Published online before print August 31, 2006, doi: 10.1161/01.STR.0000240508.28625.2c
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Stroke. 2006;37:2579.)
© 2006 American Heart Association, Inc.
Original Contributions |
Population-Based Study of Determinants of Initial Secondary Care Costs of Acute Stroke in the United Kingdom
From the Health Economics Research Centre (R.L.-F., A.M.G.), Department of Public Health and Stroke Prevention Research Unit (P.M.R.), Department of Clinical Neurology, University of Oxford, Oxford, UK.
Correspondence to Prof Peter M. Rothwell, Stroke Prevention Research Unit, University Department of Clinical Neurology, Radcliffe Infirmary, Oxford OX2 6HE, UK. E-mail peter.rothwell{at}clneuro.ox.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Background and Purpose— To determine the cost-effectiveness of specific interventions to prevent or treat acute stroke, it is necessary to know the costs of stroke according to patient characteristics and stroke subtype and etiology. However, very few such data are available and none from population-based studies. We determined the predictors of resource use and acute care costs of stroke using data from a population-based study.
Methods— Data were obtained from the Oxford Vascular study, a population-based cohort of all individuals in nine general practices in Oxfordshire, UK, which identified 346 patients with a first or recurrent stroke during April 1, 2002, to March 31, 2004. Univariate and multivariate analyses were performed to identify the main predictors of resource use and costs.
Results— Acute care costs ranged from £326 (lower decile) to £19 901 (upper decile). There were multiple important univariate interrelations of patient characteristics, stroke subtype, and stroke etiology with hospital admission, length of stay, and 30-day case-fatality. For example, patients with primary intracerebral hemorrhage were more likely to be admitted than patients with partial anterior circulation ischemic stroke and less likely to survive without disability, but length of stay was reduced as a result of high early case-fatality such that cost was substantially less. However, the majority of univariate predictors of resource use, cost, and outcome were confounded by initial stroke severity as measured by the National Institutes of Health Stroke Scale score, which accounted for approximately half of the predicted variance in cost. Cost increased approximately linearly up to an National Institutes of Health Stroke Scale score of 18 and then fell steeply at higher scores as a result of rising early case-fatality.
Conclusions— Several patient and event-related characteristics explained the wide range of initial secondary care costs of acute stroke, but stroke severity was by far the most important independent predictor.
Key Words: costs and cost analysis • outcome • stroke
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Stroke has recently been shown to cost the UK healthcare system £5.2 billion.1 Other studies have also estimated the average cost per stroke in the United Kingdom,2–6 and these costs are often used to estimate the cost-effectiveness of specific interventions to prevent or treat stroke. However, cost varies substantially between individuals and is likely to depend on the pathologic subtype of stroke,7 the particular etiology, and other patient-related characteristics such as age, sex, and comorbidity,8,9 all of which are likely to be relevant to estimates of cost-effectiveness of specific interventions. For instance, the cost-effectiveness of anticoagulation and carotid endarterectomy will be influenced by both the characteristics of patients in whom the treatments are used and the type or severity of strokes that are caused and prevented. More detailed data on the drivers of cost are therefore required.
To reliably determine the predictors of stroke resource use and costs, population-based studies with full case ascertainment (ie including minor strokes not admitted to the hospital and strokes resulting in death before, or soon after, hospital admission) are ideally required.10 We therefore studied hospitalization, subsequent length of stay, and acute care costs during the 12 months after any first incident or recurrent stroke in a population-based study in relation to baseline patient characteristics, comorbidity, premorbid handicap, pathologic subtype, etiologic factors, and severity of the neurologic deficit at first assessment.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Study Population
The Oxford Vascular Study (OXVASC) study population comprised 91 106 individuals registered with 63 family physicians in nine Oxfordshire practices. The population was 94% white, higher than the UK average (92%), but had a similar age–sex structure to that of the United Kingdom.11 Although the electoral wards containing the nine practices were less deprived than the rest of England,11 two practices were based in wards ranked in the lower third nationally in terms of deprivation. Registration of patients into the study began on April 1, 2002, and this analysis includes all cases ascertained with stroke occurring before March 31, 2004. Comprehensive overlapping methods of stroke ascertainment were used11 with direct assessments of ascertainment suggesting that it was near complete.12
Patients were assessed by a study clinician as soon as possible after stroke (median delay, 2 days; interquartile range, 1–4) in the hospital, in a dedicated daily clinic, or at home. Diagnoses of stroke were verified by CT or MRI with the rate of brain imaging or autopsy in OXVASC being 97%.12 Neurologic impairment was measured using the National Institutes of Health Stroke Scale (NIHSS)13 to determine stroke severity at baseline. Clinical subtype of ischemic stroke was categorized with the Oxford Community Stroke Project (OCSP) classification.14 Stenosis of the symptomatic carotid artery was measured with Duplex ultrasound of the carotid bifurcation and categorized as 0% to 49%, 50% to 99%, and occlusion using scales based on the North American Symptomatic Carotid Endarterectomy Trial method of measurement of degree of stenosis.15 Stroke severity at 1 month was classified according to handicap as measured by Rankin score,16 and for the purposes of this analysis, a nonfatal disabling stroke was classified as 1-month Rankin score 3 to 5.
Resource Use and Costs
OXVASC provided information on resource use. Use of diagnostic tests (which included multiple imaging in individual patients) was collected prospectively or retrieved retrospectively from case notes, if required. Unit costs were derived from national reference costs,17 including CT (£67), MRI (£313), electrocardiogram (£25), echocardiogram (Echo) (£59), and carotid Doppler (£99). We assumed that those patients not requiring hospitalization or being hospitalized more than 48 hours after initial stroke visited their general practitioner on the day of the stroke. Unit costs for general practitioner (£24) and outpatient (£111) visits and emergency ambulance services (£237) were derived from Netten et al.18 Hospital stay costs were derived and attached to each day spent in each of the following wards: general and long-term care (£269),4 stroke unit (£331),4 and rehabilitation ward (£213).17 Initial assessment at the hospital was determined to be an emergency visit (£112).17 The perspective of the UK National Health Service (NHS) was adopted in the study. All costs were standardized to 2004–2005 prices by use of the NHS hospital and community health services inflation index.18
Statistical Analysis
Outcome measures included the proportion of patients admitted to hospital, subsequent length of stay, and total acute care costs. We also estimated the proportion of patients with substantial disability (as measured with Rankin score 3–5 at 30 days) and death within 30 days after stroke onset. Resource use and total costs were reported as means. To account for the skewed nature of resource use and cost data, 95% CIs were calculated from 1000 bootstrap estimates.19 Categorical outcomes are reported as proportions and exact 95% CIs computed.
Follow-up data up to March 31, 2004, were used; therefore, for those patients still in the hospital, it was not possible to determine when they were discharged or if they were still alive after this date. We therefore examined the effect of censoring on our cost results using the method developed by Bang and Tsiatis.20 This method partitions the study period into smaller time periods within each of which the total cost incurred for all patients alive at the beginning of the period is calculated. The estimated costs of patients with complete costs for each time period were weighted by the Kaplan Meier sample average estimator using reverse censoring (ie the probability of not being censored at the beginning of each period), which were then summed over all periods and divided by the total number of patients to obtain an estimate of the mean total study cost. In our study, we partitioned periods by days. 95% CIs were reported around the mean censored adjusted costs using 1000 bootstrap estimates.
All predictor variables were first examined by means of univariate analysis to assess the importance of each on outcomes. For all continuous variables (NIHSS score, age, and degree of carotid stenosis), appropriate cutoff points were used to stratify patients to detect nonlinear relations. Subsequent analysis of any linear associations was done using the continuous variables as appropriate. All comparisons for categorical variables between groups were performed using 
2 tests. Because length of stay in hospital and total costs were not normally distributed, all comparisons between groups were performed using Mann-Whitney U or the Kruskal-Wallis test to examine the differences between/among stratified groups. Statistical significance was set at P<0.05.
Multiple regression analyses were then performed. Only age was initially entered as a continuous variable, whereas NIHSS score and degree of carotid stenosis were stratified into the same groups as those used in the univariate analysis. To determine the predictors of case-fatality and disability at 1 month, logistic regressions were used. To assess the main predictors of admission and length of stay, a two-part model was used. A logistic regression model was used to assess the predictors of hospital admission and, conditional on admission, the length of hospital stay was logarithmically transformed to normalize the distribution and regressed over the same variables.
Like with length of stay, total costs were logarithmically transformed. We then assessed the significance (F-test) and effect of each predictor variable on log total costs adjusting by stroke severity (NIHSS score). Second, the log total costs were regressed using White’s robust standard errors over all predictor variables. Because we found a nonlinear relationship between log total costs and NIHSS score, we decided to perform a further regression model combining all the effects of predictor variables but including NIHSS score as a continuous variable and also including the quadratic term of the initial NIHSS score.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
A total of 346 patients had either a first-ever or recurrent stroke during the study period (280 ischemic strokes, 21 primary intracerebral hemorrhages [PICH], 17 subarachnoid hemorrhages [SAH], 28 undetermined pathology). Patients were followed up for a minimum of 1 month and a maximum of 2 years (mean, 394 days; SD, 209). Baseline characteristics and risk factors are reported in Table 1. NIHSS scores at baseline were available in 333 (96%) patients. Of the 13 patients without an NIHSS score, 11 died shortly after stroke onset.
|
A total of 42 (12%) patients had a recurrent stroke during the study period, and a further six (2%) had two recurrences. The mean time between the first stroke in the study period and the second and third recurrent strokes were 54 days (SD, 93) and 214 days (SD, 122), respectively. Our aim was to report only the costs associated with the first stroke in the study period, but for those patients already hospitalized when a second stroke occurred, it was not possible to separately attribute costs to the multiple strokes and so all costs were combined together as part of the initial event.
A total of 215 (62%; 95% CI, 57%–67%) patients were admitted to the hospital with 131 (38%) being managed in the community. Of those patients admitted to the hospital, 31% (25%–37%) were admitted to a stroke unit and the remainder managed on general medical wards. Forty-one percent (34%–48%) were subsequently transferred to a rehabilitation unit. There were significant differences in rates of hospital admission according to stroke severity, degree of carotid stenosis, presence of atrial fibrillation, and stroke subtype groups (Table 2). Patients with a total anterior circulation ischemic stroke (TACI) (93%; 78%–99%), PICH (95%; 76%–100%) and SAH (88%; 64%–99%) were the most likely to be admitted. Logistic regression showed that significant independent predictors of hospital admission were initial stroke severity (NIHSS score, P<0.001), PICH (P=0.022), and SAH (P=0.028). Patients with a history of atrial fibrillation tended to have higher hospital admission rates than those without (P=0.085) as did patients with a history of hypertension (P=0.096).
|
Conditional on admission, mean length of stay was 41 days (95% CI, 35–48 days). The majority of bed-days were spent in the rehabilitation ward (53%), whereas 27% were spent in the general wards and 12% in a stroke unit. There were significant differences in length of stay between groups in terms of stroke subtype, stroke severity, and degree of carotid stenosis (Table 2). Patients with partial anterior circulation ischemic stroke (PACI) and TACI had the longest length of stay conditional on hospital admission (56 days; 95% CI, 43–71 days and 44 days; 95% CI, 28–65 days, respectively). However, after adjusting for all variables, the only significant predictor of length of stay was initial stroke severity (NIHSS score). The association with NIHSS score was nonlinear and was best modeled as a quadratic expression (P<0.001). Although carotid occlusion did not predict length of stay in the hospital, such patients tended to spend more days in hospital than those with 0 to 49% stenosis: 117 versus 37 days (P=0.066).
The acute care cost per patient was £6607 (95% CI, £5597–7882) (Table 3). Inpatient rehabilitation was the major cost component, accounting for 44% of costs. Initial investigations represented 2.6% of costs, with initial general practitioner and outpatient assessment and ambulance services representing 3% of costs. Adjusting for censoring increased the mean cost per patient to £6906 (95% CI, £5707–8109), an increase of 4.5% when compared with mean unadjusted costs (Table 3).
|
As shown in the Figure, there was considerable variation between patients in the total acute care cost of stroke (£326 in lower decile versus £19 901 in upper decile), mainly as a result of lower costs in patients with minor stroke, who were often not admitted to the hospital, and the wide range of severity of stroke, and hence length of stay, in patients who were admitted. Patients who were hospitalized incurred acute care costs of £10 474 (95% CI, £8891–12 251), which was significantly more than for those who were not hospitalized (£338; 95% CI, £327–350). As also shown in Table 3, such differences between hospitalized and nonhospitalized patients remain when stratifying by initial severity levels (ie NIHSS score). It should be noted, however, that the eight nonhospitalized patients with NIHSS scores over >10 or not determined died on the same day as that of their stroke.
|
Univariate analysis showed that patients with a history of atrial fibrillation incurred significantly higher costs (£9667 versus £5824, P<0.001). There were also significant differences in total costs according to stroke severity, degree of carotid stenosis, and stroke subtype (Table 4). By OCSP classification, patients with TACI incurred the highest mean costs (£10 782; 95% CI, £6549–16 429), whereas patients with a lacunar infarct incurred the lowest mean costs (£3426; 95% CI, £2004–5272). However, after adjusting for stroke severity, the only variable that significantly predicted total costs was degree of carotid stenosis (Table 4). Similar results were found when adjusting for all baseline variables with carotid occlusion and stroke severity being the only significant a priori predictors of cost (Table 4).
|
Like with length of stay, NIHSS score between 3 and 20 was the main cost predictor (Table 4). Patients with NIHSS scores of 3 to 10 incurred mean costs of £9284 (95% CI, £6897–12 547) and those with scores of 11 to 20 incurred mean costs of £13 694 (95% CI, £10 649–17 257). When NIHSS score was included as a continuous variable together with its quadratic form, log total cost increased linearly up to a score of 18 (P<0.0001) and then fell at higher scores (P<0.0001).
The nonlinear association between NIHSS and costs is explained by the increase in admission, length of stay, and survival with disability (Table 5) with increasing scores in the range of 0 to 20 and the sharp increase in case-fatality at scores above 20 (Table 5) with the corresponding reduction in length of stay and longer-term hospital rehabilitation. As a consequence, although only 8% of patients with PICH had no disability at 1 month (Table 5), costs were less than those for PACI strokes because of the higher early case-fatality and short mean length of stay in patients with PICH.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Ours is the first population-based study to estimate the predictors of hospital admission, length of stay, and acute care costs of first or recurrent stroke in relation to baseline characteristics, stroke subtype, etiology, and measures of initial stroke severity. Most recent studies estimating and predicting the acute costs of stroke have either been based on data derived from trials with strict inclusion criteria or from hospitalized series, usually including only patients with ischemic strokes and excluding patients with less severe strokes.8,9 Using trial data, which excluded less severe strokes (NIHSS<7), Caro and colleagues9 concluded that severity of stroke was the strongest predictor of treatment costs after acute ischemic stroke. In a hospital-based study, Diringer and colleagues8 found that the main predictors of costs included NIHSS score, heparin treatment, presence of atrial fibrillation, female sex, and presence of ischemic heart disease. The only previous population-based study of the cost of stroke was done in Melbourne, Australia, which has a different healthcare system with higher rates of hospital admission and only looked at cost in relation to OCSP subtype of stroke and stroke recurrence.7
In the United Kingdom, a large proportion of patients with minor stroke are managed in the community and investigated only in outpatient clinics. The high levels of ascertainment of such cases in OXVASC12 allowed us to determine the proportion of patients not admitted to hospital reliably and to study the predictors of hospital admission in the United Kingdom. We were also able to estimate the acute care costs for patients treated in the community and in the hospital and found that, not surprisingly, acute care costs were significantly higher for those treated in the hospital (£10 474 versus £338), which was true for each level of initial stroke severity. Differences in hospitalization rates, and therefore costs, may be explained by nonclinical factors such as patients’ living arrangements, distance to general practitioner, or other social characteristics. Furthermore, our study was an observational study, which makes comparisons between hospitalized and nonhospitalized patients difficult to interpret.
Our study showed that there was very considerable variation between patients in the total acute care cost of stroke. As a result, reporting the average cost of stroke, without taking into account its severity or subtype, may be meaningless when assessing the cost-effectiveness of stroke prevention or treatment interventions. For example, we found that strokes in patients with atrial fibrillation incurred more costs than those who did not as they tended to experience more severe strokes. Therefore, if we were to use our results to assess the cost-effectiveness of stroke prevention in patients with atrial fibrillation, using the average cost of a stroke would bias the results against prevention as the savings generated from preventing strokes in such patients would be higher than those for preventing the "average" stroke.
Our study showed that the main baseline determinant of acute care resource use and costs of stroke was the severity of the neurologic deficit (NIHSS score) at initial assessment. Although severity of carotid disease was also an independent predictor of cost, interpretation is difficult as a result of the nonimaging of patients with severe ischemic stroke either as a result of death soon after admission or on the basis that the patient was not a candidate for endarterectomy, usually because their stroke was too severe. The apparently low costs of stroke in patients with symptomatic carotid stenosis are therefore likely to be at least partly an artifact.
Previous studies, all of which have used hospitalized patients, have assessed stroke treatment costs in the United Kingdom over a 12-month period. These have ranged from approximately £6800 to £11 450.2–5 Another study, which used modeling, estimated the 5-year cost to the NHS of a stroke case to be £15 306 with 59% of these costs attributable to acute hospitalization.6 Others have estimated the costs of stroke using a prevalence approach (ie all costs attributable to stroke during a given time period and regardless of disease onset).1,21,22 For example, in 2004, stroke was found to cost the UK economy £8 billion, including healthcare, productivity, and informal care costs, of which, £4.6 billion (58%) were incurred by the NHS.1
Our study had some limitations. First, we did not include the costs of neurosurgery or intensive care, which might underestimate the costs of SAH; like in the United Kingdom, these are rarely used for ischemic stroke or PICH. Second, we did not include the costs of ongoing care after the acute phase such as subsequent carotid endarterectomy or hospital readmission resulting from late complications of the stroke. Although acute treatment costs are a major cost component, subsequent outpatient and community care also account for a substantial proportion of costs.6 As a result of the NHS perspective adopted, our analysis did not include indirect costs such as productivity losses resulting from early mortality or absence from work. Third, we did not include the costs of any recurrent stroke after discharge. Because the risk of recurrence is highest after a minor stroke or stroke resulting from carotid stenosis,23,24 including the costs of recurrence may affect the estimates of the predictors. Third, our overall costs are, of course, only applicable to the United Kingdom and possibly to similar healthcare systems in which a high proportion of patients with minor stroke are investigated and treated in the outpatient setting. However, our predictors of cost are more likely to be generalizable to other healthcare settings.
In summary, our study reports up-to-date estimates of the predictors of acute care resource use associated with stroke in the United Kingdom using data from a well-conducted population-based study. Our results highlight the importance of initial stroke severity as a predictor of resource use and cost and should be of use to analysts assessing the burden of stroke and the cost-effectiveness of interventions for prevention of particular stroke subtypes.
|
Acknowledgments |
|---|
The comments from three anonymous reviewers are acknowledged.
Sources of Funding
This project was funded by the UK Medical Research Council, the UK Stroke Association, and partly by an unrestricted educational grant from Astra Zeneca, UK. Ramon Luengo-Fernandez is funded by a UK Department of Health Research and Development award.
Disclosures
None.
Received June 30, 2006; accepted July 13, 2006.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
- Luengo-Fernandez R, Leal J, Gray A, Petersen S, Rayner M. Cost of cardiovascular diseases in the United Kingdom. Heart. 2006; May 15 [Epub ahead of print].
- Beech R, Rudd AG, Tilling K, Wolfe CDA. Economic consequences of early inpatient discharge to community-based rehabilitation for stroke in an inner-London teaching hospital. Stroke. 1999; 30: 729–735.
[Abstract/Free Full Text] - Grieve R, Porsdal V, Hutton J, Wolfe C. A comparison of the cost-effectiveness of stroke care provided in London and Copenhagen. Int J Technol Assess Health Care. 2000; 16: 684–695.[CrossRef][Medline] [Order article via Infotrieve]
- Kalra L, Evans A, Perez I, Knapp M, Swift C, Donaldson N A randomised controlled comparison of alternative strategies in stroke care. Health Technol Assess. 2005; i-94.
- Forbes JF, Dennis MS. Costs and Health Outcomes of Stroke Patients: A Prospective Study. Final Project Report submitted to the Health Service and Public Health Research Committee; Edinburgh; 1995.
- Youman P, Wilson K, Harraf F, Kalra L. The economic burden of stroke in the United Kingdom. Pharmacoeconomics. 2003; 21: 43–50.[Medline] [Order article via Infotrieve]
- Dewey HM, Thrift AG, Mihalopoulos C, Carter R, Macdonell RAL, McNeil JJ, Donnan GA. Lifetime cost of stroke subtypes in Australia: findings from the North East Melbourne Stroke Incidence Study (NEMESIS). Stroke. 2003; 34: 2502–2507.
[Abstract/Free Full Text] - Diringer MN, Edwards DF, Mattson DT, Akins PT, Sheedy CW, Hsu C, Dromerick AW. Predictors of acute hospital costs for treatment of ischemic stroke in an academic center. Stroke. 1999; 30: 724–728.
[Abstract/Free Full Text] - Caro JJ, Huybrechts KF, Kelley HE. Predicting treatment costs after acute ischemic stroke on the basis of patient characteristics at presentation and early dysfunction. Stroke. 2001; 32: 100–106.
[Abstract/Free Full Text] - Feigin V, Hoorn SV. How to study stroke incidence. Lancet. 2004; 363: 1920.[CrossRef][Medline] [Order article via Infotrieve]
- Rothwell PM, Coull AJ, Giles MF, Howard SC, Silver LE, Bull LM, on behalf of the Oxford Vascular Study. Change in stroke incidence, mortality, case-fatality, severity, and risk factors in Oxfordshire, UK from 1981 to 2004 (Oxford Vascular Study). Lancet. 2004; 363: 1925–1933.[CrossRef][Medline] [Order article via Infotrieve]
- Coull AJ, Silver LE, Bull LM, Giles MF, Rothwell PM, on behalf of the Oxford Vascular Study. Direct assessment of completeness of ascertainment in a stroke incidence study. Stroke. 2004; 35: 2041–2045.
[Abstract/Free Full Text] - Brott T, Adams HP, Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J, Holleran R, Eberle R, Hertzberg V, Rorick M, Moomaw CJ, Walker M. Measurements of acute cerebral infarction: a clinical examination scale. Stroke. 1989; 20: 864–870.
[Abstract/Free Full Text] - Bamford J, Sandercock P, Dennis M, Burn J, Warlow C. Classification and natural history of clinically identifiable subtypes of cerebral infarction. Lancet. 1991; 337: 1521–1526.[CrossRef][Medline] [Order article via Infotrieve]
- Rothwell PM, Gibson RJ, Slattery J, Sellar RJ, Warlow CP. Equivalence of measurements of carotid stenosis. A comparison of three methods on 1001 angiograms. European Carotid Surgery Trialists’ Collaborative Group. Stroke. 1994; 25: 2435–2439.[Abstract]
- Van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJA, Van Gijn J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1988; 19: 604–607.
[Abstract/Free Full Text] - Department of Health. NHS Reference Costs 2005. Available at: www. dh.gov.uk/PolicyAndGuidance/OrganisationPolicy/FinanceAndPlanning/ NHSReferenceCosts/fs/en. Accessed April 15, 2006.
- Netten A, Curtis L. Unit Costs of Health and Social Care 2005. Canterbury, UK: Personal Social Services Research Unit; 2005.
- Briggs A, Gray A. The distribution of healthcare costs and their statistical analysis for economic evaluation. J Health Serv Res Policy. 1998; 3: 233–245.[Medline] [Order article via Infotrieve]
- Bang J, Tsiatis AA. Estimating medical costs with censored data. Biometrika. 2000; 87: 329–343.
[Abstract/Free Full Text] - Reducing Brain Damage: Faster Access to Better Stroke Care. National Audit Office, Department of Health; 2005.
- Payne K, Huybrechts K, Caro J, Green T, Klittich W. Long term cost-of-illness in stroke: an international review. Pharmacoeconomics. 2002; 20: 813–825.[CrossRef][Medline] [Order article via Infotrieve]
- Coull AJ, Rothwell PM. Under-estimation of the early risk of recurrence after first stroke by the use of restrictive definitions. Stroke. 2004; 35: 1925–1929.
[Abstract/Free Full Text] - Lovett JK, Coull AJ, Rothwell PM. Early risk of recurrence by subtype of ischemic stroke in population-based incidence studies. Neurology. 2004; 62: 569–573.
[Abstract/Free Full Text]
This article has been cited by other articles:
 |
|
 |
|
  W. Lo, K. Zamel, K. Ponnappa, A. Allen, D. Chisolm, M. Tang, B. Kerlin, and K. O. Yeates The Cost of Pediatric Stroke Care and Rehabilitation Stroke, January 1, 2008; 39(1): 161 - 165. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||