doi: 10.1161/01.STR.0000013663.27776.DB
(Stroke. 2002;33:1341.)
© 2002 American Heart Association, Inc.
Original Contributions |
Guideline Compliance Improves Stroke Outcome
A Preliminary Study in 4 Districts in the Italian Region of Lombardia
From the Cerebrovascular Unit, C. Mondino Foundation (G.M., A.C.), and Department of Computer Science and Systems, University of Pavia (S.Q.), Pavia, Italy.
Reprint requests to Giuseppe Micieli, MD, Cerebrovascular Unit, IRCCS C. Mondino, 27100 Pavia, Italy. E-mail giuseppe.micieli{at}mondino.it
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionAppendix 1Appendix 2References |
|---|
Background and Purpose— Guidelines for medical practice in stroke have been proposed in different countries, but their impact on stroke outcome has not been verified to date. The aim of this study was to evaluate the impact of the American Heart Association guidelines for acute stroke and for transient ischemic attack on first-ever stroke patients.
Methods— Three hundred eighty-six first-ever ischemic stroke patients were admitted to the study. Those observed within 6 hours from stroke onset were eligible for the acute clinical phase of the study, while all were admitted to the early clinical phase. The follow-up lasted 6 months. Primary end points were survival and the effectiveness of treatment on disability, measured as the proportion of potential improvement in the Barthel Index score achieved during treatment. A rating of noncompliance with the guideline recommendations was calculated for each patient, and its association with the end points was investigated. The Kaplan-Meier method and log-rank test were used to estimate and compare survival curves between groups; Cox proportional hazards model and logistic regression were used to identify risk factors for mortality; and correlation tests and regression analysis were used to evaluate the influence of guideline compliance on disability. Both univariate and multivariate statistical analyses were performed.
Results— Survival and treatment effectiveness were directly correlated with guideline compliance. The relative risk of death for patients with a noncompliance rating 
5 was 2.26 with respect to patients with a noncompliance rating <5 (95% CI, 1.51 to 4.67; P<0.0007). In this latter group, at 6 months we detected a 15% decrease in mortality (95% CI, 9.1% to 17.5%). Treatment effectiveness showed a Spearman’s rank correlation with the noncompliance rating of -0.3 (P<0.001). At discharge we observed a 13% increase in treatment effectiveness, while no significant differences were detectable at 3 and 6 months. These associations were confirmed by the multivariate analysis, in which we included, together with the noncompliance rating, all the variables previously identified as independent predictors of mortality and disability.
Conclusions— This study demonstrates an association between adherence to guidelines and stroke outcome, and it can be viewed as a study that prepares the way for a randomized controlled trial in this area. It also emphasizes the need to develop personnel and structures devoted to stroke care because an evidence-based clinical approach could significantly reduce the risk of death.
Key Words: disability evaluation • mortality • practice guidelines • stroke
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1Appendix 2References |
|---|
Stroke is the second leading cause of death and the leading cause of chronic disability in industrialized countries. It is thus one of the most pressing medical and economic problems.1–3 Guidelines for medical practice in stroke have been proposed in different countries with the aim of improving the management of patients and hence reducing direct and indirect costs, but the impact of their application on mortality, disability, and costs has not yet been investigated.
In 1994 the Ad Hoc Committee of the Stroke Council of the American Heart Association (AHA) proposed guidelines for the management of patients presenting with acute ischemic stroke4 and with transient ischemic attacks (TIAs).5 These provide recommendations based on available data from clinical trials and graded on the strength of scientific evidence. While the guidelines for acute ischemic stroke include recommendations for initial care, no recommendation is given for medical and/or surgical measures to prevent stroke recurrence. In contrast, the TIA management guidelines provide recommendations for diagnostic procedures and specific treatment to minimize the risk of stroke.
We sought to evaluate, in 4 districts in the northern Italian region of Lombardia, associations between guideline compliance and survival and disability in patients with first-ever ischemic stroke and to identify which guideline-recommended actions appear to be strongly related to survival. We also analyzed the applicability of these guidelines in an Italian healthcare setting to plan a multicenter, randomized national study.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1Appendix 2References |
|---|
We performed a prospective, observational, multicenter study. We enrolled 386 consecutive subjects with a confirmed diagnosis of first-ever ischemic stroke admitted to the 4 centers participating in the study (the neurological departments of the hospitals of Pavia, Bergamo, Brescia, and Monza) between February 1998 and April 1999. Although the neurologists and nurses at the 4 centers were trained in acute stroke care, at the time of the study these centers were not organized specifically as stroke units (ie, they did not have homogeneous guidelines covering all the aspects of acute stroke management and rehabilitation, and monitoring of physiological parameters was not always available).6 To minimize variability across centers and between observers, a user manual was produced, and meetings of the study group were held every 3 months to discuss the collection, interpretation, and quality of data. Moreover, the AHA guidelines were made available online through a computerized system.
Clinical first-ever ischemic stroke was defined as an acute focal neurological deficit lasting >24 hours with no cause other than cerebrovascular disease in patients who have not had a previous stroke. On the basis of the clinical features of the initial neurological examination, the patients were grouped, according to the clinical criteria proposed by the Oxfordshire Community Stroke Project,7 into the following subtypes: total anterior circulation infarct (TACI); partial anterior circulation infarct (PACI); lacunar infarct (LACI); and posterior circulation infarct (POCI). Clinical diagnosis was confirmed by CT or MRI examination. The first 6 hours after the onset of symptoms was defined as the acute stroke phase, and the subsequent period, up to 7 days after stroke, was considered the early clinical phase.
Written informed consent and details of personal history were obtained from the patients or, if their level of consciousness and/or neurological deficit precluded this, from relatives.
The following data were collected: (1) baseline characteristics: age, sex, living conditions, drug use before stroke; (2) vascular risk factors and comorbid conditions: hypertension, atrial fibrillation, previous myocardial infarction, cardiac insufficiency, TIAs, diabetes mellitus, smoking (current or past), alcohol consumption, hypercholesterolemia; (3) clinical condition at admission: level of consciousness (Glasgow Coma Scale8), neurological deficit (National Institutes of Health Stroke Scale9), and level of disability (Barthel Index [BI]10); and (4) diagnostic tests, therapeutic interventions, amount of inpatient rehabilitation, and, of great importance in our study, compliance of these areas with guideline recommendations.
Outcome data were collected at discharge and at 3 and 6 months after stroke onset; they included information on vital status, recurrence of stroke, and disability measured by means of the BI, with a score ranging from 0 to 20. These assessments were made through a direct interview. In the event of a patient’s death, the date and cause of death were recorded from information given by relatives or general practitioners.
As suggested elsewhere,11,12 the parameter treatment effectiveness (TE) was used as a measure of residual disability. TE at discharge reflects the proportion of potential improvement achieved during hospitalization; it was calculated according to the following formula: 100x (discharge BI score-initial BI score)/(BI maximum score-BI initial score). Similarly, TE was calculated at 3 and 6 months after the stroke onset. According to the formula, TE was 100% when a patient achieved the BI maximum score (ie, 20).
Three hundred seventy-six (97.4%) and 368 patients (95.3%) were evaluated at 3 and 6 months, respectively. Eighteen patients were thus lost during the follow-up.
Management guidelines for acute ischemic stroke were made available for the 110 patients (28.5%) observed within 6 hours of stroke onset, while diagnostic procedures and specific treatments aimed at minimizing stroke recurrence were derived from the guidelines for TIAs and made available for all the subjects. As mentioned, these guidelines were entered into a computerized system and rendered accessible online: first, they were represented as a flow chart with the use of a graphical editor13,14 that allows the guidelines to be set out, in varying detail, as a sequence of recommended actions. Users could browse through them both for educational purposes and to obtain real-time suggestions, since we integrated the guidelines with the electronic patient record. The latter, implemented by means of the MS-ACCESS relational database management system, was shared by the 4 study centers and stored all the aforementioned information. In addition, when physicians applied specific diagnostic and/or therapeutic strategies that were at variance with guideline indications, the database was also used to store information detailing the reasons for these decisions. We refer to this behavior as physician "noncompliance" with the guideline.
The detailed electronic patient record facilitated the calculation, for each patient, of a noncompliance rating (NCR) during his/her management: the NCR is initially set at 0 and is then incremented by 1 for each guideline recommendation not acted on. The NCR could range from 0 to 47 (acute phase, 0 to 17; early clinical phase, 0 to 23; management of medical and/or neurological complications, 0 to 7), but the maximum value we observed was 19. The actions recommended by the guidelines and considered in the calculation of the NCR (ie, all the recommendations supported by the highest degree of scientific evidence, as reported in the guidelines themselves) are listed in Appendix 1.
Statistical Analysis
Data analysis was performed with the use of the S-Plus statistical package from Mathsoft Inc.15
Median, range, and quartiles were used as descriptive statistics because of the nonnormal distribution of most variables. The nonparametric Wilcoxon test was used to compare numerical and ordinal variables in 2 patient groups. The Bonferroni method was used for correction of probability values in multiple comparisons.
The cumulative survival curves were obtained by the Kaplan-Meier method,16 and the difference between 2 survival curves was tested by the log-rank or Mantel-Haenszel test.17 The influence of prognostic factors on survival was investigated by means of the Cox proportional hazards regression model.18 Logarithmic transformation was used on variables showing skewed distribution before they were entered in the regression models. Both univariate and multivariate analyses were performed. The Harrel z test on Schoenfeld residuals was used to validate the proportional hazards assumption for each covariate.19 A threshold value was found for some of the numerical variables that were significant under the Cox regression model, and for illustrative purposes the Kaplan-Meier estimates for the 2 groups of patients distinguished by that threshold are shown. Relative risks (RRs) and their 95% CIs were derived, after fitting the Cox model, as the antilogarithm of the coefficient estimated for each covariate included in the model itself. Note that for numerical variables, such as age and NCR, the RR refers to a unit change in the variable itself. While the Cox model was used for long-term survival, a logistic regression analysis was performed for early death, with the patient status at discharge (alive/deceased) considered as the binary variable to be predicted.
For the analysis of functional outcome, Spearman’s rank correlation test was used to investigate the univariate relationship between the TE (the effect of treatment on the BI score) and the NCR, and a multivariate regression model was used to test whether the NCR retained its significance after correction for additional covariates.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1Appendix 2References |
|---|
Three hundred eighty-six patients (202 men and 184 women; median age, 72 years; interquartile range, 62 to 78 years) were enrolled in the study. Deaths numbered 26 at 7 days, 70 at 3 months, and 91 at 6 months. The number of subjects lost to follow-up was 0 at 7 days, 10 at 3 months, and 18 at 6 months. Clinical diagnosis was TACI in 10.7% of patients, PACI in 31.7%, LACI in 38.9%, and POCI in 18.7%. The case fatality rate for the entire population was 7% (95% CI, 4.4% to 9.5%) at 7 days, 18.6% (95% CI, 13.6% to 20.1%) at 3 months, and 24.7% (95% CI, 22.1% to 27.3%) at 6 months. The prestroke characteristics and the clinical status on admission in patients at 6 months, both living and deceased, are reported in Table 1.
|
First, we analyzed the data without considering compliance. On univariate Cox analysis, the overall mortality was found to be significantly correlated with age (P<0.00001; RR=1.05; 95% CI, 1.03 to 1.07), type of stroke (P<0.0001, log-rank test), history of atrial fibrillation (P<0.001; RR=2.38; 95% CI, 1.46 to 3.87), cardiac insufficiency (P<0.02; RR=2.47; 95% CI, 1.23 to 4.24), previous TIA (P<0.05; RR=1.92; 95% CI, 1.08 to 3.43), and initial BI score (P<0.0001; RR=1.15; 95% CI, 1.11 to 1.20). In particular, mortality was significantly higher in TACI patients than in the other subgroups taken as a whole (P<0.0001; RR=5.06; 95% CI, 3.06 to 8.35). On multivariate Cox analysis, only age, type of stroke, and initial BI score were still significant independent predictors (age: P<0.027; RR=1.03; 95% CI, 1 to 1.05; type of stroke [TACI versus others]: P<0.014; RR=2; 95% CI, 1.15 to 3.48; initial BI score: P<0.00001; RR=1.11; 95% CI, 1.07 to 1.16).
The Kaplan-Meier survival curves for each clinical subtype of stroke showed significant differences (Figure 1; P<0.0001). In particular, the highest risk of death was detectable in the TACI subgroup (6 months: TACI, 57.1%; PACI, 20.4%; LACI, 7.8%; POCI, 14.6%). Increasing age significantly correlated with death only in PACI subjects (P<0.002; RR=1.07; 95% CI, 1.03 to 1.12) and LACI subjects (P<0.05; RR=1.06; 95% CI, 1.00 to 1.12).
|
The causes of death during hospitalization were intracranial hypertension in 35.8% (n=14) of cases, pneumonia in 20.5% (n=8), acute cardiac insufficiency in 12.8% (n=5), pulmonary embolism in 7.7% (n=3), ventricular arrhythmia in 7.7% (n=3), recurrence of stroke or hemorrhagic transformation in 10.3% (n=4), gastric bleeding in 2.6% (n=1), and acute renal insufficiency in 2.6% (n=1).
During the follow-up, 9.6% (n=5) of patients had a stroke recurrence, 3.8% (n=2) died of cardiac causes (acute myocardial infarction, lethal arrhythmia), and 86.5% (n=45) died of nonvascular causes.
In the second phase of the statistical analysis, we considered compliance with the guidelines. The major areas of noncompliance, the number of times each appears, and relative percentages are reported in Table 2. The Cox model showed that mortality was directly correlated with the NCR (P<0.0007; RR=2.26; 95% CI, 1.51 to 4.67). The median of the NCR, for the whole treatment, was 5 (range, 0 to 19). While the NCR was used as a numerical variable in the regression model, for illustrative purposes the survival curves of the 2 groups of patients with NCRs <5 and 5 are reported in Figure 2. The case fatality rate for the group with an NCR value <5 (n=220) was 1% (95% CI, 0% to 2.2%) at 7 days and 8% (95% CI, 3.9% to 12%) at 6 months; for those with a threshold value 5 (n=141), the case fatality rate was 2% (95% CI, 0.1% to 3.4%) and 23% (95% CI, 13% to 25.5%), respectively. Thus, at 6 months we detected a difference in mortality of 15%. The influence of noncompliance on mortality was also confirmed when the multivariate analysis was performed, including in the model age, type of stroke, and initial BI score, which were identified as independent predictors of mortality (NCR: P<0.03; RR=1.07; 95% CI, 1.01 to 1.15; age: P<0.02; RR=1.03; 95% CI, 1.01 to 1.06; BI: P<0.0001; RR=1.10; 95% CI, 1.05 to 1.16). Compliance with guideline-recommended actions was inversely correlated with age (R=-0.42, P<0.0001). However, NCR was still significant for survival when corrected by age, and the median age of the 2 groups of patients with NCR <5 and 5 was very similar (71 versus 74 years).
|
|
When we refined the analysis according to stroke type, evaluating the relationship between noncompliance and death, a very significant correlation was detectable only in the LACI group (P<0.003; RR=1.23; 95% CI, 1.07 to 1.41), but in TACI subjects we also observed a trend suggesting an increase in mortality on increase in the NCR (P<0.07). In the POCI group there were too few deaths to allow a correlation to be computed.
We then analyzed separately the guidelines for the acute and for the early clinical phases to verify both their applicability and association with outcomes. This analysis can also be used as a means of identifying possible different behaviors of physicians when faced with medical emergencies versus nonemergency situations.
In the acute phase, the median NCR during the acute phase was 5 (range, 3 to 13). None of the patients was treated entirely in accordance with the guidelines. The most common violations were in relation to admission to a thrombolysis/neuroprotection study protocol (86.4%), 24-hour ECG monitoring (80%), physical medicine evaluation (60.9%), and rapid execution of general and neurological assessment (52.7%). The NCR in this phase did not correlate significantly with long-term survival but was found to correlate with early death (P<0.02, multivariate logistic analysis).
In the early clinical phase, the median NCR during the early clinical phase was 3 (range, 0 to 10); only 5 patients were treated entirely in accordance with the guidelines. Survival continued to be significantly correlated with guideline compliance in this phase. In Figure 3, the survival curves of the 2 groups of patients are characterized by a threshold NCR of 5 (P<0.00004; RR=4.2; 95% CI, 2.5 to 7). At 6 months we detected a decrease in mortality of 21%.
|
We then analyzed the effects of each guideline recommendation on survival. Early mobilization (P<0.001), echo-duplex scanning of extracranial vessels (P<0.004), cardiological assessment (P<0.02), and secondary prevention treatment (P<0.001) were significantly correlated with survival. The echo-duplex scanning of extracranial vessels (P<0.007; RR=2.5; 95% CI, 1.3 to 5.0), cardiological assessment (P<0.01; RR=2.1; 95% CI, 1.2 to 3.8), and secondary prevention treatment (P<0.05; RR=1.9; 95% CI, 1.0 to 3.5) were still significantly correlated with survival even when the multivariate analysis included age and BI score at the time of maximum impairment, variables that were strongly correlated with mortality. The most frequent motivation for noncompliance was lack of resources or of intrahospital coordination. Physicians failed to agree with guideline recommendations only in the choice of pharmacological treatment for the control of infectious complications and for secondary prevention (Table 2), preferring to be guided by personal experience rather than to adopt an evidence-based approach. On no occasion did clinicians justify their noncompliance with guidelines on the basis of the severity of a patient’s condition.
For functional outcome, the median initial BI scores in the 2 groups of patients with NCR <5 and 5 were 12 (interquartile range, 5 to 18) and 8 (interquartile range, 3 to 16), respectively. TE at discharge was significantly correlated with NCR (Spearman’s rank correlation=-0.3; P<0.001) and was higher in the group of patients with NCR of <5 (NCR <5: median, 33%; NCR 5: median, 20%; P<0.02, Wilcoxon test) (Figure 4).
|
The influence of noncompliance on TE was also confirmed when the multivariate analysis model included NCR, age, and initial BI, variables identified as independent predictors of poor outcome (Table 3). More precisely, the squared value of the initial BI was included in the model because it showed a better significance in relation to the rough value, possibly because it takes into account the nonlinearity due to the upper-threshold effect.
|
At the 3- and 6-month follow-ups, TE was still higher in the group of patients with NCR of <5, but the difference was not significant (median, 75% versus 72% at 3 months and 79% versus 76% at 6 months).
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1Appendix 2References |
|---|
This study demonstrates an association between stroke outcome and compliance with guidelines, even though a randomized trial is needed to demonstrate a direct cause and effect relationship. The most striking result is the reduction of mortality at the 6-month follow-up, with risk of death found to be directly related to the level of noncompliance with guideline-recommended actions. In particular, compliance with guideline recommendations in the acute phase is able to reduce the number of early deaths, while good compliance during both the acute and the early clinical phases improves stroke survival at 6 months. Furthermore, a better functional outcome is also observable in the presence of a higher level of compliance with the guidelines. TE (calculated with the use of the BI) at discharge is significantly higher in the group showing less noncompliance, and this trend is confirmed at 3 and 6 months (even though statistical significance is not reached, probably because the functional recovery after hospital discharge is influenced by a series of other variables, such as length and type of physical rehabilitation and support of relatives, which we did not quantify in this study).
A complete neurological and cardiovascular assessment, early mobilization, and an evidence-based secondary prevention therapy seem to be the procedures able to strongly influence the outcome of stroke patients. It is important to note that these procedures are recommended by the AHA guidelines regardless of a patient’s clinical status. In this study we did not evaluate whether noncompliance in single areas was associated with evidence of failure to proceed with further intervention, but the importance of specific actions, such as early mobilization, relevant investigations, or acute monitoring, has been confirmed by recent studies.6,20–22 However, the real significance of these results probably does not lie solely in this statistical evidence: it is likely that good clinical practice, achievable through the application of evidence-based guidelines is, per se, capable of modifying and improving the medical and nursing approach to stroke. It means that patients are monitored and treated better than in routine settings, which, as is known, are characterized by a lack of effective treatments for the acute phase of the disease. In view of this lack of effective treatments, we also considered the recommendation of admission to thrombolysis/neuroprotection study protocol because it has previously been emphasized that supporting these protocols through patient referral and participation is crucial for evaluating additional treatment options23 that could positively influence patient outcome. Thus, if a patient reaches the hospital within the "therapeutic window," the possibility of including him or her in a study protocol must be considered.
Some behavioral biases were detected. First, the frequency distribution of noncompliance shows that much more attention is currently devoted to the early clinical phase and to young patients, suggesting that the same level of attention is not perceived as particularly useful in older patients or in the acute phase. However, when the average age of the 2 groups (good/poor compliance) is considered, it seems that the age difference is not great enough to justify the difference in mortality. Moreover, it is also possible that some physicians still prefer a medical approach based on personal experience to an evidence-based one, especially in the choice of pharmacological treatments. These aspects, together with the high rate of deviation from guideline recommendations due to lack of coordination or resources, could reflect the limited scenario of Italian healthcare and may not be directly applicable to other healthcare systems.
We performed a statistical evaluation based on univariate and multivariate analyses aimed at identifying possible confounding factors because the objective of our study was not to compare different types of care but to ascertain whether greater compliance with guidelines could modify the outcome of stroke patients. Additionally, the "sum score" of deviations is not to be read as an absolute measure of the care process but only as a quantification of deviations used for the purpose of investigating a correlation between noncompliance and mortality. We demonstrated that this correlation does exist, which means that good clinical practice, based on scientific evidence, is able to modify favorably the outcome of these patients, even in the absence of efficacious pharmacological treatments.
Our study is the first attempt to date to verify the possible effects of guideline application on stroke outcome. The positive results observed in the study must of course be confirmed through a larger multicenter randomized study comparing "guideline care" with "conventional care." The electronic support for medical decision making used in this project could also be applied in a large-scale evaluation of the application of stroke guidelines, conducted to obtain more definite information about their applicability in individual countries and to establish the effectiveness of their use in clinical practice.
Although preliminary, these data emphasize the need for professional education on stroke and for much more wide-spread use of guidelines among those involved in stroke care. This could lead to better general management of stroke patients in the acute and early clinical phases, more specific identification of the causes of stroke, and more rational secondary prevention therapy based on scientific evidence and well-defined stroke pathogenic mechanisms. The low impact of guideline compliance on long-term stroke survival in the acute phase could be explained by the lack of specific treatment currently available.
Our study stresses the need to raise awareness, particularly among general practitioners and neurologists, of stroke as a medical emergency. At the present time, however, many hospitals do not have the personnel, equipment, and organization necessary for the treatment of stroke patients. A recent study revealed that in North Carolina 66% of surveyed hospitals did not have stroke protocols, and 82% did not have procedures in place for the rapid identification of patients experiencing acute stroke.24
In conclusion, there is an absolute need to develop personnel and structures devoted to stroke care because even though there is, as yet, no "ideal" pharmacological treatment, an evidence-based clinical approach, supported by online help, can significantly reduce the risk of death.
|
Appendix 1 |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1Appendix 2References |
|---|
|
|
Appendix 2 |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1Appendix 2References |
|---|
GLADIS Study Group
G. Micieli, A. Cavallini, Cerebrovascular Unit, IRCCS C. Mondino, Pavia, Italy (Coordinating Center); A. Mamoli, M. Camerlingo, Department of Neurology, Ospedali Riuniti, Bergamo, Italy; G.P. Anzola, P. Zavarise, Department of Neurology, Spedali Civili, Brescia, Italy; L. Frattola, S. Bassi, Department of Neurology, IRCCS S. Gerardo, Monza, Italy; S. Quaglini, M. Stefanelli, Department of Computer Science and Systems, University of Pavia, Italy (informatics and statistical support).
|
Footnotes |
|---|
A complete list of the members of the GLADIS Study Group appears in Appendix 2.
Received October 31, 2001; revision received December 20, 2001; accepted January 16, 2002.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionAppendix 1Appendix 2References |
|---|
1. Murray CJL, Lopez AD. Mortality by cause for eight regions of the world: global burden of disease study. Lancet. 1997; 349: 1269–1276.[CrossRef][Medline] [Order article via Infotrieve]
2. Isard PA, Forbes JF. The cost of stroke to the National Health Service in Scotland. Cerebrovasc Dis. 1992; 2: 47–50.
3.
Taylor TN, Davis PH, Torner JC, Holmes J, Meyer JW, Jacobson MF. Lifetime cost of stroke in the United States. Stroke. 1996; 27: 1459–1466.
4. Adams HP Jr, Brott TG, Crowell RM, Furlan AJ, Gomez CR, Grotta J, Helgason CM, Marler JR, Woolson RF, Zivin JA, Feinberg W, Mayberg M. Guidelines for the management of patients with acute ischemic stroke. Stroke. 1994; 25: 1901–1914.[Medline] [Order article via Infotrieve]
5. Feinberg WM, Albers GW, Barnett HJM, Biller J, Caplan LR, Carter LP, Hart RG, Hobson RW, Kronmal RA, Moore WS, Robertson JT, Adams HP Jr, Mayberg M. Guidelines for the management of transient ischemic attacks. Stroke. 1994; 25: 1321–1335.
6. Evans A, Perez I, Harraf F, Melbourn A, Steadman J, Donaldson N, Karla L. Can differences in management processes explain different outcomes between stroke unit and stroke-team care? Lancet. 2001; 358: 1586–1592.[CrossRef][Medline] [Order article via Infotrieve]
7. 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]
8. Teasdale G, Jennett B. Assessment of coma and impaired consciousness: a practical scale. Lancet. 1974; 2: 81–84.[CrossRef][Medline] [Order article via Infotrieve]
9.
Brott TG, Adams HP, Olinger CP, Marler JR, Barsan WG, Biller J, Spilker J, Holleran R, Eberle R, Herztberg V, Rorick M, Moomaw CJ, Walker M. Measurements of acute cerebral infarction: a clinical examination scale. Stroke. 1989; 20: 864–870.
10. Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Md State Med J. 1965; 14: 61–65.[Medline] [Order article via Infotrieve]
11.
Surya S, Vanclay F, Cooper B. Efficiency, effectiveness, and duration of stroke rehabilitation. Stroke. 1990; 21: 241–246.
12. Paolucci S, Antonucci G, Grasso MG, Morelli D, Troisi E, Coiro P, Bragoni M. Early versus delayed inpatients stroke rehabilitation: a matched comparison conducted in Italy. Arch Phys Med Rehabil. 2000; 81: 695–700.[Medline] [Order article via Infotrieve]
13. Quaglini S, Dazzi L, Gatti L, Stefanelli M, Fassino C, Tondini C. Supporting tools for guideline development and dissemination. Art Intell Med. 1998; 14: 119–137.
14. Quaglini S, Stefanelli M, Cavallini A, Micieli G, Fassino C, Mossa C. Guideline-based careflow systems. Art Intell Med. 2000; 20: 5–22.
15. Becker RA, Chambers JM, Wilks AR. The New S Language: A Programming Environment for Data Analysis and Graphics. Pacific Grove, Calif: Wadsworth & Brooks; 1988.
16. Kaplan EL, Meier P. Non-parametric estimation from incomplete observations. J Am Stat Assoc. 1958; 53: 457–481.[CrossRef]
17. Kalbfleisch JD, Prentice RL. The Statistical Analysis of Failure Time Data. New York, NY: Wiley; 1980.
18. Cox DR, Oakes D. Analysis of Survival Data. New York, NY: Chapman and Hall; 1984.
19. Harrel F. The PHGLM procedure.In: SAS Supplementary Library User’s Guide (Version 5). Cary, NC: SAS Institute Inc; 1986.
20.
Indredavik B, Bakke F, Slordahl SA, Rokseth R, Haheim LL. Treatment in a combined acute and rehabilitation stroke unit: which aspects are most important? Stroke. 1999; 30: 917–923.
21.
Karla L, Yu G, Wilson K, Roots P. Medical complications during stroke rehabilitation. Stroke. 1995; 26: 990–994.
22. Warlow CP, Dennis MS, van Gijn J, Hankey GJ, Sandercock PAG, Bamford JM, Wardlaw J. Stroke: A Practical Guide To Management. London, UK: Blackwell Science; 1996; 364–365.
23. Adams JR. Management issues for patients with ischemic stroke. Neurology. 1995; 45 (suppl 1): S15–S18.[Abstract]
24.
Goldstein L, Hey LA, Laney R. North Carolina stroke prevention and treatment facilities survey: statewide availability of programs and services. Stroke. 2000; 31: 66–70.
This article has been cited by other articles:
 |
|
 |
|
  B. Wettermark, A. Persson, and M. von Euler Secondary Prevention in a Large Stroke Population: A Study of Patients' Purchase of Recommended Drugs Stroke, October 1, 2008; 39(10): 2880 - 2885. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Norrving, P. Wester, K. S. Sunnerhagen, A. Terent, A. Sohlberg, F. Berggren, P.-O. Wester, K. Asplund, for the Stroke Guidelines Working Group,, and National Board of Health and Welfare, Stockholm, S Beyond Conventional Stroke Guidelines: Setting Priorities Stroke, July 1, 2007; 38(7): 2185 - 2190. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. K. Kapral, A. Laupacis, S. J. Phillips, F. L. Silver, M. D. Hill, J. Fang, J. Richards, J. V. Tu, and for the Investigators of the Registry of the Canad Stroke Care Delivery in Institutions Participating in the Registry of the Canadian Stroke Network Stroke, July 1, 2004; 35(7): 1756 - 1762. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Cavallini, G. Micieli, S. Marcheselli, and S. Quaglini Role of Monitoring in Management of Acute Ischemic Stroke Patients Stroke, November 1, 2003; 34(11): 2599 - 2603. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||