XVII. European Stroke Conference
Nice, France

Oral Session:

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 10:30 - 10:40 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

01
Birth weight and childhood BMI among 191,163 Danish schoolchildren and the risk of ischemic stroke in adulthood
J. Baker    L. Olsen    T. Truelsen    T.I.A.Sørensen                                          
 

Institute of Preventive Medicine, Centre for Health and Society

DENMARK

Background: It has been shown that higher childhood BMI raises the risk of coronary heart disease in adulthood (Baker JL, Olsen LW, Sørensen TIA. NEJM 2007;357:2329-37). Although birthweight has been associated with stroke, less is known about how childhood BMI may affect the risk. As stroke subtypes have different etiologies, we focused on ischemic stroke and investigated if birthweight and childhood BMI from 7-13 years of age was associated with ischemic stroke in adulthood. Methods: 97.391 boys and 93.772 girls born from 1936-1976 in the Copenhagen School Health Records Register were included in the study. BMI values were calculated and transformed to z scores. Stroke events were obtained from Danish Hospital and Cause of Death Registers. Cox regressions were performed. Results: Birthweight was associated with ischemic stroke in adulthood among boys and girls. Compared with the reference category (3251-3750g) boys in the lowest (2000-2750g) had a relative risk (RR) of 1.60 (95%CI: 1.30-1.98) of stroke, and girls in the lowest category had a RR of 1.34 (95%CI: 1.02-1.75). BMI at 13 years was independently associated with ischemic stroke only in boys. In a model adjusted for birthweight, compared to those with a z score in the reference category (-0.25-0.25), boys with a BMI z-score  0.75 had a RR of 1.35 (95%CI: 1.10-1.67). Conclusion: Birthweight is associated with ischemic stroke in adulthood among boys and girls. BMI at 13 years of age was associated with ischemic stroke in adulthood only among boys. These results suggest that some risk for ischemic stroke is already present at birth but that childhood BMI among boys matters as well; thus prevention possibilities remain. Funding: NIH NRSA F32DK070491 (to JLB) & the Danish National Research Foundation

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 10:40 - 10:50 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

02
Transition into Summer Time abruptly shifts the circadian pattern of stroke onset
C. Foerch    H.W.Korf    H. Steinmetz    M. Sitzer                                          
 

Goethe-University, Frankfurt am Main

GERMANY

Background: Stroke onset shows a circadian pattern with highest incidence in morning hours. Diurnal oscillations of blood pressure, hormones and platelet function are supposed to be underlying mechanisms, but exogenous factors like increasing physical activity after awakening and recognition of so far undetected night time strokes may also be of relevance. Transition into Summer Time (ST) and back constitutes an arbitrary interference into the natural time pattern. This study was performed to characterize the effects of the transitions on the circadian pattern of stroke onset. Methods: Our study relied on a prospective hospital based stroke registry in Germany containing 44 251 datasets with admission dates between year 2000 and 2005 and a known time point of stroke onset. To gain a uniform timeline, time points of stroke onset were set back from Central European Summer Time (CEST) to Central European Time (CET) for patients admitted in ST periods. Results: As compared to the last week prior to the clock change, transition into ST resulted in a drop of the time points of stroke onset in the first week after the clock change in reference to the uniform timeline (i.e. CET; for the 25th percentile of the diurnal pattern -60min, for the 50th percentile -60 min; Mann Whitney U p<0.001, patients pooled on a weekly basis). Vice versa, transition from CEST to CET led to increasing stroke onset time points (+60min, +60min, respectively; p=0.001). A significant shift was already present at the first and second day after the transitions (i.e. Mon/Tue). A gradual adjustment within the first week was not apparent. Discussion: Transition into ST and back is coupled with sudden shifts of stroke onset time points in reference to the timeline effective prior to the shifts. As clock change is likely to abruptly shift the sleep-wake cycle of most individuals, our results support exogenous factors associated with awakening to be important determinants of the circadian pattern of stroke onset rather than endogenous factors, as molecular circadian clocks would need at least a few days to gradually adjust.

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 10:50 - 11:00 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

03
Patients with cardiac, cerebral or peripheral artery disease have different vascular risk factors and different risks of future vascular events
S.W.Achterberg    M.J. Cramer    L.J. Kappelle    G.J.deBorst    Y.van derGraaf    A. Algra                            
for the SMART study group

University Medical Center Utrecht

THE NETHERLANDS

Background: Patients with coronary artery disease (CAD), cerebrovascular disease (CVD), or peripheral artery disease (PAD) have different expressions of atherothrombotic disease. Limited data are available on a direct comparison of their vascular risk factors at baseline and their risk of future vascular events during follow-up. Methods and Results: This study involved 3114 consecutive patients from the same hospital based series (the Second Manifestations of ARTerial disease (SMART) study) who were treated because of clinical manifestation of cerebral, cardiac or peripheral atherothrombotic disease,. Prevalence of baseline characteristics was compared between the three SMART subgroups with linear or logistic regression analysis. Primary outcome was the composite of myocardial infarction, stroke and vascular death, The incidence of outcome events was analysed with Cox regression. All analyses were adjusted for age and sex. At baseline, CAD patients were the most obese (BMI 27.3 kg/m2), PAD patients smoked most (54%) and had higher prevalence of hypertension and hypercholesterolemia compared with the other groups. CVD and PAD patients had more frequent aortas >3 cm than CAD patients (OR 2.21; 95%CI 1.40-3.47), more often carotid stenosis >50% (OR 8.7; 95%CI 6.8-11.1), more frequent ankle brachial index <0.90 (OR 12.8; 95%CI 10.2-16.0), and more renal impairment (OR 1.46; 95%CI 1.22-1.74). Patients were followed during a mean of 4.1 years. Primary outcome events were less common in patients with CAD (6.8% vs. 14.5% in CVD and 14.1% in PAD). Patients with PAD had the highest death rate (3.9% vs. 3.0% for CVD and 1.4% for CAD) and significantly more cardiac events than patients with CAD (HR 1.4; 95%CI 1.0-1.9). Patients with CAD had striking less major bleeding complications compared with the other two patient groups (HR 0.42; 95%CI 0.26-0.67). Conclusions: Patients with different manifestations of atherosclerosis have a different vascular risk factor profile. Prognosis in patients with clinical manifestations of CAD is more benign than in patients with CVD or PAD.

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 11:00 - 11:10 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

04
Long-term appearance of lacunar infarction on imaging: what proportion become CSF-containing cavities?
G.M.Potter    F. Doubal    C.A.Jackson    C. Sudlow    M.S.Dennis    J.M.Wardlaw                            
 

University of Edinburgh, Western General Hospital

UNITED KINGDOM

Introduction: On T2 or FLAIR MR imaging, established lacunar infarcts are often defined as small CSF-containing cavities, but when acute, many lacunar infarcts would be indistinguishable from age-related white matter hyperintensities (WMH) without DWI, especially in older patients. The time taken for acute lacunes to go from resembling a WMH to a cavity, and the proportion that do, is unknown. Methods: From two studies, we identified all patients with acute lacunar stroke, who had DWI, T2W and FLAIR imaging at presentation and FLAIR/T2 at least two weeks later. The acute infarct on FLAIR/T2 was identified by reference to DWI; its appearance on follow-up FLAIR was classified as resembling WMH/high signal only, or possibly/probably cavitated (ie containing low signal CSF on FLAIR). On the acute scan, we recorded acute infarct size (mm), presence of CSF-containing “old lacunar infarcts”, WMH score (Fazekas), and enlarged perivascular spaces (PVS, <2mm on T2WI, on a local scale). Results: Amongst 30 patients, 8 (27%) showed changes on the follow-up scan consistent with developing cavitation (median 57, range 39-81 days), vs median 53 (range 30-168) days for the 22 (73%) with no cavitation. Average infarct size for patients with cavitation was 12.4mm vs 9.5mm in those showing no cavitation (p=0.13). CSF-containing “old lacunar infarcts” were present in 6/8 cavitating vs 9/22 non-cavitating infarcts (Chi squared p=0.09). Moderate/severe WMH and moderate/severe PVS were not more frequent in those with cavitating infarcts (Chi squared p=0.77 and p=0.42 respectively). Conclusions: Cavitation occurs in <1/3 of lacunar infarcts within the first 2-3 months but some may never cavitate and so permanently resemble WMH. Lacunar infarcts which retain the appearance of WMH (up to 2/3) may be overlooked where only CSF-containing cavities are counted as old lacunar infarcts. This has important implications for epidemiology and observational studies of the causes and associations of lacunar stroke.

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 11:10 - 11:20 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

05
Incidence and survival in symptomatic lacunar stroke in Dijon, France, from 1989 to 2006: a population-based study
Y. Bejot    G.V.Osseby    M. Caillier    A. Catteau    O. Rouaud    G. Couvreur    J. Durier    T. Moreau    M. Giroud       
 

Dijon Stroke Registry, EA 4184

FRANCE

Background: Lacunar infarcts are usually regarded as benign, but population-based studies are required to assess the exact impact of this stroke subtype in cerebrovascular pathology. Methods: We evaluated incidence and survival in symptomatic lacunar stroke in a prospective well-defined population-based study, in Dijon, France (150,000 inhabitants), from 1989 to 2006. Results: A total of 2536 ischemic strokes were recorded. Among these, 715 (28%) were lacunar infarcts (354 men and 361 women). From 1989 to 2005, we observed a significant rise in the incidence of lacunar stroke in the two sexes considered together ([RR] 1.02; 95% CI 1.005-1.035; p=0.007) whereas the variation was not significant in either men or women when considered separately. The incidence significantly increased in patients under 65 years old ([RR] 1.049, 95% CI 1.0175-1.0817; p=0.002). For lacunar infarcts, survival rates were 96% at 1 month (95% CI 0.94-0.97), 86% at 1 year (95% CI 0.83-0.89), and 78% at 2 years (95% CI 0.75-0.81), and were significantly higher than those for non-lacunar stroke (HR 2.05; 95% CI 1.70-2.47, p<0.001). Conclusion: Our results suggest a significant increase in the incidence of lacunar stroke even when any classification biases are taken into account. Even though lacunar stroke is associated with a relatively good short-term prognosis in terms of survival, there is a demonstrated risk of reccurence and cognitive dysfunction. Lacunar stroke should therefore not be considered as benign.

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 11:20 - 11:30 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

06
THE EFFECT OF VERY EARLY MOBILISATION ON MOOD AFTER STROKE
J.M.Collier    T.B.Cumming    A.G.Thrift    J. Bernhardt                                          
 

National Stroke Research Institute

AUSTRALIA

Background After stroke, patients spend over 50% of the day resting in bed. Immobility may increase negative mood symptoms, such as depression and anxiety. Increasing mobility early after stroke may have important psychological benefits. Aims To determine the effect of very early mobilisation (VEM) on depression and anxiety after stroke. Methods Patients with confirmed stroke admitted within 24 hours of symptom onset were included. Data were collected as part of A Very Early Rehabilitation Trial (AVERT), a multi-centre randomised controlled trial. VEM patients were mobilised earlier (within 24 hours of stroke) and more frequently than standard care (SC) patients. The Irritability, Depression and Anxiety (IDA) scale was administered at 7 days and 12 months post stroke. Results Mean age of patients (n=71) was 74.7 years. At 7 days, VEM patients were less depressed (P=0.012) and marginally less anxious (P=0.073) than SC patients. Adjusting for potential confounding factors (age, sex, NIH score, premorbid disability and 7-day walking status), VEM patients was associated with less depression (OR 0.14, 95% CI 0.03 to 0.61, P=0.009), and no association for lower anxiety (P=0.385). Classifying IDA scores into ‘normal’ and ‘borderline/morbid’ depression, and using multivariable logistic regression, VEM was associated with less depression at 7 days (OR 2.2, 95% CI 1.3–3.7; P=0.01). There were no significant differences between groups for depression or anxiety at 12 months. Conclusion VEM may reduce depressive symptoms in stroke patients at 7 days post stroke. Reduced depression may be attributed to the exercise intervention, or the additional attention provided during VEM.

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 11:30 - 11:40 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

07
Falls in first hospitalizations for stroke in Scotland conceal an increase in rates amongst the young, 1986-2005
J.D.Lewsey    P.S.Jhund    M. Gillies    A. Redpath    L. Kelso    A. Finlayson    M. Walters    P. Langhorne    J. McMurray    K. MacIntyre
 

University of Glasgow

UNITED KINGDOM

Introduction Population based studies of stroke incidence suggested that incidence was increasing until the late 1990s. Recently a number of population based studies have suggested that the incidence of stroke is now falling. It is not known if first hospitalizations for stroke have fallen in a whole country and in all age groups. Methods Using the Scottish Linked Morbidity Record Database we identified all patients with a first episode of stroke hospitalized between 1986 and 2005 in Scotland. Rates were calculated using denominators supplied by the General Registrar Office of Scotland with interpolation of counts between census years. Rates were standardized by the direct method using the 2001 census population estimates. Poisson regression was used to model rates over time. Results Over the study period 162484 individuals (73676 men and 88808 women) were discharged from hospital after a first stroke. The mean age at admission did not change over the study period in men (69.2 years in 1986/7 vs. 68.8 years in 2004/5) or women (73.6 years in 1986/7 vs 74.2 years in 2004/5). Age standardized rates of first hospitalization for stroke (per 100,000 population) increased in men from 157 in 1986 to 177 in 1993 before falling to 132 in 2005. In women rates of first hospitalization for stroke (per 100,000 population) increased from 180 in 1986 to 195 in 1993 before falling to 145 in 2005. These overall falls concealed an increase in the rates amongst younger individuals. In men aged <55 years the relative risk (RR) of admission in 2005 vs. 1986 was 1.43 (95% CI 1.33-1.53). Similarly, rates increased in women aged <55 years in 2005 vs. 1986, RR 1.36 (1.26-1.47). Conclusions We report in a whole country that the rates of first hospitalizations for stroke are now falling, after increasing until the early 1990s. However, rates increased in under 55s. We hypothesise this trend is due to a lower threshold for recognition and investigation of focal neurological symptoms in the young.

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 11:40 - 11:50 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

08
All cause mortality rate after stroke in 31,821 patients with atrial fibrillation characterised by CHADS2 scores
K.M.Henriksson    B. Farahmand    S. Johansson    S. Åsberg    N. Edvardsson    A. Terént                            
 

The Stroke Unit at Uppsala University

SWEDEN

This study examined all cause mortality rate after stroke in patients with atrial fibrillation (AF) and assessed, retrospectively, the impact by CHADS2 score. The CHADS2 score (0-6 points) is formed by assigning 1 point each for congestive heart failure (CHF), hypertension, age ≥75 years (yrs) and diabetes, and 2 points for a previous stroke. The Swedish National Quality Register for Stroke Care (RS) 2001-5 was, by records, linked to the Inpatient and Cause of Death registries. A cause-specific mortality rate after stroke by CHADS2 score and different combinations of the diagnoses in each CHADS2 score was calculated. The cohort included 31,821 (30% of all subjects in RS) AF patients. During the mean 10 months follow-up, 11,285 deaths occurred (9,326 due to cardiovascular diseases, 4,385 to strokes). Overall mortality rate increased from 102/1000 person-yrs in the lowest, to 760 for those with the highest CHADS2 score. Mortality rate varied within each CHADS2 score. The diagnoses most frequently contributing to a CHADS2 >4 were hypertension, age >/=75 yrs and previous stroke. CHF was the most common extra fifth point whereas the influence of diabetes seems to play a lesser role. In patients with a CHADS2 score of 4, the total and stroke mortality were approximately the same in patients with the co-morbidity CHF, age>/=75 yrs and previous stroke as the combination of CHF, hypertension, age >/=75 yrs and diabetes. Hypertension was not as prominent in patients with a CHADS2 score of 5 as with score 4. Conclusions: Total mortality rate increased more than 7-fold between subjects with the highest compared to the lowest CHADS2 score. In excess of just the numeric CHADS2 score, the combination of co-morbidity adding to the score seems to play a role.

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 11:50 - 12:00 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

09
Frequency, Severity and Outcomes of Stroke Associated with Atrial Fibrillation – The North Dublin Population Stroke Study
N. Hannon    A. Merwick    L.A.Kelly    O. Sheehan    J. Duggan    A. Moore    P.MMcCormack    E. Williams    L. Daly    P.J.Kelly
 

Neurovascular Clinical Science Unit, Mater University Hospital/University College Dublin

IRELAND

Prospective cohort studies in large population are important to determine the epidemiology of clinically-relevant stroke subgroups, to avoid selection bias which may complicate hospital-based studies. Although atrial fibrillation (AF) is a recognised risk factor for stroke, relatively few population-based data exist regarding the frequency, characteristics and outcome of AF-associated stroke. The North Dublin Population Stroke Study is a prospective population-based cohort study of TIA and stroke in an urban population of 294,592 individuals over 1 year, using multiple overlapping hospital and community sources of case ascertainment and detailed clinical and follow-up assessment, according to recommended criteria for stroke epidemiology studies. Of all 544 individuals with new stroke, AF was present in 162 (29.8%). The crude incidence rate of AF-associated stroke was 55/100,000. While ischaemic stroke was commoner in individuals with AF (92% vs 75.9%, p<0.001), haemorrhagic stroke was an important contributor to AF-associated stroke (8% of all). 57.8% had known AF prior to stroke onset. One-third of these were taking warfarin at stroke onset. Compared to non-AF ischaemic stroke, individuals with AF were older (77.1 vs 68.7 years, p<0.001), had more recurrent stroke (22.2% vs 12.8%, p=0.01), greater acute severity (NIHSS 8.7 vs 6.6, p<0.01) and disability (RS 3.8 vs 3, p<0.001), and greater disability at 7 and 28 days (p<0.001). AF was not associated with higher 7 or 28-day case-fatality or recurrence. AF was common and associated with a distinct profile of severe, recurrent and disabling stroke in older adults in our population. Targeted strategies to improve prevention of AF-stroke may have substantial benefits to population health.

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 12:00 - 12:10 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

10
Comparison of incidence and case-fatality rates for stroke and myocardial infarction in a French population-based study, from 2001 to 2006: the Dijon Vascular (DIVA ) Project
Y. Bejot    A. Gentil    G.V.Osseby    M. Caillier    O. Rouaud    Y. Cottin    M. Zeller    T. Moreau    M. Giroud       
 

Dijon Stroke Registry, EA 4184

FRANCE

Background: The burden of cardiovascular disease in developed countries is dramatic and is thus a health care priority. Nevertheless, few population-based studies have provided data to compare myocardial infarction with stroke in terms of incidence and survival. Methods: We compared incidence and 1-month case fatality rates of stroke and myocardial infarction in a prospective well-defined population-based study, in Dijon, France (150,000 inhabitants), from 2001 to 2006. Results: Over the 5 years, we recorded 1020 first-ever strokes (485 in men and 535 in women) and 640 first-ever myocardial infarctions (403 in men, 237 in women). Mean age at vascular event onset was respectively 71.5 and 64.9 years old. Age-standardised incidence rates were 122/100,000/year for stroke and 70.2/100,000/year for myocardial infarction. For Europe as a whole, the figures were respectively 82.9 and 62.1/100,000/year. In women, the incidence of stroke was significantly higher than that of myocardial infarction, whatever the age ([RR] 2.26, 95% CI 1.94-2.63; p<0.001). In contrast,the incidence of stroke was significantly lower in men <55 years old ([RR] 0.60, 95% CI 0.42-0.86; p=0.007) and significantly higher in men >75 years old ([RR] 2.01, 95% CI 1.48-2.71; p<0.001). One-month case fatality rates were 9.8% for stroke and 9.84% for myocardial infarction and did not differ significantly. Conclusion: Our study demonstrates a higher incidence of stroke compared with myocardial infarction. It provides evidence for the need to set up stroke units in France, following the model of cardiology acute care units.

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 12:10 - 12:20 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

11
Secular trends in the incidence of atrial fibrillation during 17 years in Sweden - indications of a growing health problem
K.M.Henriksson    N. Edvardsson    S. Johansson    N. Hammar                                          
 

Laboratory Medicine Lund University Hospital

SWEDEN

Atrial Fibrillation (AF) is the most common arrhythmia and a rising incidence has been reported. This study presents recent national trends of hospital-treated AF in Sweden. Information on patients with AF (ICD-9 427.3 or ICD-10 148.9) from 1987-05 was acquired from the national inpatient register which covers all AF patients treated in hospital.Person years at risk were estimated from population register. Incidence rates were calculated by year, sex and age and age adjusted incidence rates were calculated. Overall, the age adjusted incidence of hospital-treated AF increased by 1.3% in men and 0.9% in women, with a substantial part occurring in the last 5 years. In relative terms this represents a more than doubled incidence as compared to the early period. This increase was less pronounced in subjects <64 years of age, but was consistently about twice as high in men (~0.5%) as in women (~0.25%). In subjects 65-74 yrs the incidence of AF continued to be stable in women (~1%) but men showed an increasing trend from ~1% at the beginning of the period to 2% towards the end. The two oldest age groups displayed a more dramatic development of AF starting in the beginning of the 1990s, the incidence in women 75-84 years increased from 2 to 4%, and in women ≥85 years from 3 to 7%. The increases in the corresponding age groups in men were from 2 to 5.5% and from 3 to almost 9%, respectively. Conclusions: Over the last two decades, there was a substantial increase in the incidence of hospital-treated AF in Sweden, especially in the oldest subjects and particularly marked in recent years. Possible explanations include an aging population, an improved survival in subjects with coronary heart disease and an increased awareness of the importance of detecting and treating AF.

Epidemiology of stroke I
Date: Thursday 15 May 2008  
Time: 12:20 - 12:30 -  Room: Erato/Uranie
Chair: D. Leys, France and N. Venketasubramanian, Singapore

12
Patients enrolled in landmark trials of antiplatelet treatment for secondary prevention are not representative of patients in clinical practice
E. Maasland    R.J.van Oostenbrugge     C. Franke    W.J.M.Scholte op Reimer     P.J.Koudstaal    D.W.J.Dippel                            
for the Netherlands Stroke Survey Investigators

Erasmus MC University Medical Center Rotterdam

THE NETHERLANDS

Background: Many randomised clinical trials (RCT) have evaluated the benefit of long-term use of antiplatelet drugs in reducing the risk of vascular events. Evidence from these trials forms the basis for many cardiovascular guidelines. Abundant and strict enrolment criteria may limit the generalisability and the applicability of results of RCTs to clinical practice. Aim: To identify the proportion of patients with stroke or TIA in clinical practice who are eligible for participation in trials investigating antiplatelet treatment for secondary prevention. Methods: In 11 centers in the Netherlands 972 patients with TIA or ischemic ischemic stroke were prospectively and consecutively enrolled. We applied 5 large antiplatelet trials’ enrolment criteria to the patients in our survey. We distinguished between enrolment criteria aimed at selection of high-risk patients and criteria aimed at safety. Results: In total, 886 patients were available for secondary prevention. Mean follow-up was 2.5 years. The annual rate of TIA, stroke or myocardial infarction was 6.7%. The proportions of patients fulfilling the trial enrolment criteria were 25% (MATCH), 32% (CAPRIE), 36% (PERFORM), 58% (ESPRIT) and 63% (ESPS-2). Mortality was significantly higher in ineligible patients (27-41%) than in patients fulfilling enrolment criteria (16-20%). Rates of vascular events were not increased in eligible patients. Conclusions: Our data confirm that RCT enrol patients are only partially representative of patients with TIA or stroke in clinical practice. Use of less strict enrolment criteria could result in valid and more efficient selection of patients for randomised clinical trials.