countries. The settings and methods of case ascertainment in this study have been previously described.15 ,16 Population-based stroke registers were established in six selected European countries representing populations in central (Dijon, France), southern (Sesto Fiorentino, Italy; Menorca, Spain), eastern (Kaunas, Lithuania; Warsaw, Poland) and western Europe (London, UK) centres were selected on the basis of previous experience in running stroke registers and were therefore, not necessarily representative of their countries as a whole. The study was approved by ethics committees of each of the centres involved.
Inclusion and exclusion criteria
All patients with a first-ever stroke were identified using overlapping sources of information. Patients admitted to hospitals were identified by screening all acute hospitals serving the source population, including reviews of acute wards by the study team, checks of brain imaging referrals and reviews of hospital discharge registers. Patients who were not admitted to hospital were identified by regular screening of all primary care facilities in the study area. In addition, nursing homes and community therapists in the study were contacted and death certificates were checked regularly.
Patients with subarachnoid haemorrhage were excluded from these analyses as the level of evidence for management differs significantly from that for patients with other types of stroke.
Data collection
Data were collected between 2004 and 2006. The cohort comprised 1918 patients with a first-ever stroke with cerebral infarction (CI) or intracerebral haemorrhage and was followed up for 1 year. Stroke was defined according to the WHO definition,17 with or without brain imaging confirmation. Data collection was standardised across all centres and included patient sociodemographic characteristics (age, sex, employment status—full time, part time, unemployed, unable to work owing to disability, pre-stroke activities of daily living (Barthel Index (BI)18 and living conditions before the stroke (private household alone, private household with others, supportive environment)). Although inter-rater reliability was not tested, checks on data quality were made by other methods, including agreement between postal and site visits.19 All centres were trained to use the data collection tools before the start of the study.
Clinical markers of the case mix included the Glasgow Coma Scale (GCS),20 a previously described and validated six simple variables model, to predict outcome that included age, pre-stroke function and living circumstances, the verbal component of the GCS, arm power and the ability to walk, and these were also used in adjustments to the models.21 ,22 Stroke was classified as CI or intracerebral haemorrhage (ICH) based on brain imaging within 30 days of stroke onset. Patients without pathological confirmation of stroke subtype were unclassified (table 1).
Sociodemographic and case-mix characteristics of patients with cerebral infarction and intracerebral haemorrhage in the acute phase
Patterns of care
The EROS collaborators met several times before data collection to agree the processes of care across countries that should be recorded to deal with the study questions. These were based on European and local guidelines current at the start of the study and were influenced by the results of previous EU projects with which many of the collaborators had been involved that demonstrated variations in processes of care.3–5 23–25 The processes of care were either evidence based/included in guidelines or had no evidence base but were known to be used in in these European centres. Some were based on evidence of effectiveness (eg, stroke unit care, aspirin use after CI), whereas others were not (eg, neuroprotection). The latter were included because we wished to describe the breadth of processes of care.
The following processes of hospital admission and organisation of care were recorded: admission to hospital; stroke while an inpatient; delay in hospital admission; stroke unit admission; 50% of total length of hospital stay on stroke unit.26
Acute care interventions included use of antiplatelet therapy in patients with CI; thrombolysis (and route of administration); surgical interventions for stroke—for example, carotid endarterectomy, hemicraniectomy; drugs used to increase or decrease blood pressure; secondary prevention while in hospital in the acute phase; continuation of blood pressure lowering drugs; newly started blood pressure lowering medicine for those with newly diagnosed hypertension; cholesterol-lowering drugs.
Acute care investigations included the proportion receiving brain scans and date of the first scan (CT) or MRI; carotid Doppler imaging; transcranial Doppler imaging; clinical assessment of swallowing test; acute management by specialists—a record was made of assessment by personnel in the multidisciplinary team, including specialist stroke physician, occupational therapist (OT), physiotherapist (PT), speech and language therapists (SALT); enteral tube feeding, nasogastric tube, percutaneous endoscopic gastrostomy in managing dysphagic patients.
Use of neuroprotection—for example, N-methyl-D-aspartate agonist; routine use of intravenous fluids; drugs used to lower intracranial blood pressure; use of anticoagulants (regardless of dosage) for non-atrial fibrillation indications (oral, intravenous, subcutaneous) table 2.
Stroke care indicators
Outcomes
Mortality at 7 days, 30 days, 3 and 12 months after stroke were recorded. For the survival analysis we used 3 and 12 months as outcomes.
Organised care
In addition to the items above a summary score, the Organised Care Index (OCI) that was developed by Saposnik and colleagues,7 was calculated. Assessment by a PT, OT and SALT individually or collectively was treated as one element, stroke physician as another element and admission to the stroke unit as a third. A score of 0 indicates access to none of these services, 1 indicates access to any one service, 2 to two services and 3 to all. As with the original OCI score these scores were not weighted.
Statistical analysis
Analyses included univariate comparisons between centres for sociodemographic characteristics and case mix. χ2 Tests were used to compare categorical variables, and analysis of variance or Kruskal–Wallis tests were used for continuous variables. Associations between the processes of care and populations were investigated using univariate and multivariate logistic regression models. Multivariate models were employed to adjust for potential confounders including the six simple variables: age, pre-stroke living conditions, pre-stroke functional status (measured by BI score), arm power, ability to walk and the verbal components of GCS.21 In addition, adjustment was made for gender, employment, ability to swallow, BI score 5–10 days after the stroke and whether the patient was incontinent or not in the acute phase. BI scores were dichotomised so that 0–19 indicates physical dependency, and 20, full independence.27 Each model included all the potential confounding variables plus the centre. Variables that showed no association at a p value ≤0.2 were excluded from the final models. A p value of 0.01 was chosen as a criterion for significance.
Unadjusted and adjusted ORs and their 95% CI were presented for the six populations using London as a reference (OR=1) (table 3).
Multivariate analysis: variations in patterns of care between six European centres
To describe the relationship between mortality and OCI, the percentages of deaths by 7 days, 30 days, 3 and 12 months after stroke, were calculated for each level of OCI (table 4). Kaplan–Meier survival curves and Cox regression models (adjusting for age, dependency, arm power, ability to walk and the verbal component of the GCS) were used to calculate the unadjusted and adjusted probabilities of survival at 3 and 12 months after a stroke. Deaths during the first 72 h of stroke were excluded from this analysis as they have a different prognosis. Kaplan–Meier curves were displayed by admission to the stroke unit and by levels of OCI for the whole population as power to stratify by centre was limited by the small numbers (figure 1).
Numbers and percentages of deaths by levels of Organised Care Index (OCI).
Menorca was excluded from this analysis as no stroke unit was available. Organised care index levels were 1, 2 and 3; 3 is the highest level. Deaths in the first 72 h were excluded. Admitted, admitted to stroke unit.
All models were prespecified to investigate variations of care process across centres and the impact of care on mortality.
Results
A total of 1918 patients with stroke were registered in the six populations—1673 with CI and 245 with primary ICH. Sociodemographic and case-mix variables are described in table 1. The average age was 71.7 years (SD=13.2), and 51.7% were female. There were significant differences between centres for all sociodemographic and case-mix variables.
Table 2 illustrates the variations in the processes of care between populations. Admission to hospital rates were over 93% in all centres. The overall proportion of patients spending ≥50% of their hospital stay in a stroke unit was low at 30.7%, and ranged between none and 65.4%.
Antiplatelet therapy use in the acute phase ranged between 30% in Menorca and 77.1% in Warsaw. Use of anticoagulation as an acute treatment was high overall but ranged from 10.6% in London to 83.7% in Dijon. Thrombolysis rates were generally low—1.5% average—ranging from none in Menorca and Kaunas to 6.8% in Warsaw.
Looking in more detail at management by the MDT, the proportion of patients being assessed by a specialist stroke physician ranged from 7.1% in Menorca to 96.6% in Warsaw. The proportion having seen a specialist stroke physician and therapists (PT, OT, SALT), regardless of need, ranged from none in Menorca to 31% in London.
Table 3 illustrates the variations in the range of processes of care for stroke, including, having access to a MDT, and a combined item (access to MDT and spending more than 50% of hospital stay in a stroke unit). The unadjusted estimates show significant variations (p<0.001) between populations for all processes of care examined. Adjustment for case mix and other confounders reduced the variations but these remained significant. Significant associations between care indicators and case-mix variables were also found. For example, patients with more severe stroke were more likely to arrive at hospital within 4 h of symptoms and significantly more likely to spend a higher proportion of their inpatient stay in a stroke unit. Patients with ICH were more likely to arrive at hospital within 4 h of symptoms and less likely to spend a longer time in a stroke unit than patients with CI. Patients over the age of 84 were less likely to spend a higher proportion of their inpatient stay in a stroke unit.
Table 4 presents an analysis across the whole cohort, showing decreasing proportions of deaths as the OCI level increases. The trend was significant for the early outcomes at 7 days, 30 days and 3 months after stroke (p<0.01) but no differences were found for the 1-year outcome. The Kaplan–Meier curves show higher probabilities of survival at higher levels of care. For the 1-year outcome, the curves for OCI=1 and OCI=2 exchanged position at an early stage but both were lower than the curve for OCI=3 (figure 1). Adjustment for confounders resulted in a sharper drop in the probability of survival but a similar pattern was maintained.
Discussion
This six-population study in the mid-2000s has identified significant variations in both evidence-based and non-evidence-based acute stroke care interventions. These variations remain for the most part after adjustments for case mix and other potential confounders.
This study aimed at reporting in more detail than previously the case mix and process of care in populations. It identified significant variations in evidence-based care between populations and also showed positive associations between organised stroke care and survival.
Previous reports on variations in stroke care in Europe have either relied on analyses of trials data28 acute thrombolysis data29 ,30 or selected hospital populations, but although they all indicated variations unexplained by case mix, they did not deal with the components of care or case mix in detail. This study has attempted to remedy deficiencies in previous studies by using less biased population-based samples, more detailed assessment of case mix and acute care processes and longer-term follow-up.15 ,16 A weakness in the recording of the acute process of care is that although the EROS data collection were standardised and fieldworkers from all centres were trained at one centre (London), there was still scope for local interpretation of variables, as in any study. This is particularly so for classification of ‘type of ward/bed’, ‘type of healthcare professional’ and ‘hospital length of stay’, which vary considerably across Europe.5 We validated these by site visits and would argue that use of the EROS Quality Assessment Tool score, limited the misclassification of processes in these areas.19
The variations in care bring into question what barriers to implementing evidence-based care exist across and within countries, particularly where high-level evidence supports their implementation. We are examining this topic in a follow-up European project. Although the populations are not necessarily representative of their countries and the results relate only to the centres involved, the methodological advantages of the improved design and detailed documentation undertaken in this study make this approach feasible for describing variations in stroke care and the impact on outcome.
The study was undertaken 15 years after initial systematic review evidence showed that stroke units are effective.31 Only 30.7% of patients spent more than 50% of their hospital stay in a stroke unit and just over a third were admitted to hospital within 4 h of their stroke. Variations in admissions to stroke units were wide and significant with none in the island of Menorca, only 26% in Kaunas and 100% in Dijon. Variations in the proportion of hospital stay that was spent in a stroke unit were between 16% in a suburb of Florence, Sesto Fiorentino, to 65.4% in London and adjustment for case mix showed significantly increased odds of admission to a stroke unit in certain populations. The drivers for this variation—local health system policy, guideline implementation, audit and patient expectation—require further investigation. For example, are financial incentives in place for stroke unit admission and how is the evidence of effectiveness of stroke unit care diffused in different countries? What are the reasons behind variations in delay in admissions that do or do not influence treatment with thrombolysis?
Similar variations and odds of delivery of other evidence-based interventions were identified—in particular, being seen by a stroke specialist and by members of MDT, regardless of need, with patients in London being most likely to receive MDT care after adjustment for case mix. However, overall only 1 in 10 patients had access to all members of the specialist team for stroke care. As the evidence-based guidelines used in these countries and internationally7–10 advocate multidisciplinary care, this level of evidence uptake is of particular concern. There are assumptions and weaknesses in our classification of assessment and treatment by members of the MDT. Although treatment is likely to be determined by need, assessment might be a routine procedure for identifying need. It is difficult to distinguish between the two, with the exception of those where such distinction is more clear—for example, the use of antiplatelets for patients with ischaemic stroke, being seen by a SALT if the swallow test failed. It is worth emphasising that this investigation was not intended to deal with appropriateness of different interventions for various subgroups because of the methodological limitations of such an approach, particularly for smaller groups. Instead the study wished to examine variations in delivery of care after adjustment for severity and case mix, assuming that such an adjustment would correct for need at least partially, though recognising that full adjustment is not possible.
Although at site visits for the development of the EROS Quality Assessment Tool score we could classify uniformly what centres meant by types of physician, nurse and therapist, this study used these variables as indicators of quality.23 We did not quantify the seniority and input of the team across centres, although a small observational study by Wellwood et al32 did quantify the inputs to stroke unit care across centres. In a detailed European study of physiotherapy and occupational therapy inputs into stroke care, De Wit et al33 showed that the content of the two disciplines was consistent between the four centres studied, with clearly demarcated roles.
Examination of the use of drug in acute care in the mid-2000s showed that thrombolysis rates were extremely low and presumably mirrored by the low access to specialist stroke care.28 ,29 However, the use of thrombolysis for stroke was only licensed in 2003. Of interest, was the use of treatments for which there was no evidence in the guidelines. Overall, 5% of patients were given drugs to raise blood pressure acutely, but this level was as high as 17.7% in Dijon. Acute lowering of blood pressure was recorded in 40% overall, but as high as 67% in Lithuania. The use of anticoagulants in the acute phase in patients without atrial fibrillation was of particular interest and was as high at 1 in 3 and particularly high in Dijon. It is difficult to establish with such high rates of acute anticoagulation, despite the evidence that this should not be prescribed,11 the validity of the responses. Further analyses of whether anticoagulants were prescribed for ‘acute treatment’, deep venous thrombosis or for other reasons should be investigated, but the high rates are not justified by the evidence. Despite evidence that neuroprotective agents are not effective, 7% of all patients received an agent, and neuroprotection was particularly high in Kaunas (14.7%). Of the evidence-based, cost effective drugs overall, 71% of patients were appropriately prescribed an antiplatelet during the hospital admission, but the reasons for not prescribing this cheap drug remain unclear.
The OCI developed by Saposnik and colleagues7 was used to investigate survival, and similarly to findings in Canada in the mid-2000s, this European study shows clearly that receipt of more elements of evidence-based stroke unit care is associated with better survival. Nonetheless, these data need to be interpreted with caution because this was an observational study and because factors used to differentiate patients as having good or poor outcomes, might have been unobservable, immeasurable or unknown. In addition, these curves were drawn for all populations combined, hence, would not reflect features of individual centres. The OCI scale makes no assumptions about weights for each component of the score and the simple additive score can at best seen as an indicator of quality, and the nature of the relationship between scores requires cautious interpretation.
The reduction in the survival probability among those with low levels of care despite the exclusion of deaths within the first 72 h might be explained by early deaths of patients; some of those patients would have been admitted to intensive care units but possibly had not had the chance of admission to stroke units, or access to other care elements. Stroke unit care is a ‘black box’ of several processes that should include assessment by members of the MDT, which are two separate components of the score. The EROS definition of MDT also did not include elements advocated in some guidelines—for example, health psychologist and social worker, as these services were not available at certain centres.10 However, the association between more elements of evidence-based care and improved outcome demonstrates to policy makers and clinicians that implementation of evidence-based care is associated with saving lives in a variety of European settings.
This six-population study highlights the variations and inequalities in acute care for patients with stoke. The implementation of evidence-based interventions is suboptimal and ways of understanding how to improve the dissemination of research evidence in different healthcare settings is a priority.
Acknowledgments
We thank all the patients and their families and the healthcare professionals at the different centres. We acknowledge the following: Sesto Fiorentino: M Lamassa (collection of data), P Nencini (collection of data), A Poggesi (collection of data), F Pescini (collection of data), A Cramaro (collection of data), E Magnani; I Romani (collection of data) (Department of Neurological and Psychiatric Sciences, University of Florence); M Baldereschi (Institute of Neurosciences, Italian National Research Council, Florence, collection of data); Kaunas: D Sopagiene (collection of data), D Kranciukaite (collection of data),(Institute of Cardiology c/o Kaunas University of Medicine, Kaunas); Menorca: J Rodriguez-Mera (Area de Salud de Menorca, ib-salut, Menorca, collection of data); Warsaw: M Głuszkiewicz (2nd Department of Neurology, Institute of Psychiatry and Neurology, Warsaw, collection of data); J Pniewski (Neurology Department, Medical Research Centre, Polish Academy of Sciences/CSK MSWiA, Warsaw, collection of data). Data centre: V Moltchanov (National Public Health Institute, Helsinki, Finland, technical assistance).
References
Footnotes
-
Collaborators M Giroud (Stroke Registry of Dijon, University of Burgundy, University Hospital of Dijon, local PI); Sesto Fiorentino: D Inzitari (Department of Neurological and Psychiatric Sciences, University of Florence; Institute of Neurosciences, Italian National Research Council, Florence, local PI); Menorca: M Torrent (Area de Salud de Menorca, ib-salut, Menorca, local PI); Warsaw: H Sienkiewicz-Jarosz (1st Department of Neurology, Institute of Psychiatry and Neurology, Warsaw, local PI); A, Czlonkowska (2nd Department of Neurology, Institute of Psychiatry and Neurology, Warsaw, local PI). Data centre: C Sarti (National Public Health Institute, Helsinki, Finland, local PI).
-
Contributors CDAW and AGR, conceived and designed the study and obtained funding. They drafted the first version of the paper with equal contributions from IW and AB. BC managed the data for the study, contributed to the data interpretation, revision and critical review of the paper. SAA carried out the statistical analysis and interpretation of results and contributed to writing of the second draft. All coauthors contributed to the interpretation of data, critically revised the paper for important intellectual content. All authors saw and approved the final version of the manuscript.
-
Funding The work was supported by the European Union Fifth Framework Programme and was partially supported by the National Institute for Health Research (NIHR) Biomedical Research Centre at Guy's and St Thomas’ NHS Foundation Trust and King's College London, the Stanley Thomas Johnson Foundation, the Stroke Association and the NIHR Programme grant funding (RP-PG-0407-10184) and DH HQIP funding. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.
-
Competing interests None.
-
Patient consent Obtained.
-
Ethics approval The study was ethically approved by ethics committees of each of the centres.
-
Provenance and peer review Not commissioned; externally peer reviewed.