Abstract Abstract. Boysen G (University of Copenhagen, Copenhagen, Denmark). Bleeding complications in secondary stroke prevention by antiplatelet therapy: a benefit–risk analysis (Review). J Intern Med 1999; 246: 239–245. This review analyses the benefit–risk ratio of antiplatelet drugs in secondary stroke prevention and is based on the published data from eight large stroke prevention trials. In patients with prior transient ischaemic attack (TIA) or stroke, aspirin prevented one to two vascular events (stroke, AMI, or vascular death) per 100 treatment-years with an excess risk of fatal and severe bleeds of 0.4–0.6 per 100 treatment-years. The gastrointestinal bleeding risk was significantly lower with ticlopidine and clopidogrel, which were both somewhat more effective than aspirin in the prevention of vascular events. The combination of dipyridamole and aspirin prevented 2.82 strokes at the expense of an excess risk of 0.61 (95% CI = 0.27–0.95) fatal or severe bleeds per 100 treatment-years. In the acute phase of stroke, the aspirin-associated risk of haemorrhagic complications was much increased compared with that in the stable phase after stroke, with 0.48 (95% CI = 0.13–0.83) fatal or severe bleeds per 100 treated patients for the first 4 weeks after stroke in the Chinese Acute Stroke Trial and 0.41 (95% CI = 0.05–0.77) in the International Stroke Trial. Still, there was a net benefit with the prevention of about one death or non-fatal ischaemic stroke per 100 treated patients.

1. REVIEW
Bleeding complications in secondary stroke prevention by
antiplatelet therapy: a benefit±risk analysis
G. BOYSEN
From the Department of Neurology, Bispebjerg Hospital, University of Copenhagen, Copenhagen, Denmark
Abstract. Boysen G (University of Copenhagen,
Copenhagen, Denmark). Bleeding complications in
secondary stroke prevention by antiplatelet therapy:
a benefit±risk analysis (Review). J Intern Med 1999;
246: 239±245.
This review analyses the benefit±risk ratio of
antiplatelet drugs in secondary stroke prevention
and is based on the published data from eight large
stroke prevention trials. In patients with prior
transient ischaemic attack (TIA) or stroke, aspirin
prevented one to two vascular events (stroke, AMI,
or vascular death) per 100 treatment-years with an
excess risk of fatal and severe bleeds of 0.4±0.6 per
100 treatment-years. The gastrointestinal bleeding
risk was significantly lower with ticlopidine and
clopidogrel, which were both somewhat more
effective than aspirin in the prevention of vascular
events. The combination of dipyridamole and aspirin
prevented 2.82 strokes at the expense of an excess
risk of 0.61 (95% CI = 0.27±0.95) fatal or severe
bleeds per 100 treatment-years.
In the acute phase of stroke, the aspirin-associated
risk of haemorrhagic complications was much
increased compared with that in the stable phase
after stroke, with 0.48 (95% CI = 0.13±0.83) fatal
or severe bleeds per 100 treated patients for the first
4 weeks after stroke in the Chinese Acute Stroke
Trial and 0.41 (95% CI = 0.05±0.77) in the
International Stroke Trial. Still, there was a net
benefit with the prevention of about one death or
non-fatal ischaemic stroke per 100 treated patients.
Keywords: antiplatelet therapy, aspirin, benefit±risk
analysis, haemorrhagic complications, secondary
stroke prevention, stroke prevention trial.
Introduction
Several antiplatelet drugs are available for second-
ary stroke prevention and each drug has its own
spectrum of side-effects. Bleeding is one of the most
important side-effects, which may limit or offset a
beneficial effect of the intervention. From presenta-
tions of odds ratios or percentages of adverse events
in the individual trials, it is often difficult to get a
clear impression of the benefit±risk ratio of a given
treatment. Focusing on antiplatelet therapy for
stroke prevention in patients with previous transient
ischaemic attack (TIA) or stroke, this review will
analyse the benefit±risk relation by presenting,
whenever possible, the number of vascular events
avoided by a given therapy per 100 treatment-years
based on placebo-controlled trials confronted with
the excess number of severe or fatal bleeds per 100
treatment-years. In trials comparing two treatment
regimens without a placebo group, the rate of
vascular events and the rate of bleeding episodes
will be given.
Bleeding complications have been given fairly
little attention in aspirin treatment in general. The
Journal of Internal Medicine 1999; 246: 239±245
# 1999 Blackwell Science Ltd 239

2. first hint of the problem was given by the
Physicians' Health Study [1], where more haemor-
rhagic strokes occurred in the aspirin-treated group
than in the placebo group. Although this was not
statistically significant, it was a contributing factor
in the decision not to advocate aspirin as a primary
preventive measure. Because the risk of haemor-
rhagic complication is small, it takes large studies to
detect it. Bleeding complications do, however,
become of importance when the large-scale use of
aspirin or other antithrombotic drugs is contem-
plated.
This review is based on the published data from
the following large stroke prevention trials: Swedish
Aspirin Low-dose Trial [2], Dutch TIA Trial [3],
European Stroke Prevention Study 2 (ESPS 2) [4],
Canadian American Ticlopidine Study (CATS) [5],
Ticlopidine Aspirin Stroke Study (TASS) [6], Clopi-
dogrel vs. Aspirin in Patients at Risk of Ischaemic
Events (CAPRIE) [7], International Stroke Trial (IST)
[8], Chinese Acute Stroke Trial (CAST) [9], and on
the Antiplatelet Trialists' Collaboration data from
1988 [10] and 1994 [11]. The data on haemor-
rhagic complications as well as on recurrent strokes
in these studies are of variable quality. However,
they provide the best sources of information avail-
able on these important questions. Generally, for
intracerebral haemorrhage to be diagnosed, a CT
scan was required, and such bleedings were rated as
severe. Asymptomatic haemorrhagic transformation
of cerebral infarcts was not likely to be counted as a
haemorrhagic complication, since the event ana-
lysed was a recurrent stroke. For each trial the
definitions used will be quoted. Severe systemic
haemorrhagic complications usually mean a bleed
requiring either a transfusion or an operation, or a
fatal bleed.
Results
In an overview of 13 individual studies in patients
with prior stroke or TIA [11], antiplatelet therapy
conveyed a reduction of risk of vascular events
(stroke, AMI or vascular death) from 22.2 to 18.4%
over 33 months, an absolute risk reduction of 3.8%
and a relative risk reduction of 17%. The risk
reduction of non-fatal stroke alone was 20%. The
usually cited reduction of 22% is the odds reduction,
which is different from the relative risk reduction.
With antiplatelet treatment, 1.3 events of AMI,
stroke or vascular death were avoided per 100
treatment-years, and 0.7 strokes were avoided per
100 treatment-years. The risk of haemorrhagic
stroke in all the studies compiled by the Antiplatelet
Trialists' Collaboration, including those in patients
with AMI as qualifying event, was 0.3% in the
antiplatelet group, significantly higher than in the
control group, where it was 0.2%, 2P , 0.05. In
patients with TIA or stroke as qualifying event, the
risk of haemorrhagic stroke was 0.6% in the
antiplatelet group compared with 0.4% in the
control group. These haemorrhagic strokes were
included in the total number of strokes and therefore
do not detract from the effect cited above. Of all
strokes, 13% were haemorrhagic in the antiplatelet
group compared with 7% in the control group.
There were no data on systemic haemorrhages. In a
derived meta-analysis [12] excluding studies on
antiplatelet drugs other than aspirin, the relative
risk reduction of vascular events was found to be
13%, which means the avoidance of approximately
1.05 vascular events per 100 treatment-years. It is
not possible from any of these meta-analyses to get a
clear impression of the magnitude of bleeding
complications. Later studies have more precisely
described the risk of bleeding associated with
antiplatelet treatment.
The Swedish Aspirin Low-dose Trial [2] compared
aspirin 75 mg daily with placebo in patients with
TIA or minor stroke. In that study, CT scan or
necropsy was carried out in 98% of patients who
had a fatal or non-fatal stroke. There were six fatal
intracranial bleeds in the aspirin group compared
with nil in the placebo group. Gastrointestinal
bleeding was not clearly defined; nine instances
were listed as severe in the aspirin group compared
with four in the placebo group. Table 1 shows that
two vascular events were avoided per 100 treat-
ment-years at the expense of an excess of 0.6 (95%
CI = 0.04±1.20) severe or fatal bleeds. The Dutch
TIA Trial [3] compared 30 mg aspirin with 283 mg
aspirin daily in 3131 patients (Table 2). Fatal
bleeding had to be documented by convincing
clinical evidence or autopsy. Non-fatal bleeds were
considered major if a hospital visit and treatment
were necessary. Incidences of major bleeding events
of 2.6% in the 30 mg group and 3.2% in the
283 mg group were reported. Calculated as events
per 100 treatment-years, given a mean duration of
follow-up of 31 months, the risk of major bleeds was
240 G. BOYSEN
# 1999 Blackwell Science Ltd Journal of Internal Medicine 246: 239±245

3. REVIEW: BLEEDING IN SECONDARY STROKE PREVENTION 241
# 1999 Blackwell Science Ltd Journal of Internal Medicine 246: 239±245
Table1Antiplateletrandomizedplacebo-controlledtrialsinpatientswithtransientischaemicattack(TIA)andischaemicstroke
TrialsDrugsn
Patient-years
atrisk
Outcome
eventsper100
treatment-years(n)
Vascular
eventsper100
treatment-years
Vascular
events
avoided(n)
Severe
orfatal
bleeds(n)
Bleeds
per100
treatment-years
Excessbleeds
per100
treatment-years95%CI
Strokes,MI,VD
SALTAspirin75mg
daily
67618021377.602.05201.110.620.04±1.20
Placebo6841824176*9.6590.49
Strokes
ESPS2Aspirin50mg16493298206*6.251.33200.610.400.09±0.71
Dipyridamole
retard200mg
twicedaily
16543308211*6.381.2060.180.610.27±0.95
Aspirin+
dipyridamole
16503300157*4.742.82*270.82
Placebo164932982507.58±70.21
Strokes,MI,VD
CATSTiclopidine
500mgdaily
5256837410.84.520.290.16±0.29±0.61
Placebo52877311815.310.13
*Statisticallysignificantdifference,P<0.05.
VD,vasculardeath.

4. 0.99 and 1.3 per 100 treatment-years, respectively.
Without a placebo group, the excess risk is
unknown, as well as the number of vascular events
avoided. However, this study shows that a dose of
aspirin as low as 30 mg daily is associated with an
appreciable risk of bleeding complications.
ESPS 2 [4] evaluated the effect of dipyridamole,
aspirin, and dipyridamole plus aspirin versus place-
bo in patients with TIA or ischaemic stroke. Bleeding
episodes were recorded as mild, moderate, severe or
fatal, but were not otherwise defined. As shown in
Table 1, aspirin prevented 1.33 strokes per 100
treatment-years at the expense of an excess risk of
0.4 (95% CI = 0.09±0.71) severe or fatal bleeds.
Dipyridamole alone prevented 1.2 strokes without
any excess bleeds. The combination of dipyridamole
and aspirin prevented 2.82 strokes at the expense of
an excess of 0.61 (95% CI = 0.27±0.95) severe or
fatal bleeds. The risk of bleeding complications was
maintained over the 2 years of treatment, in
contrast to the side-effects to dipyridamole, head-
ache and diarrhoea, which mainly occurred in the
early phase of treatment.
Ticlopidine [5], another antiplatelet drug, signifi-
cantly reduced stroke risk in patients with ischaemic
strokes when compared with placebo (Table 1). The
risk of bleeding seemed to be twice as high as in the
placebo group, 34 events vs. 16, but severe bleeds
were rare in both groups. However, there was no
clear definition of severe bleeds, and it may have
differed from other studies. In TASS, ticlopidine
500 mg daily was compared with aspirin 1300 mg
daily (Table 2) [6]. In this study, peptic ulcer and
gastrointestinal haemorrhage were listed amongst
adverse experiences without precise definitions and
the severity of intracerebral haemorrhage was not
described. The ticlopidine group had a significantly
lower rate of vascular events, and a lower gastro-
intestinal bleeding risk, than that of aspirin, whilst
the number of intracerebral bleeds was equal. A rare
but severe adverse event, granulocytopenia, is
associated with ticlopidine. Clopidogrel [7] was
compared with aspirin 325 mg daily in a large
study in which the study population was composed
of one-third with TIA and ischaemic stroke, one-
third with AMI and one-third with peripheral
vascular disease. The risk of vascular events was
slightly, but significantly, lower in the clopidogrel
group than in the aspirin group. However, when the
group which entered the study with TIA and
ischaemic stroke was analysed separately, the risk
reduction by clopidogrel was not statistically sig-
nificant, nor was the occurrence of ischaemic stroke
in all patients significantly lower in the clopidogrel
arm. The gastrointestinal and intracerebral haemor-
rhages were reported as `any ever occurring' or
`severe' without any further definition of severity.
Clopidogrel was accompanied by a lower risk of
gastrointestinal bleeding than aspirin (Table 2),
whereas the risk of intracerebral haemorrhage was
not significantly different. Clopidogrel may be
considered an alternative to aspirin, but can hardly
be said to be superior to aspirin in patients with TIA
and ischaemic stroke.
Studies comparing antithrombotic therapy with
placebo are getting rarer, since it is now considered
unethical to withhold antithrombotic therapy from
patients at risk of ischaemic stroke. However, the IST
242 G. BOYSEN
# 1999 Blackwell Science Ltd Journal of Internal Medicine 246: 239±245
Table 2 Antiplatelet randomized aspirin-controlled trials in patients with transient ischaemic attack (TIA) and ischaemic stroke
Trials Drugs n
Patient-
years
at risk
Outcome
events (n)
Vascular events
per 100
treatment-years
Severe or
fatal bleeds (n)
Bleeds per 100
treatment-years
Strokes, MI, VD
Dutch TIA Trial Aspirin 30 mg daily 1555 4017 228 5.67 40 1.00
Aspirin 283 mg daily 1576 4071 240 5.89 53 1.30
TASS Ticlopidine 500 mg daily 1529 5014 306 6.10 14 0.28
Aspirin 1300 mg daily 1540 5008 349 6.97* 28 0.56
CAPRIE Clopidogrel 75 mg daily 9599 17636 939 5.32 77 0.44
Aspirin 325 mg daily 9586 17519 1021 5.83* 109 0.62
*Statistically significant difference, P < 0.05.
VD, vascular death.

5. [8] and CAST [9] studies of acute stroke did contain
large groups of placebo-treated patients providing
ample evidence of the bleeding risk associated with
aspirin in the acute phase of stroke. In IST (Table 3),
symptomatic intracranial haemorrhage within
14 days was a secondary outcome as well as major
extracranial haemorrhage including any bleed that
required transfusion or caused death within
14 days. Fourteen days of aspirin 300 mg daily
following acute stroke prevented 1.2 ischaemic
strokes or strokes of unknown aetiology per 100
patients treated at the expense of an excess of 0.41
(95% CI = 0.05±0.77) intracerebral haemorrhages,
transfused or fatal bleeds. In CAST [9], 160 mg of
aspirin for 4 weeks after acute stroke (Table 3)
prevented 0.89 deaths or non-fatal ischaemic
strokes per 100 patients treated at the expense of
0.48 (95% CI = 0.13±0.83) severe or fatal intra-
cranial or extracranial bleeds.
The relationship between gastrointestinal bleed-
ing and aspirin intake has been dealt with in several
studies. The UK-TIA trial [13] reported that the risk
of gastrointestinal bleeding was significantly higher
on 1200 mg aspirin daily than on 300 mg. A
review of the gastrointestinal toxicity of aspirin [14]
included all vascular disease prevention trials from
the Antiplatelet Trialists' Collaboration in which
direct aspirin to placebo comparison was possible.
An odds ratio of 2.0 (99% CI = 1.5±2.8) for
gastrointestinal bleeding was found. The odds ratio
for developing peptic ulcer was 1.3 (99% CI = 1.0±
1.6). From this analysis, however, it is not possible
to extract the number of severe and fatal bleeds for
any given period of treatment. The risk of hospita-
lization for bleeding peptic ulcer was analysed in
1121 patients [15], who were matched with
hospital and community controls. In the peptic
ulcer group, 12.8% had been regular users of aspirin
compared with 9.0% of the hospital controls and
7.8% of the community controls. Odds ratios for
gastrointestinal bleeding were significantly raised for
all doses of aspirin between 75 and 300 mg daily.
Addition of non-steroidal anti-inflammatory drug to
aspirin increased the risk of gastric bleeding. Use of
enteric-coated or buffered aspirin [16] may not
reduce the risk of bleeding, as these formulations in
a case±control study were found to be as likely to
cause gastrointestinal bleeding as plain aspirin;
however, the number of cases taking enteric coated
aspirin was small.
REVIEW: BLEEDING IN SECONDARY STROKE PREVENTION 243
# 1999 Blackwell Science Ltd Journal of Internal Medicine 246: 239±245
Table3Antiplateletrandomizedplacebo-controlledtrialsinacuteischaemicstrokeduring2±4weeks
TrialsDrugsn
Strokesof
ischaemicor
unknownaetiology
Eventsper100
treatedpatients
Eventsavoided
per100patients
Severeor
fatalbleeds
Bleedsper100
treatedpatients
Excessbleeds
per100
treatedpatients95%CI
ISTAspirin300mgdaily
Placebo
4858
4859
156
214
3.2
4.4*
1.249
29
1.01
0.60
0.410.05±0.77
Deathornon-fatalischaemicstroke
CASTAspirin160mgdaily
Placebo
10335
10320
430
521
4.16
5.04*
0.89201
151
1.94
1.46
0.480.13±0.83
*Statisticallysignificantdifference,P<0.05.
VD,vasculardeath.

6. Discussion
This review emphasizes that antithrombotic therapy
is associated with an increased risk of bleeding
complications. Even without any antithrombotic
intervention, the stroke-prone population carries a
certain risk of severe or fatal bleeds. Aspirin is the
most widely used antithrombotic drug, and unfortu-
nately its benefit±risk ratio is quite small. It is
therefore of utmost importance that doctors who
advise patients to take aspirin inquire about history
of dyspepsia and peptic ulcer and institute proper
treatment before embarking on antithrombotic
therapy. Other NSAID drugs should be avoided
and the patients adequately informed about the
increased risk, when these two types of NSAID drugs
are combined.
The dose±response effect of aspirin with regard to
bleeding complications that was found in individual
studies [3, 13] comparing two doses of aspirin was
not apparent across the different studies in this
review. The TASS [6], investigating 1300 mg of
aspirin daily, had approximately the same rate of
severe bleeding events as the ESPS 2 [4], testing a
dose of 50 mg aspirin daily, whilst the intermediate
doses of aspirin had a somewhat higher rate of
bleeding episodes. This lack of dose±response rela-
tionship could possibly be due to variations in the
definition of severe bleeding events. It might also
reflect that once the platelet cyclooxygenase is
inhibited, there is little further effect of increasing
doses of aspirin.
We now have sufficient data to state that aspirin
in stroke prevention is associated with the avoidance
of 1±2 vascular events per 100 treatment-years and
with the occurrence of 0.6±1.3 fatal and severe
bleeds per 100 treatment-years, and an excess risk of
these bleeds of 0.4±0.6 per 100 treatment-years.
The bleeding risk is slightly lower with ticlopidine
and clopidogrel. The best benefit±risk ratio amongst
the above studies was obtained by ticlopidine in the
CATS study with 4.5 vascular events avoided per
100 treatment-years at the expense of an excess risk
of bleeding of 0.16. However, the CATS study
differed from the other studies by having an
exceptionally high rate of ischaemic events in the
placebo group and by an exceptionally low bleeding
risk. Further, the analysis was not by intention to
treat. Dipyridamole did not in itself increase bleeding
risk, and also the ESPS 2, by the combination of
dipyridamole and aspirin, had a favourable benefit±
risk ratio, with 2.82 strokes avoided at the expense
of an excess risk of bleeds of 0.61.
Future antithrombotic agents in secondary stroke
prevention will have to be tested against aspirin or
another known drug. It is therefore of great
importance to know as precisely as possible the risk
reduction as well as the risk of side-effects of the
drug to which new therapies are compared.
It is likely that the higher the thromboembolic
event rate in a given population, the more favour-
able will be the benefit±risk ratio. If the event rate is
low, the number of events that can be prevented will
be low and the benefit±risk ratio may approach
unity. It is therefore unlikely that aspirin will ever be
beneficial as a primary preventive measure.
In the acute phase of stroke, the risk of haemor-
rhagic complications is much increased compared
with that in the stable phase weeks or months after
a stroke. In IST and CAST, the 2- and 4-week rate of
excess severe bleeds was about 0.4 per 100 patients
treated, which is similar to the risk during a whole
year of aspirin treatment in the stable phase. Thus,
prevention of stroke and other vascular events by
aspirin, although economically very favourable, has
a cost in the way of severe or fatal bleeds. An ideal
alternative antithrombotic therapy should be more
effective than aspirin with fewer side-effects at an
affordable price.
Conclusion
Antiplatelet therapy in patients with prior TIA or
ischaemic stroke has a positive benefit±risk ratio.
Aspirin prevents about 1±2 vascular events per 100
treatment-years at the expense of an excess risk of
0.4±0.6 fatal or severe bleeds. In this patient group,
ticlopidine is more effective than aspirin, but has
more serious side-effects. Clopidogrel is an alter-
native to aspirin which is not significantly more
effective. Both drugs have a slightly lower bleeding
risk than aspirin.
References
1 Physicians' Health Study Research Group. Final report on
the aspirin component of the ongoing physicians' health
study. N Engl J Med 1989; 321: 129±35.
2 SALT Collaborative Group. Swedish aspirin low-dose trial
(SALT) of 75 mg aspirin as secondary prophylaxis after
cerebrovascular events. Lancet 1991; 338: 1345±49.
244 G. BOYSEN
# 1999 Blackwell Science Ltd Journal of Internal Medicine 246: 239±245

7. 3 Dutch TIA Trial Study Group. A comparison of two doses of
aspirin (30 mg vs. 283 mg a day) in patients after a
transient ischaemic attack or minor ischaemic stroke. N Engl
J Med 1991; 325: 1261±66.
4 Diener HC, Cunha L, Forbes C, Sivenius J, Smets P,
LoÈwenthal A. European Stroke Prevention Study 2. Dipyr-
idamole and acetylsalicylic acid in the secondary prevention
of stroke. J Neurol Sci 1996; 143: 1±13.
5 Gent M, Easton JD, Hachinski VC et al. The Canadian
American Ticlopidine Study (CATS) in thromboembolic
stroke. Lancet 1989; 1: 1215±20.
6 Hass WK, Easton JD, Adams HP et al. For the Ticlopidine
Aspirin Stroke Study Group. A randomised trial comparing
ticlopidine hydrochloride with aspirin for the prevention of
stroke in high-risk patients. N Engl J Med 1989; 321: 501±
507.
7 CAPRIE Steering Committee. A randomised, blinded, trial of
clopidogrel versus aspirin in patients at risk of ischaemic
events (CAPRIE). Lancet 1996; 348: 1329±39.
8 International Stroke Trial Collaborative Group. The Inter-
national 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.
9 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.
10 Antiplatelet Trialists' Collaboration. Secondary prevention of
vascular events by prolonged antiplatelet therapy. Br Med J
1988; 296: 320±31.
11 Antiplatelet Trialists' Collaboration. Collaborative overview
of randomised trial of antiplatelet therapy ± I. Prevention of
death, myocardial infarction, and stroke by prolonged
antiplatelet therapy in various categories of patients. Br
Med J 1994; 301: 81±106.
12 Algra A, van Gijn J. Aspirin at any dose above 30 mg offers
only modest protection after cerebral ischaemia. J Neurol
Neurosurg Psychiat 1996; 60: 197±99.
13 UK-TIA Study Group. United Kingdom transient ischaemic
attack (UK-TIA) aspirin trial: interim results. Br Med J 1988;
296: 316±20.
14 Roderick PJ, Wilkes HC, Meade TW. The gastrointestinal
toxicity of aspirin: an overview of randomised controlled
trials. Br J Clin Pharmac 1993; 35: 219±26.
15 Weil J, Colin-Jones D, Langman M et al. Prophylactic aspirin
and risk of peptic ulcer bleeding. Br Med J 1995; 310: 827±
30.
16 Kelly JP, Kaufman DW, Jurgelon JM, Sheehan J, Koff RS,
Shapiro S. Risk of aspirin-associated major upper-gastro-
intestinal bleeding with enteric-coated or buffered product.
Lancet 1996; 348: 1413±16.
Received 2 June 1998; accepted 29 December 1998.
Correspondence: Professor Gudrun Boysen MD, Department of
Neurology, Bispebjerg Hospital, University of Copenhagen,Copen-
hagen, Denmark (fax: + 45 35312595).
REVIEW: BLEEDING IN SECONDARY STROKE PREVENTION 245
# 1999 Blackwell Science Ltd Journal of Internal Medicine 246: 239±245