Articles

Effect of urgent treatment of transient ischaemic attack and

minor stroke on early recurrent stroke (EXPRESS study):
a prospective population-based sequential comparison
Peter M Rothwell, Matthew F Giles, Arvind Chandratheva, Lars Marquardt, Olivia Geraghty, Jessica N E Redgrave, Caroline E Lovelock,
Lucy E Binney, Linda M Bull, Fiona C Cuthbertson, Sarah J V Welch, Shelley Bosch, Faye Carasco-Alexander, Louise E Silver, Sergei A Gutnikov,
Ziyah Mehta, on behalf of the Early use of Existing Preventive Strategies for Stroke (EXPRESS) study

Summary
Lancet 2007; 370: 1432–42 Background The risk of recurrent stroke is up to 10% in the week after a transient ischaemic attack (TIA) or minor
Published Online stroke. Modelling studies suggest that urgent use of existing preventive treatments could reduce the risk by 80–90%,
October 9, 2007 but in the absence of evidence many health-care systems make little provision. Our aim was to determine the effect of
DOI:10.1016/S0140-
6736(07)61448-2
more rapid treatment after TIA and minor stroke in patients who are not admitted direct to hospital.
See Comment page 1398
Methods We did a prospective before (phase 1: April 1, 2002, to Sept 30, 2004) versus after (phase 2: Oct 1, 2004, to
See Lancet Neurol 2007;
6: 953–60
March 31, 2007) study of the effect on process of care and outcome of more urgent assessment and immediate
Stroke Prevention Research
treatment in clinic, rather than subsequent initiation in primary care, in all patients with TIA or minor stroke not
Unit, University Department of admitted direct to hospital. The study was nested within a rigorous population-based incidence study of all TIA and
Clinical Neurology, Radcliffe stroke (Oxford Vascular Study; OXVASC), such that case ascertainment, investigation, and follow-up were complete
Infirmary, Oxford, UK and identical in both periods. The primary outcome was the risk of stroke within 90 days of first seeking medical
(Prof P M Rothwell FRCP,
M F Giles MRCP,
attention, with independent blinded (to study period) audit of all events.
A Chandratheva MRCP,
L Marquardt MD, Findings Of the 1278 patients in OXVASC who presented with TIA or stroke (634 in phase 1 and 644 in phase 2),
O Geraghty MRCP, 607 were referred or presented direct to hospital, 620 were referred for outpatient assessment, and 51 were not
J N E Redgrave MRCP,
C E Lovelock FRACP,
referred to secondary care. 95% (n=591) of all outpatient referrals were to the study clinic. Baseline characteristics and
L E Binney BMBCh, L M Bull RGN, delays in seeking medical attention were similar in both periods, but median delay to assessment in the study clinic
F C Cuthbertson MCSP, fell from 3 (IQR 2–5) days in phase 1 to less than 1 (0–3) day in phase 2 (p<0·0001), and median delay to first
S J V Welch RGN, S Bosch, prescription of treatment fell from 20 (8–53) days to 1 (0–3) day (p<0·0001). The 90-day risk of recurrent stroke in the
F Carasco-Alexander MSc,
L E Silver MSc,
patients referred to the study clinic was 10·3% (32/310 patients) in phase 1 and 2·1% (6/281 patients) in phase 2
S A Gutnikov DPhil, (adjusted hazard ratio 0·20, 95% CI 0·08–0·49; p=0·0001); there was no significant change in risk in patients treated
Z Mehta DPhil) elsewhere. The reduction in risk was independent of age and sex, and early treatment did not increase the risk of
Correspondence to: intracerebral haemorrhage or other bleeding.
Prof Peter M Rothwell, Stroke
Prevention Research Unit,
University Department of Clinical
Interpretation Early initiation of existing treatments after TIA or minor stroke was associated with an 80% reduction
Neurology, Radcliffe Infirmary, in the risk of early recurrent stroke. Further follow-up is required to determine long-term outcome, but these results
Woodstock Road, Oxford have immediate implications for service provision and public education about TIA and minor stroke.
OX2 6HA, UK
peter.rothwell@clneuro.
ox.ac.uk
Introduction including aspirin,13 other antiplatelet agents,14–16 blood-
The risk of recurrent stroke in the week after a transient pressure-lowering drugs,17 statins,18 anticoagulation for
ischaemic attack (TIA) or minor stroke is up to 10%.1–3 atrial fibrillation,19 and endarterectomy for 50% or greater
About 150 000 suspected TIAs and minor strokes are symptomatic carotid stenosis.20,21 Assuming that these
referred to secondary care for assessment and investigation effects are independent, use of all of these interventions in
in England alone each year,4 and rates are similar in the appropriate patients would be predicted to reduce the
USA.5 These warning events provide a short window of long-term group risk of recurrent stroke by 80–90%.22
opportunity for prevention,6 but there is considerable Moreover, evidence from trials of treatment in acute stroke
variation in service provision, ranging from emergency and acute coronary syndromes suggests that the relative
inpatient care to non-urgent outpatient clinic assessment,7,8 benefits of several of these interventions are, if anything,
with little consensus about the most cost-effective likely to be even greater in the acute phase.23–28 However,
strategy.9,10 In the UK, it is recommended that patients are there are no large randomised trials of these, or any other
seen in specialist outpatient clinics,11 but many hospitals interventions, in the acute phase after TIA and minor
offer only a weekly clinic and about half of all patients in ischaemic stroke; this lack of evidence has undoubtedly
the UK wait more than 14 days to be seen, and have even contributed to the variation in service provision.
longer delays to investigation, and treatment.12 In 2002, in light of increasing evidence of the high early
Several treatments are effective in preventing stroke in risk of stroke after TIA,1,29 and of high rates of stroke
the long term after a TIA or minor ischaemic stroke, before assessment in an audit of referrals to our weekly

1432 www.thelancet.com Vol 370 October 20, 2007


TIA clinic,30 we aimed to determine the effect of more
rapid treatment after TIA and minor stroke in patients
who are not admitted direct to hospital. Given the
predicted benefits of early treatment, and emerging
evidence of substantial benefits of early endarterectomy
for symptomatic carotid stenosis,31 we considered
randomisation to early versus delayed assessment and
treatment to be unethical and unfeasible. We therefore
instigated a rigorous observational study of the phased
introduction of early assessment and treatment (Early
use of EXisting PREventive Strategies for Stroke;
EXPRESS), nested within a population-based study of all
incident and recurrent TIA and stroke in Oxfordshire,
UK (Oxford Vascular Study; OXVASC).32,33
Methods
Patients
The methods of OXVASC and details of the study
population have been reported previously.32,33 Briefly, the
study population consisted of all 91 000 individuals,
irrespective of age, registered with 63 primary-care
physicians in nine primary-care practices in Oxfordshire,
UK. In the UK, almost all individuals register with a
primary-care practice, which holds a lifelong medical
record. OXVASC was approved by our local research
ethics committee and began on April 1, 2002, after a
3-month pilot to ensure reliable case ascertainment. This
report concerns all events identified in the study
population in the first 5 years, with follow-up to June 30,
2007.
Procedures
At the start of the EXPRESS study (April 1, 2002), we
introduced a daily TIA and minor stroke clinic to which
collaborating primary-care physicians were asked to refer
all patients with suspected TIA and minor stroke in the
study population. However, in keeping with usual UK
practice, the clinic was appointment-based, with inherent
delays in receiving referrals and contacting patients, and
did not initiate treatment, but made treatment
recommendations to the referring primary-care
physician. The EXPRESS study intervention was a
reduction in delay to clinic assessment and initiation of
treatment in the study clinic from Oct 1, 2004 onwards.
Between April 1, 2002, and Sept 30, 2004 (phase 1),
primary-care physicians referred, by fax, any patient
they suspected had had a TIA or stroke but whom they
did not consider required immediate hospital admission,
to our daily TIA and minor stroke clinic. The study team
contacted the patient at home (usually by telephone) to
arrange a clinic appointment as soon as possible. The
patient was seen in a daily (weekdays only) hospital
outpatient clinic (or at home if too frail to attend
hospital), and brain imaging (usually CT) and ECG were
obtained on the same day or shortly thereafter. Carotid
ultrasound imaging (all patients) and trans-thoracic or
trans-oesophageal echocardiography (when clinically
www.thelancet.com Vol 370 October 20, 2007
Articles
indicated) were arranged during the following week. A
report of the initial clinical assessment, with treatment
recommendations, was faxed to the primary-care
physician after the clinic (usually within 24 h), but no
treatment was given in the study clinic and no
prescription was issued. However, patients were
instructed to contact their primary-care physician as
soon as possible.
The treatment protocol recommended to the
primary-care physician was tailored to the individual
patient but generally included: aspirin in patients not
already on antiplatelet therapy (75 mg daily), or clopidogrel
if aspirin was contraindicated; simvastatin (40 mg daily);
blood pressure lowering unless systolic blood pressure
was below 130 mm Hg on repeated measurement (either
by increases in existing medication, or by commencement
of perindopril 4 mg daily with or without indapamide
1·25 mg daily); and anticoagulation as required. In
patients seen within 48 h of their event, or those seen
within 7 days who were thought to be at particularly high
early risk,1 clopidogrel (75 mg daily, to be stopped after
30 days) was recommended in addition to aspirin. Brain
imaging was required before starting combination
antiplatelet treatment or anticoagulation after a minor
stroke.
In phase 2 (Oct 1, 2004, to March 31, 2007), we moved
to a clinic at which no appointments were necessary, and
at which treatment was initiated immediately if the
diagnosis was confirmed. Thus, the mode of access to
the clinic and time of treatment initiation were changed,
but recommendations on who should be referred to the
clinic were not. Primary-care physicians were requested
to send all patients directly to the study clinic each
weekday afternoon immediately after they presented to
medical attention (ie, no written referral or appointment
was necessary). Patients were assessed in the same way
as in phase 1, but all those who were considered to have
had a TIA or stroke were given aspirin 300 mg to take in
the clinic, together with a prescription for a 4-week supply
of any other study medication (using the same treatment
protocol as in phase 1) to start on the same day. A loading
dose of clopidogrel (300 mg) was also prescribed in cases
in whom this drug was initiated. A CT brain scan was
obtained during the clinic for patients with incomplete
resolution of symptoms at the time of assessment to
exclude intracerebral haemorrhage before giving aspirin,
clopidogrel, or anticoagulants. A report of the assessment,
investigations, and treatment given was faxed to the
primary-care physician as soon as possible after the clinic
(usually within 24 h).
In both study periods, patients were assessed by a study
physician in a standardised manner.32,33 Detailed data
were gathered on the times and methods of first contact
with medical attention after the presenting event.34
Severity of stroke was assessed with the National
Institutes of Health Stroke Scale (NIHSS).35 All cases
were reviewed with the study senior neurologist (PMR)
1433
 Articles

and classified as TIA, stroke, or other condition by use of Although our primary interest in this study was in the
standard definitions.32,33 However, because of the potential outcome of patients referred to the neurovascular clinic,
inaccuracy in clinical diagnosis of TIA and to avoid any all patients presenting to medical attention with TIA or
bias due to possible changes in diagnostic practice over stroke in the study population were ascertained and
time, any referral to the study clinic that was thought followed up, irrespective of mode of presentation or
after assessment not to be a definite TIA but that went on referral (ie, all patients presenting to the emergency
to have a recurrent stroke within 90 days of clinic department, or referred for direct admission to hospital,
assessment was included in the analysis of recurrence or managed at home by the primary-care physician).
within 90 days of first presentation (this occurred in two Ascertainment was achieved as part of the OXVASC
cases in phase 1 and in one case in phase 2). study by several overlapping methods of hot and cold
pursuit, including prospective daily searches for acute
Phase 1 Phase 2 Total events and retrospective searches of hospital and
Whole population 634 644 1278 primary-care administrative and diagnostic coding data.
Presented with TIA 233 252 485 Details of the methods of ascertainment have been
Presented with stroke 401 392 793
published previously,32,33 and direct assessment of
Outpatient care 323 297 620
ascertainment has shown it to be complete.33,36
Referred to EXPRESS clinic 310 281 591
All patients in OXVASC were followed up by a research
Referred to other clinics 13 16 29
nurse or clinical research fellow after 1, 6, 12, and
24 months and asked about any new neurological
Presented with TIA 165 172 337
symptoms and any bleeding that had required medical
Presented with stroke 158 125 283
attention. Severity of bleeding events was classified as in
Hospital-based care 285 322 607
the CURE trial.26 Current medication was recorded and
Discharged from A&E department 33 54 87
resting blood pressure was measured twice, with
Inpatient care 252 268 520
measurements 5 min apart. In patients referred to the
Presented with TIA 56 67 123
study neurovascular clinic, data about the date of
Presented with stroke 229 255 484
prescription of all medications after the presenting TIA
Not referred to secondary care 26 25 51
or stroke were obtained from the collaborating
Presented with TIA 12 13 25
primary-care practices.
Presented with stroke 14 12 26
If a recurrent vascular event was suspected at a
Table 1: Number of patients seeking medical attention after a first event, follow-up visit, the patient was reassessed and investigated
stratified by place of initial referral, or by type of presenting event by a study neurologist. Details of all potential recurrent
strokes within 90 days of seeking medical attention,
All patients in study population Patients referred to EXPRESS clinic whether or not they had been seen in the neurovascular
clinic initially, were reviewed at the end of the study by an
Phase 1 (n=634) Phase 2 (n=644) Phase 1 (n=310) Phase 2 (n=281)
independent neurologist who was blinded to the phase of
Age (years) the study. The definition of recurrent stroke used in
<80 383 (60%) 376 (58%) 207 (67%) 189 (67%) OXVASC has been reported previously,2,37 and required a
≥80 251 (40%) 268 (42%) 103 (33%) 92 (33%) sudden new symptomatic neurological deterioration on a
Male sex 301 (47%) 299 (46%) 141 (45%) 132 (47%) background of stability or improvement after the
TIA 233 (37%) 252 (39%) 156 (50%) 160 (57%) presenting event. Gradual or stuttering progression after
Hypertension 365 (58%) 378 (59%)† 167 (54%) 164 (59%)¶ an initial event was not coded as a recurrent stroke.
Diabetes 74 (12%) 78 (12%)‡ 36 (12%) 37 (13%)¶ Strokes that occurred within 24 h of an initial TIA or
Previous myocardial 83 (13%) 74 (12%)‡ 36 (12%) 29 (10%)¶ minor stroke were included. If recurrence occurred
infarction before complete symptomatic resolution of the initial
Previous angina 114 (18%) 94 (15%)‡ 57 (18%) 42 (15%)¶ event, the initial event was coded as a stroke. The blinded
Previous PVD 59 (9%) 51 (8%)‡ 18 (6%) 20 (7%)¶ independent neurologist classified all potential recurrent
Prior antiplatelet 292 (46%)* 283 (44%)‡ 139 (45%) 109 (39%)¶ strokes as definite, probable, or not stroke. Probable
Prior statin 126 (20%)* 176 (28%)† 63 (20%) 89 (32%)¶ recurrent strokes were mainly sudden deteriorations
Smoking after a major acute stroke for which a clinical diagnosis
Current smoker 73 (12%)* 83 (13%)§ 42 (14%) 41 (15%)¶ of recurrent stroke had been made by the patient’s
Ex-smoker 267 (42%)* 268 (43%)§ 125 (40%) 120 (43%)¶ physicians but for which imaging evidence was equivocal
Never 293 (46%)* 277 (44%)§ 143 (46%) 118 (42%)¶ or imaging was unavailable.
Data are n (%). Data missing for *one case, †six cases, ‡three cases, §16 cases, ¶two cases. TIA=transient ischaemic
attack. PVD=peripheral vascular disease. Statistical analysis
The study denominator and rate of recurrence were
Table 2: Baseline clinical characteristics of all patients in the study population with first presentations with
defined from the point at which patients with TIA or
TIA or stroke during each study period, and of all patients who were referred to EXPRESS study clinic
stroke first sought medical attention (eg, contact with their

1434 www.thelancet.com Vol 370 October 20, 2007

 Articles

primary-care practice, emergency department, or other

Phase 1 Phase 2 p*
services). Rates of recurrence cannot be calculated from
the time of the event that led to presentation because an Whole population 63/634 (9·9%) 27/644 (4·2%) <0·0001
unknown number of patients in the population do not Presented with TIA 29/233 (12·4%) 11/252 (4·4%) 0·0015
ever seek medical attention after a TIA. Presented with stroke 34/401 (8·5%) 16/392 (4·1%) 0·0077
To avoid any potential exclusion bias, patients in the Outpatient care 33/323 (10·2%) 6/297 (2·0%) <0·0001
study population who had an incident or recurrent event Referred to the EXPRESS clinic† 32/310 (10·3%) 6/281 (2·1%) 0·0001
while temporarily away from Oxfordshire (eg, on holiday) Referred to other clinics 1/13 (7·7%) 0/16 (0%) 0·45‡
were included. We restricted analyses to the first Presented with TIA 16/165 (9·7%) 1/172 (0·6%) 0·0001
presentation in each phase of the study. However, since Presented with stroke† 17/158 (10·8%) 5/125 (4·0%) 0·037
exclusion of patients from analysis in phase 2 if they had Hospital-based care 23/285 (8·1%) 19/322 (5·9%) 0·23
also presented in phase 1 could have caused bias (patients Discharged from A&E department 5/33 (15·2%) 5/54 (9·3%) 0·42
were not excluded from phase 1 if they had had a previous Inpatient care 18/252 (7·1%) 14/268 (5·2%) 0·30
event before the study), patients who presented in both Presented with TIA 9/56 (16·1%) 8/67 (11·9%) 0·56
phases were included in the analysis of each. To avoid Presented with stroke§ 14/229 (6·1%) 11/255 (4·3%) 0·28
any possible bias due to carry-over effects of improved Not referred to secondary care 7/26 (26·9%) 2/25 (8·0%) 0·14‡
long-term treatment in phase 1 (compared with before Presented with TIA 4/12 (33·3%) 2/13 (15·4%) 0·38‡
the study) on outcome of any subsequent events in Presented with stroke 3/14 (21·4%) 0/12 (0%) 0·22‡
phase 2, all analyses were also done for incident first-ever
*Log-rank test unless otherwise specified. †Includes six patients with intracerebral haemorrhage (three in phase 1 and
events only. three in phase 2). ‡Fisher’s exact test. §Includes 66 patients with intracerebral haemorrhage (32 in phase 1 and 34 in
The primary outcome was the proportion of patients phase 2).
with recurrence of stroke within 90 days of first
Table 3: Risk of stroke during the 90 days after first seeking medical attention, stratified according to the
presentation. Patients who were referred to the study
place of initial referral, or by type of presenting event
neurovascular clinic but had a recurrent stroke
necessitating hospital admission before clinic assessment
were coded as study clinic cases. To exclude any bias due
to temporal changes in referral patterns and to determine A
the effect of the study clinic on the whole population, we 14
Phase 1
also analysed the rate of recurrence in all patients who Phase 2
12
sought medical attention with a TIA or stroke of any
severity, both overall and stratified by place of treatment: 10
Risk of recurrence (%)

acute hospital services (ie, patients who self-referred to

8
the emergency department or emergency ambulance
services or who were referred directly to hospital after 6
contacting their primary-care practice); outpatient
4
services (ie, referred to the study clinic or other clinic);
or management in primary care without referral to 2
p<0·0001
hospital. Analyses were also stratified according to the
0
type of presenting event (TIA or stroke).
The study was powered on the basis of analysis of the
B
expected rate of recurrence in patients referred directly to 14
the study clinic. With about 300 patients with probable or
definite TIA or minor stroke (from about 600 referrals) 12
predicted in each 30-month period, and with an expected
10
Risk of recurrence (%)

90-day risk of recurrence of 15% in phase 1,1,29 the study

had 80% power to detect a 50% reduction in the 90-day 8
risk of recurrence in phase 2 at the 95% level of confidence.
6
All analyses were done with SPSS version 15.0.
4
Role of the funding source
2
None of the agencies that funded OXVASC had any input p=0·0015

into the design, performance, analysis, or reporting of 0

either the OXVASC or EXPRESS studies, or saw the 0 10 20 30 40 50 60 70 80 90
manuscript before final acceptance for publication. PMR Time (days)
and ZM had access to all the data, and PMR had final
Figure 1: Risk of recurrent stroke after first seeking medical attention in all
responsibility for the decision to submit the manuscript patients with TIA or stroke in the whole study population (A) and in all
for publication. patients with TIA (B)

www.thelancet.com Vol 370 October 20, 2007 1435

 Articles

12
A
100
Phase 1 Phase 1
Phase 2 90
Phase 2

Cumulative proportion (%)

10 80
70
Risk of recurrence (%)

8 60
50
6 40
30
4 p=0·0001 20
10
2 0

0 B
100
0 10 20 30 40 50 60 70 80 90 90
Time (days)

Cumulative proportion (%)

80
70
Figure 2: Risk of recurrent stroke after first seeking medical attention in all
60
patients with TIA or stroke who were referred to the study clinic
50
40
Phase 1 (n=310) Phase 2 (n=281) p 30
20
First call for medical attention*
10
≤12 h 128 (41·3%) 105 (37·5%) 0·35
0
≤24 h 184 (59·4%) 160 (57·1%) 0·62
First call for attention to assessment in study clinic† C
100
≤6 h 5 (1·7%) 80 (29·0%) <0·0001 90
Cumulative proportion (%)

≤24 h 70 (23·4%) 163 (59·1%) <0·0001 80

70
*Data unavailable for one patient in phase 2. †Data unavailable for two patients in 60
phase 2, and not applicable in a further 11 patients in phase 1 and three patients
50
in phase 2 who were referred to the EXPRESS clinic but had a stroke and were
40
admitted to hospital before the clinic assessment.
30
Table 4: Delay to seeking medical attention and subsequent delay in being 20
seen in the clinic for all patients who were referred to the study clinic 10
0

D
Results 100
Table 1 shows the number of patients seeking medical 90
Cumulative proportion (%)

attention after a first event, stratified by place of 80

70
presentation and by type of first event (probable or
60
definite TIA or stroke). 51 patients sought medical 50
attention after an event in both phases. 40 OXVASC 40
patients presenting with subarachnoid haemorrhage and 30
ten patients with stroke who died in the community 20
before reaching medical attention were excluded; 10
23 patients who first sought medical attention outside 0
0 10 20 30 40 50 60 70 80 90
Oxfordshire were included.
Time from seeking medical attention (days)
For all patients seeking medical attention with TIA or
stroke in the study population, and for all patients Figure 3: Cumulative proportions of patients prescribed new medication
referred to the study clinic, baseline characteristics were since first seeking medical attention
(A) Any statin drug (in patients not already on a statin); (B) clopidogrel (usually
much the same in the two phases of the study, except for in addition to aspirin); (C) initiation of a first blood-pressure-lowering drug (in
a small increase in the proportion of patients using patients not already on such medication); (D) initiation of two blood-
statins at presentation in phase 2 compared with in pressure-lowering drugs (in patients previously on none or only one drug).
phase 1 (p=0·002 for both comparisons; table 2). There Clopidogrel was given for 4 weeks only.
were no significant differences between the phases in the
numbers of patients not referred to secondary care, or in differences in any of the components of the ABCD² score
the proportion of patients who were referred or who (data not shown).3
self-referred to acute hospital services versus those 90-day follow-up data were available for all patients
referred to outpatient services (table 1). In patients who presented with TIA or stroke in the study population,
presenting with TIA, there were also no significant irrespective of where they initially sought medical

1436 www.thelancet.com Vol 370 October 20, 2007


attention or were subsequently referred. After
independent blinded adjudication, 90 follow-up events
were classified as first recurrent strokes within 90 days
of first medical attention (table 3); 80 events were
classified as definite and ten as probable (seven after
major disabling strokes and three after TIA or minor
stroke). An analysis based on definite events alone is
shown in the webtable. The results of this analysis were
qualitatively identical to our analyses with all 90 cases of
recurrent stroke; thus all analyses reported here are
based on all 90 probable or definite recurrent strokes.
There were also no qualitative differences between
analyses of 90-day rates of recurrence in patients
presenting with incident events versus that in all patients
(data not shown); the results presented here thus refer to
all events.
The risk of recurrent stroke within 90 days of first
presentation with TIA or stroke in the whole population
was significantly higher in phase 1 than it was in phase 2
(p<0·0001; figure 1 and table 3). Likewise, the 90-day
risk of stroke in all patients in the study population
presenting with TIA was significantly higher in phase 1
than in phase 2 (p=0·0015; figure 1 and table 3).
607 patients presented direct to emergency services
(either via the emergency department or emergency
ambulance, with or without first contacting primary
care), or were already in hospital at the time of stroke
(table 1). There were no differences in the baseline
clinical characteristics of these cases between the two
study periods (data not shown) and the number of
intracerebral haemorrhages was similar (32 in phase 1
and 34 in phase 2). There was no significant difference
in the 90-day risk of recurrent stroke between phase 1
and phase 2 (table 3 and webtable).
Of the 620 patients with TIA or stroke who were
referred to outpatient services during the study period,
591 (95%) were referred direct to the study clinic (table 1).
The 29 cases referred to other clinics had mainly had
ocular ischaemic events. The 591 referrals to the study
clinic with probable or definite TIA or stroke made up
58% of the 1024 referrals of suspected events to the study
clinic, and consisted of 316 TIAs and 275 minor strokes
(including six patients with intracerebral haemorrhage;
three in each study period). The median NIHSS score for
minor strokes at the time of assessment in the study
clinic was 1 (IQR 0–3) in both study periods; 236 (86%)
patients had a score of 3 or less, while 264 (96%) had a
score of 5 or less.
In patients referred to the study clinic, the risk of stroke
during the 90 days after presentation was significantly
lower in phase 2 than it was in phase 1 (2·1% vs 10·3%,
p=0·0001; figure 2 and table 3). During the same period,
there were three acute myocardial infarctions in phase 1
and one in phase 2, and seven deaths in phase 1 and four
in phase 2. The overall risk of non-fatal stroke, myocardial
infarction, or death at 90 days was 11·9% (37/310) in
phase 1 and 3·6% (10/281) in phase 2 (p=0·0002).
www.thelancet.com Vol 370 October 20, 2007
Articles
Phase 1 (n=302)* Phase 2 (n=278)* p
On antiplatelet or anticoagulant 292 (97%) 269 (97%) 1·00
On aspirin and 30-day course of clopidogrel 26 (10%) 137 (49%) <0·0001
On a statin 196 (65%) 233 (84%) <0·0001
On one or more blood-pressure-lowering drugs 187 (62%) 231 (83%) <0·0001
On two or more blood-pressure-lowering drugs 103 (34%) 168 (60%) <0·0001
Systolic blood pressure (mm Hg) 142 (20) 136 (21) 0·0019
Diastolic blood pressure (mm Hg) 80 (10) 75 (10) <0·0001
Time to carotid surgery n=17 n=15
<7 days 0 (0%) 6 (40%) 0·006
<30 days 2 (12%) 10 (67%) 0·001
Data are n (%), n/N (%), or mean (SD). *Analysis excludes patients who died before 30-day follow-up and seven
patients with incomplete data (five in phase 1 and two in phase 2).
Table 5: Measures of process of care, assessed at 1-month follow-up, in all patients with TIA or stroke
who were referred to the study clinic
There was no difference in the delay from the
presenting event to seeking medical attention in patients
subsequently referred to the study clinic between the
two study periods (table 4). However, the median delay
from seeking medical attention in primary care to
assessment in clinic was 3 (IQR 2–5) days in phase 1,
compared with less than 1 (0–3) day in phase 2
(p<0·0001); the proportion seen within 6 hours was
significantly greater in phase 2 than it was in phase 1
(p<0·0001, table 4). There were therefore fewer recurrent
strokes after presentation to primary care but before
assessment in clinic in phase 2 (3/281) than there were
in phase 1 (11/310; p=0·048).
Of the 341 patients referred to the study clinic who
were not already taking antiplatelet treatment at
presentation, the primary-care physician initiated See Online for webtable
treatment before the clinic in 71 (45 in phase 1 and 26 in
phase 2). Median time from seeking medical attention to
first prescription of one of the other treatments
recommended in the faxed letter from the study clinic to
primary care in phase 1 was 20 (IQR 8–53) days,
compared with a delay of 1 (0–3) day in phase 2
(p<0·0001). The cumulative proportions of patients
prescribed medication since first seeking medical
attention are shown in figure 3.
The reduction in the 90-day risk of recurrent stroke in
referrals to the study clinic was present for patients
presenting with TIA and for those with stroke (TIA:
16/156 in phase 1 vs 1/160 in phase 2; stroke: 16/154 vs
5/121), for both sexes (men: 16/141 vs 2/132; women:
16/169 vs 4/149), and for patients of all ages (<70 years:
11/100 vs 3/116; 70–79 years: 13/107 vs 1/73; ≥80 years:
8/103 vs 2/92). Of the 17 patients in either phase who
were referred to the study clinic after presenting with a
TIA and who had a stroke within 90 days of their
presenting event, 15 had an ABCD² score of 5 or more.
Rather than decreasing gradually over the 5-year study
period, the risk of recurrent stroke in patients referred
to the study clinic dropped substantially after the first
1437
 Articles
80
TIA
70 Minor stroke
60
Number of patients
50
40
30
20
10
0 Discussion
Mon Tue Wed Thu Fri Sat Sun
Day of the week
Figure 4: Day of presentation for all patients with TIA or minor stroke in the study population who first
sought medical attention via their primary-care physician
30 months of the study (9·3% [14/151] in months 1–15;
11·3% [18/159] in months 16–30; 2·5% [3/121] in
months 31–45; and 1·9% [3/160] in months 46–60).
Early treatment did not increase the 30-day risk of
bleeding events requiring medical attention in referrals to
the study clinic in phase 2. No symptomatic intracerebral
or other intracranial haemorrhages were identified in
either phase of the study, and there was no symptomatic
haemorrhagic transformation of infarction. There were
three episodes of gastrointestinal bleeding in phase 1 (one
major and two minor by CURE trial criteria26) and four
episodes in phase 2 (one major and three minor). There
was one other episode of bleeding (menorrhagia) in
phase 2 that necessitated medical attention.
Process of care in patients referred to the study clinic,
assessed at 1-month follow-up, showed that those referred
in phase 2 were more likely to be taking a statin, to have
been treated with aspirin and clopidogrel, and to be on
blood-pressure-lowering medication than those referred
to the study clinic in phase 1 (table 5). Mean systolic and
diastolic blood pressures at 1 month were lower in
phase 2 than in phase 1 (table 5). Similar numbers of
patients had carotid endarterectomy for recently
symptomatic stenosis during the two periods, but surgery
was done earlier in phase 2 than in phase 1 (table 5).
The number of outcome events was too small to allow
adjustment for all baseline characteristics, but adjustment
for differences in premorbid use of statin and antiplatelet
drugs did not affect the results. Unadjusted hazard ratios
for the rate of recurrence at 90 days in phase 2 versus in
phase 1 were 0·20 (95% CI 0·08–0·48) for those
individuals referred to the study clinic and 0·40
(0·26–0·63) for the whole population. The corresponding
adjusted hazard ratios were 0·20 (0·08–0·49) for those
referred to the study clinic and 0·41 (0·26–0·65) for the
whole population.
1438
All patients who sought medical attention after a TIA
or stroke were included in our analyses. However,
41 patients did not seek medical attention after an initial
TIA or minor stroke, and only presented after a recurrent
stroke that occurred within 90 days of the initial event
(22 in phase 1 and 19 in phase 2). Of the 485 patients
with TIA who did seek medical attention after the initial
event, 210 (43·4%, data missing in one case) delayed
doing so for more than 24 hours. There was little
variation in occurence of TIA or minor stroke by day of
the week (data not shown), but patients were substantially
less likely to seek medical attention at the weekend
(figure 4, p<0·0001).
Our data indicate that urgent assessment and early
initiation of a combination of existing preventive
treatments can reduce the risk of early recurrent stroke
after TIA or minor stroke by about 80%, and reduce the
total number of all early recurrent strokes in the whole
population by over half. Extrapolated across the UK
population, this equates to the prevention of nearly
10 000 strokes per year.
Our study was not a randomised comparison, but it has
produced a reliable estimate of the effect of urgent
treatment of TIA and stroke, and should not be confused
with an historical controlled study. Historical control
studies use control groups that are assembled
retrospectively, with potential for both incomplete
ascertainment of cases and selection bias. Such studies
can sometimes produce unreliable estimates of the
effects of interventions.38–40 By contrast, the control group
in our study (phase 1) was accrued prospectively, and
included all patients presenting to medical attention in
the whole study population, thus minimising any
potential for selection bias. Indeed, phase 1 of the study
was itself a pre-planned prospectively assessed
improvement (ie, daily clinic) on usual secondary care in
the UK (ie, weekly clinic).
Because this study was nested in a rigorous and well
established population-based study of all incident and
recurrent TIA and stroke, with identical methods of case
ascertainment, diagnosis, investigation, and follow-up
in both phases, we could reasonably reliably determine
any clinically significant effect on outcome of the more
rapid assessment and treatment in phase 2. Of particular
importance, we could identify any temporal changes in
referral patterns, patient characteristics, or other
potential sources of bias, and assess the effect of the
clinic on the rate of recurrent stroke in the whole
population, thereby minimising any potential selection
or referral bias.
There was little evidence of such bias. There was no
significant change in rates of referral to the clinic over
the course of the study (table 1 and table 2), nor was there
any change in the delay to seeking medical attention after
a TIA or minor stroke, or in the number of patients
www.thelancet.com Vol 370 October 20, 2007

having early recurrent stroke before seeking medical
attention. Further, there was no change in the methods
(or efficiency) of either our face-to-face follow-up of
patients or our daily ascertainment of all strokes in the
population. There was also no change in our definition of
recurrent stroke, and all outcomes were independently
audited at the end of the study, blind to the phase of the
study.
There was also no evidence of confounding (ie, that
some other factor had changed between phase 1 and
phase 2, in addition to the study intervention). First, the
risk of recurrent stroke seen in phase 1 was very similar
to that seen between 1981 and 1986 in a population-based
study in the same primary-care practices as studied here
(figure 5)41—ie, the risk of early recurrent stroke without
urgent treatment had been stable before the study for at
least 20 years. Second, the rate of recurrence changed in
October, 2004, immediately after the new clinic was
introduced, rather than slowly over the 5 years of the
study. Third, there was no evidence of any substantial
change in the characteristics of patients presenting with
TIA or stroke between the two periods, other than a small
increase in the proportion that received premorbid statin
treatment, which was due mainly to re-presentation of
patients treated in phase 1. There was no significant
difference in premorbid statin use if the analysis was
confined to first presentations in the study period (data
not shown). Moreover, the rate of recurrent stroke was
unrelated to premorbid statin use, and a 10% absolute
increase in use could account for only a small fraction of
the 80% reduction in stroke risk observed.18
Importantly, the observed reduction in risk of recurrent
stroke is also consistent with that predicted to result from
use of the combination of treatments given.22 The low
risk of stroke in phase 2 in patients with TIA who were
referred to the study clinic is also entirely consistent with
data now emerging from other studies in which patients
with TIA are treated urgently and intensively,42–44 whereas
the high rate in phase 1 is consistent with similar studies
in which patients were not treated urgently.41,45–48
The large predicted effect size and our rigorous
population-based sequential comparison approach made
randomisation less necessary than usual.49–51 Moreover,
we did not consider randomisation of individuals to be
feasible—who would consent to possible randomisation
to delayed treatment? Furthermore, we had insufficient
study primary-care practices to allow cluster
randomisation, even if ethical approval could have been
obtained. However, in certain respects our approach has
produced results that are more useful than those that
would be expected from a randomised controlled trial,
especially in relation to external validity.52 Most
randomised trials recruit only a small proportion of
patients in the population with the disease of interest,
sometimes less than 1% and usually less than 10%,52 and
those recruited often differ from those not recruited in
relation to age, sex, ethnic origin, social class, educational
www.thelancet.com Vol 370 October 20, 2007
Articles
16
14
12
Risk of recurrence (%)
10
8
6
4
2
0
1981–86 Phase 1 Phase 2
OCSP EXPRESS (2002–07)
Figure 5: Risk of recurrent stroke during the 90 days after first seeking
medical attention after an incident TIA or stroke in patients referred to the
study clinic in the Oxford Community Stroke Project (OCSP) compared with
that in patients with incident events referred to the study clinic in phase 1
and phase 2 of the EXPRESS study
Error bars are 95% CI.
status, and severity of disease,52 as well as in their
willingness to be randomised. By nesting our study in a
rigorous population-based study of all TIA and stroke we
avoided such selective recruitment. For example, a third
of the patients treated in the study clinic were aged over
80 years (and about 10% were over 90); most of these
individuals would have been excluded from many trial
protocols used in vascular disease.52 The data on the effect
of the intervention on the whole population also facilitate
much more complete assessments of the effects of the
intervention on health economics and on public health.
Our study does, however, have some shortcomings. In
particular, we do not know what the relative effect of each
of the different treatments used was on the risk of
recurrent stroke. One or more of the treatments could
have been ineffective, but it is unlikely that any of the
treatments were harmful, in view of the very low risk of
recurrent stroke and major bleeding in phase 2. Although
more evidence from randomised trials is required for
some individual classes of drug, in view of the substantial
benefit of the combined treatment regime in phase 2 of
the EXPRESS study, it would probably be unethical to
omit more than one element of the combination at a time
in any future randomised comparison.
Simply giving aspirin (with a loading dose of 300 mg)
more quickly after the presenting event probably
accounted for a proportion of the benefit. The relative
reduction in the 14-day risk of recurrent ischaemic
stroke with aspirin in inpatient acute ischaemic stroke
is about 30%,24,25 and could well be larger in patients
with TIA or minor stroke. In patients with unstable
angina or non-Q wave myocardial infarction, initiation
of treatment with a statin within 24–96 hours reduced
the early risk of ischaemic stroke by about 50% compared
with placebo,23 but benefit after TIA and minor stroke is
uncertain. The reduction in delays to carotid
endarterectomy in phase 2 will have substantially
1439
 Articles
improved the effectiveness of the procedure,21 but the
small proportion of patients operated on means that
this cannot have accounted for more than three or four
additional strokes prevented.
The greater use of clopidogrel in phase 2 was due to
the higher proportion of patients seen within the first
48 hours after TIA or minor stroke than in phase 1, when
patients are at very high risk of recurrence, and also
because we prescribed medication in the clinic rather
than simply making a recommendation to the primary
care physician—ie, it was a direct consequence of the two
changes to the study clinic that made up the study
intervention, rather than a confounding factor. There are
some data from randomised trials to suggest that the
combination of clopidogrel and aspirin might be more
effective in the acute phase after TIA and stroke than
either drug alone,28,53,54 but more evidence is required.
More evidence is also needed on the effects of the
combination of aspirin and dipyridamole in the acute
phase after TIA and stroke, given the clear benefit over
aspirin alone in the long term.15,16 More data from
randomised trials on the risks and benefits of blood
pressure lowering in the acute phase after TIA and minor
stroke could also be required, depending on the results
of ongoing randomised trials in the acute phase after
major stroke.
Another shortcoming is that we do not yet have
complete data on the management of all patients referred
directly to hospital. The high rate of early recurrent stroke
in these cases does not necessarily indicate a lack of early
or intensive treatment, especially in patients with major
ischaemic stroke, recurrence after which might be less
amenable to prevention. The proportion of patients with
intracerebral haemorrhage was also higher in cases of
stroke that were referred to hospital than in outpatients.
The high rate of recurrence in hospital-referred patients
with TIA, who mainly self-referred to the A&E
department, is in part a result of the tendency for patients
with high-risk events (eg, motor weakness and long
duration) to present in this way.34 However, further study
of process of care in this group is required.
Our results have several implications for the future
provision of clinical services for TIA and minor stroke.
First, proof that urgent assessment and treatment are
highly effective in reducing the recurrence of stroke after
TIA and minor stroke means that long delays to assessment
in TIA clinics are no longer acceptable. Many health-care
systems still provide care via weekly TIA clinics, audits of
which usually identify long delays to assessment and high
rates of recurrent stroke before the clinic appointment.12,30,55,56
Second, our results highlight the long delays to initiation
of treatment in primary care that can occur if treatment is
not started in secondary care. The high rate of recurrence
in patients referred to the study clinic in phase 1 was clearly
a reflection of the fact that, even if patients sought medical
attention immediately and were seen quickly in the clinic,
there were substantial delays to the subsequent prescription
1440
of treatment in primary care. Even at 1-month follow-up
many patients had not yet received a prescription for a
statin or blood-pressure-lowering drug. Further research is
required into the causes of this delay (eg, patient-related,
system-related, physician-related), but a policy of initiation
of treatment in secondary care, if not already given in
primary care, is clearly sensible in TIA and minor stroke.
Similar shortcomings in the speed of initiation of treatment
and in the proportion of patients appropriately treated after
TIA and stroke have been reported in North America and
elsewhere in Europe.8,10,46,56,57 Third, our results suggest that
in patients with a recent TIA or minor stroke, a combination
of preventive treatments should be initiated as soon as they
seek medical attention, either in primary care or in the
emergency department, with the exception of dual
antiplatelet treatment before brain imaging. Any
inappropriate treatment of patients with non-vascular
conditions could be stopped after subsequent specialist
assessment.
Finally, our results further highlight the need for
public education about the symptoms of TIA and minor
stroke and the need to seek medical attention urgently,
which has been highlighted previously.34,58 A nationwide
telephone survey of adults in the USA revealed
that 3·2% recalled symptoms typical of TIA that were
not brought to medical attention and, of those who
reported a physician-diagnosed TIA, 36% did not recall
seeing a physician in the first 24 hours.58 A study of
377 consecutive patients attending specialist TIA and
stroke clinics in Oxfordshire showed that 40% of those
with minor stroke and 50% of those with TIA delayed
seeking medical attention for more than 24 hours.34
Only 45% of patients thought that the event was a
medical emergency. This lack of a sense of urgency is
perhaps best illustrated by the dramatic fall-off in
patients seeking medical attention after TIA or minor
stroke at the weekend seen here (presumably not
wishing to bother the on-call primary-care service).
Having noted this occurrence during phase 1,34 we
decided not to run the study clinic at weekends in
phase 2. Part of the problem with public education
about TIA and minor stroke has been the difficulty in
conveying a succinct message about how to recognise
an event because of the wide variety of potential clinical
manifestations. However, recent studies of the relation
between the nature of the clinical event and the early
risk of stroke should allow public education to be simple
and more effectively focused.3,45
In conclusion, this rigorous population-based
sequential comparison of all patients, irrespective of
age, presenting with TIA or minor stroke has shown
that early initiation of existing treatments can prevent
about 80% of early recurrent strokes. Further follow-up
is required to determine long-term outcome, but these
results have immediate implications for service
provision and public education about TIA and minor
stroke.
www.thelancet.com Vol 370 October 20, 2007

Contributors
PMR designed the study, obtained funding, reviewed all cases referred to
the EXPRESS study clinic, supervised the analysis of data, and wrote the
manuscript. SAG was responsible for computer programming and data
management. ZM was responsible for data management and did the
analyses. All other authors were responsible for either administration of
the EXPRESS study or assessment of patients. All authors read and
commented on the manuscript.
Conflict of interest statement
PMR has received honoraria for talks, payment for occasional
consultancy, and research funding from several pharmaceutical
companies that manufacture drugs used in the secondary prevention of
stroke. All other authors declare that they have no conflict of interest.
Acknowledgments
We thank all primary-care practices and physicians who collaborated
with OXVASC and EXPRESS studies, details of which have been
published previously.32,33 We thank Charles Warlow for auditing all 90-day
recurrent strokes in the EXPRESS study and for commenting on a draft
of the manuscript. OXVASC is funded by the UK Medical Research
Council, the Dunhill Medical Trust, the Stroke Association, the BUPA
Foundation, the National Institute for Health Research, and the Thames
Valley Primary Care Research Partnership.
References
1 Johnston SC, Gress DR, Browner WS, Sidney S. Short-term
prognosis after emergency department diagnosis of TIA. JAMA
2000; 284: 2901–06.
2 Coull A, Lovett JK, Rothwell PM, on behalf of the Oxford Vascular
Study. Population based study of early risk of stroke after a
transient ischaemic attack or minor stroke: implications for public
education and organisation of services. BMJ 2004; 328: 326–28.
3 Johnston SC, Rothwell PM, Nguyen-Huynh MN, et al. Validation
and refinement of scores to predict very early stroke risk after
transient ischaemic attack. Lancet 2007; 369: 283–92.
4 Giles MF, Rothwell PM. Substantial underestimation of the need
for outpatient services for TIA and minor stroke. Age Ageing 2007;
published online July 26.
5 Kleindorfer D, Panagos P, Pancioli A, et al. Incidence and
short-term prognosis of transient ischemic attack in a
population-based study. Stroke 2005; 36: 720–23.
6 Rothwell PM, Warlow CP. Timing of transient ischaemic attacks
preceding ischaemic stroke. Neurology 2005; 64: 817–20.
7 Johnston SC, Smith WS. Practice variability in management of
transient ischemic attacks. Eur Neurol 1999; 42: 105–08.
8 Goldstein LB, Bian J, Bonito AJ, Lux LJ, Matchar DB. New transient
ischemic attack and stroke: outpatient management by primary
care physicians. Arch Intern Med 2000; 160: 2941–46.
9 Gubitz G, Phillips S, Dwyer V. What is the cost of admitting
patients with transient ischaemic attacks to hospital?
Cerebrovasc Dis 1999; 9: 210–14.
10 Ovbiagele B, Saver JL, Fredieu A, et al. In-hospital initiation of
secondary stroke prevention therapies yield high rates of adherence
at follow-up. Stroke 2004; 35: 2879–83.
11 Royal College of Physicians Intercollegiate Stroke Working Party.
National clinical guidelines for stroke, 2nd edn. London, England:
Royal College of Physicians, 2004: 134.
12 National Audit Office. Reducing brain damage: faster access to
better stroke care. London: Department of Health Publication, 2004.
13 Antithrombotic Trialists Collaboration. Collaborative meta-analysis
of randomised trials of antiplatelet therapy for prevention of death,
myocardial infarction, and stroke in high risk patients. BMJ 2002;
324: 71–86.
14 CAPRIE Steering Committee. A randomised, blinded, trial of
clopidogrel versus aspirin in patients at risk of ischaemic events
(CAPRIE). Lancet 1996; 348: 1329–39.
15 ESPRIT Study Group, Halkes PH, van Gijn J, Kappelle LJ,
Koudstaal PJ, Algra A. Aspirin plus dipyridamole versus aspirin
alone after cerebral ischaemia of arterial origin (ESPRIT):
randomised controlled trial. Lancet 2006; 367: 1665–73.
16 Diener HC, Cunha L, Forbes C, et al. European secondary
prevention study 2: dipyridamole and acetylsalicylic acid in the
secondary prevention of stroke. J Neurol Sci 1996; 143: 113.
www.thelancet.com Vol 370 October 20, 2007
Articles
17 PROGRESS Collaborative Group. Randomised trial of a
perindopril-based blood-pressure-lowering regimen among
6105 individuals with previous stroke or transient ischaemic attack.
Lancet 2001; 358: 1033–41.
18 Amarenco P, Bogousslavsky J, Callahan A III, et al. High-dose
atorvastatin after stroke or transient ischemic attack. N Engl J Med
2006; 355: 549–59.
19 European Atrial Fibrillation Trial Study Group. Secondary
prevention in non-rheumatic atrial fibrillation after transient
ischaemic attack or minor stroke. Lancet 1993; 342: 1255–62.
20 Rothwell PM, Gutnikov SA, Eliasziw M, et al. Pooled analysis of
individual patient data from randomised controlled trials of
endarterectomy for symptomatic carotid stenosis. Lancet 2003;
361: 107–16.
21 Rothwell PM, Eliasziw M, Gutnikov SA, Warlow CP, Barnett HJM,
for the Carotid Endarterectomy Trialists Collaboration. Effect of
endarterectomy for symptomatic carotid stenosis in relation to
clinical subgroups and to the timing of surgery. Lancet 2004;
363: 915–24.
22 Hackam DG, Spence JD. Combining multiple approaches for the
secondary prevention of vascular events after stroke: a quantitative
modeling study. Stroke 2007; 38: 1881–85.
23 Schwartz GG, Olsson AG, Ezekowitz MD, et al. Effects of
atorvastatin on early recurrent ischemic events in acute coronary
syndromes. The MIRACL study: a randomized controlled trial.
JAMA 2001; 285: 1711–18.
24 CAST (Chinese Acute Stroke Trial) Collaborative Group. CAST:
randomised placebo-controlled trial of early aspirin use in
20 000 patients with acute ischaemic stroke. Lancet 1997; 349:
1641–49.
25 International Stroke Trial Collaborative Group. The International
Stroke Trial (IST): a randomised trial of aspirin, subcutaneous
heparin, both or neither among 19 435 patients with acute
ischaemic stroke. Lancet 1997; 349: 1569–81.
26 Yusuf S, Zhao F, Mehta SR, Chrolavicius S, Tognoni G, Fox KK.
Effects of clopidogrel in addition to aspirin in patients with acute
coronary syndromes without ST-segment elevation. N Engl J Med
2001; 345: 494–502.
27 Schrader J, Luders S, Kulschewski A, et al. The ACCESS study:
evaluation of Acute Candesartan Cilexetil Therapy in Stroke
Survivors. Stroke 2003; 34: 1699–703.
28 Markus HS, Droste DW, Kaps M, et al. Dual antiplatelet therapy
with clopidogrel and aspirin in symptomatic carotid stenosis
evaluated using doppler embolic signal detection: the Clopidogrel
and Aspirin for Reduction of Emboli in Symptomatic Carotid
Stenosis (CARESS) trial. Circulation 2005; 111: 2233–40.
29 Lovett JK, Dennis MS, Sandercock PA, Bamford J, Warlow CP,
Rothwell PM. The very early risk of stroke after a first transient
ischaemic attack. Cerebrovasc Dis 2002; 13 (suppl 3): 45.
30 Rothwell PM, Buchan A, Johnston SC. Recent advances in the
management of transient ischaemic attacks and minor ischaemic
strokes. Lancet Neurol 2006; 5: 323–31.
31 Rothwell PM, Gutnikov SA, Mayberg MR, Warlow CP,
Barnett HJM. A pooled analysis of individual patient data from
trials of endarterectomy for symptomatic carotid stenosis: efficacy
of surgery in important subgroups. Stroke 2001; 32: 328.
32 Rothwell PM, Coull AJ, Silver LE, et al. Population-based study of
event-rate, incidence, case fatality and mortality for all acute
vascular events in all arterial territories (Oxford Vascular Study).
Lancet 2005; 366: 1773–83.
33 Rothwell PM, Coull A, Giles MF, et al, for 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: 1925–33.
34 Giles MF, Flossman E, Rothwell PM. Patient behavior immediately
after transient ischemic attack according to clinical characteristics,
perception of the event, and predicted risk of stroke. Stroke 2006;
37: 1254–60.
35 Brott T, Adams HP, Olinger CP, et al. Measurements of acute
cerebral infarction: a clinical examination scale. Stroke 1989; 20:
864–70.
36 Coull AJ, Silver L, Bull L, Giles M, Rothwell PM. Direct assessment
of completeness of ascertainment in a stroke incidence study. Stroke
2004; 35: 2041–45.
1441
 Articles
37 Coull A, Rothwell PM. Under-estimation of the early risk of
recurrence after first stroke by the use of restricted definitions.
Stroke 2004; 35: 1925–29.
38 Sacks H, Chalmers TC, Smith H Jr. Randomized versus historical
controls for clinical trials. Am J Med 1982; 72: 233–40.
39 McKee M, Britton A, Black N, McPherson K, Sanderson C, Bain C.
Methods in health service research. Interpreting the evidence:
choosing between randomised and non-randomised studies. BMJ
1999; 319: 312–15.
40 Concato J, Shah N, Horwitz RI. Randomised, controlled trials,
observational studies, and the hierarchy of research designs.
N Engl J Med 2000; 342: 1887–92.
41 Lovett J, Dennis M, Sandercock PA, Bamford J, Warlow CP,
Rothwell PM. The very early risk of stroke following a TIA. Stroke
2003; 34: e138–40.
42 Lavallée PC, Meseguer E, Abboud H, et al. A transient ischaemic
attack clinic with round-the-clock access (SOS-TIA): feasibility and
effects. Lancet Neurol 2007; published online October 9.
DOI:10.1016/S1474-4422(07)70248-X.
43 Cucchiara BL, Messe SR, Taylor RA, et al. Is the ABCD score useful
for risk stratification of patients with acute transient ischemic
attack? Stroke 2006; 37: 1710–14.
44 Lavallée PC, Meseguer E, Cabrejo L, et al. TIA clinic: description
and first year experience. Cerebrovasc Dis 2007; 23 (suppl 2): 16.
45 Rothwell PM, Giles MF, Flossmann E, et al. A simple score (ABCD)
to identify individuals at high early risk of stroke after a transient
ischaemic attack. Lancet 2005; 366: 29–36.
46 Gladstone DJ, Kapral MK, Fang J, Laupacis A, Tu JV. Management
and outcomes of transient ischaemic attacks in Ontario. CMAJ
2004; 1707: 1099–104.
47 Correia M, Silva MR, Magalhaes R, Guimaraes L, Silva MC.
Transient ischemic attacks in rural and urban northern Portugal:
incidence and short-term prognosis. Stroke 2006; 37: 50–55.
48 Tsivgoulis G, Spengos K, Manta P, et al. Validation of the ABCD
score in identifying individuals at high early risk of stroke after a
transient ischemic attack: a hospital-based case series study. Stroke
2006; 37: 2892–97.
1442
49 Vandenbroucke JP. When are observational studies as credible as
randomized trials? Lancet 2004; 363: 1728–31.
50 Glasziou P, Chalmers I, Rawlins M, McCulloch P. When are
randomized trials unnecessary? Picking signal from noise. BMJ
2007; 344: 349–51.
51 Potts M, Prata N, Walsh J, Grossman A. Parachute approach to
evidence based medicine. BMJ 2006; 333: 701–03.
52 Rothwell PM. External validity of randomised controlled trials: to
whom do the results of this trial apply? Lancet 2005; 365: 82–93.
53 Diener H-C, Bogousslavsky J, Brass LM, et al, on behalf of the
MATCH investigators. Aspirin and clopidogrel compared with
clopidogrel alone after recent ischaemic stroke or transient
ischaemic attack in high-risk patients (MATCH): randomised,
double-blind, placebo-controlled trial. Lancet 2004; 364: 331–37.
54 Kennedy J, Hill MD, Ryckborst KJ, et al. Fast assessment of stroke
and transient ischaemic attack to prevent early recurrence
(FASTER): a randomised controlled trial. Lancet Neurol 2007;
published online October 10. DOI:10.1016/S1474-4422(07)70250-8.
55 Widjaja E, Salam SN, Griffiths PD, Kamara C, Doyle C,
Venables GS. Is the rapid assessment stroke clinic rapid enough in
assessing transient ischaemic attack and minor stroke?
J Neurol Neurosurg Psychiatry 2005; 76: 145–46.
56 Scholte op Reimer WJ, Dippel DW, Franke CL, et al. Quality of
hospital and outpatient care after stroke or transient ischemic
attack: insights from a stroke survey in the Netherlands. Stroke
2006; 37: 1844–49.
57 Mouradian MS, Majumdar SR, Senthilselvan A, Khan K, Shuaib A.
How well are hypertension, hyperlipidemia, diabetes, and smoking
managed after a stroke or transient ischemic attack? Stroke 2002; 33:
1656–59.
58 Johnston SC, Fayad PB, Gorelick PB, et al. Prevalence and
knowledge of transient ischemic attack among US adults. Neurology
2003; 60: 1429–34.
www.thelancet.com Vol 370 October 20, 2007