Saturday 13 August 1988

RANDOMISED TRIAL OF INTRAVENOUS 0·2%) and of confirmed cerebral haemorrhage (0·1% vs

STREPTOKINASE, ORAL ASPIRIN, BOTH, OR 0·0%), but with fewer other strokes (0·6% vs 0·8%). These
NEITHER AMONG 17 187 CASES OF "other" strokes may have included a few undiagnosed
SUSPECTED ACUTE MYOCARDIAL cerebral haemorrhages, but still there was no increase in
INFARCTION: ISIS-2 total strokes (0·7% streptokinase vs 0·8% placebo infusion).
Aspirin significantly reduced non-fatal reinfarction (1·0%
ISIS-2 (SECOND INTERNATIONAL STUDY OF vs 2·0%) and non-fatal stroke (0·3% vs 0·6%), and was not
INFARCT SURVIVAL) COLLABORATIVE GROUP* associated with any significant increase in cerebral
haemorrhage or in bleeds requiring transfusion. An excess
17 187 patients entering 417 hospitals up to of non-fatal reinfarction was reported when streptokinase
Summary
24 hours (median 5 hours) after the onset of was used alone, but this appeared to be entirely avoided by
the addition of aspirin. Those allocated the combination of
suspected acute myocardial infarction were randomised,
with placebo control, between: (i) a 1-hour intravenous streptokinase and aspirin had significantly fewer
infusion of 1·5 MU of streptokinase; (ii) one month of 160 reinfarctions (1·8% vs 2·9%), strokes (0·6% vs 1·1%), and
mg/day enteric-coated aspirin; (iii) both active treatments; deaths (8·0% vs 13·2%) than those allocated neither. The
or (iv) neither. Streptokinase alone and aspirin alone each differences in vascular and in all-cause mortality produced
produced a highly significant reduction in 5-week vascular by streptokinase and by aspirin remain highly significant
mortality: 791/8592 (9&middot;2%) among patients allocated (2p<0&middot;001 for each) after the median of 15 months of
streptokinase infusion vs 1029/8595 (12&middot;0%) among those follow-up thus far available.
allocated placebo infusion (odds reduction: 25% SD 4;
Introduction
2p<0&middot;00001); 804/8587 (9&middot;4%) vascular deaths among
patients allocated aspirin tablets vs 1016/8600 (11&middot;8%) REDUCTIONS in mortality that are realistically moderate
among those allocated placebo tablets (odds reduction: 23% (eg, "only" 20-25%) are important, especially if produced
SD 4; 2p <0&middot;00001). The combination of streptokinase and by widely practicable treatments for common causes of
aspirin was significantly (2p<0&middot;0001) better than either death, but reliable assessment of them may require strict
agent alone. Their separate effects on vascular deaths randomisation of several thousand patients. The Second
appeared to be additive: 343/4292 (8&middot;0%) among patients International Study of Infarct Survival (ISIS-2) has
allocated both active agents vs 568/4300 (13&middot;2%) among randomly assessed, with placebo control, the separate and
those allocated neither (odds reduction: 42% SD 5; 95% combined effects of intravenous streptokinase (a single
confidence limits 34-50%). There was evidence of benefit infusion of 1 5MU over about one hour) and of oral aspirin
from each agent even for patients treated late after pain onset (160 mg/day for a month) in 17 187 patients with suspected
(odds reductions at 0-4, 5-12, and 13-24 hours: 35% SD 6, acute myocardial infarction (MI).
16% SD 7, and 21% SD 12 for streptokinase alone; 25%
SD 7, 21% SD 7, and 21% SD 12 for aspirin alone; and Streptokinase
53% SD 8, 32% SD 9, and 38% SD 15 for the combination
of streptokinase and aspirin). Streptokinase was associated During the 1960s and 1970s several trials of fibrinolytic
with an excess of bleeds requiring transfusion (0&middot;5% vs therapy (chiefly involving intravenous streptokinase) were
conducted. These studies were so small-none involved
more than 750 patients-that they yielded apparently

conflicting results, but an overview of their findings

indicated that fibrinolytic therapy could reduce mortality by
*Collaborators and participating centres are listed at the end of the report.
about a quarter.1 The prolonged regimens previously tested
Address for correspondence: ISIS, Clinical Trial Service Unit and
were not particularly convenient, and so a rapid high-dose
Department of Cardiovascular Medicine, Radcliffe Infirmary, Oxford OX2
6HE, UK. intravenous regimen known to dissolve coronary artery
8607 &copy;
 350

thrombi and recanalise occluded coronary arteries2 was government records wherever possible. As a result, 417 hospitals in
16 counties (see acknowledgments) randomised a total of 17 187
adopted by the two large trials (GISSP and the present
study) conducted to test the results of the overview. Animal patients.
models for infarction suggest that fibrinolytic therapy might
Treatment
be of value only if started within a few hours after the onset of
pain.4 But, evidence from the previous trials1 had indicated A 2 x 2 factorial study design was used.14 Half of all patients were
that mortality was reduced not only among patients treated allocated randomly to receive streptokinase (1.5MU of’Streptase’)
and half to receive matching placebo (hepatitis-B-antigen-free
early (eg, 0-6 hours after the onset of pain) but also among
those treated later (eg, between 6-12 hours, or even after a albumin), infused intravenously over about 1 hour in 50-250 ml
physiological saline, starting immediately. Half of all patients were
delay of more than 12 hours). Many patients do not reach also allocated randomly to receive oral aspirin (exact dose: 162-5 mg
hospital until several hours after the onset of chest pain, so in enteric-coated tablets) and half to receive matching placebo
reliable information is needed about any "time window" for (enteric-coated starch tablets), given daily for one month from a
benefit from fibrinolytic therapy. In ISIS-2, therefore, calendar pack, starting immediately with the first tablet crushed,
patients were eligible up to 24 hours from the onset of pain, sucked, or chewed for a rapid antiplatelet effect. Hence, there were 4
though the aim was always to treat them as promptly as treatment groups: streptokinase alone, aspirin alone, both, or neither.
The trial treatments were to be interrupted only if this was
possible.
thought to be clearly indicated by the responsible physician. In all
other respects, physicians were free to use whatever additional
Aspirin therapy they considered necessary. For example, although
An overview of the ten randomised trials of long-term physicians were asked to state, just before randomising each patient,
aspirin or other antiplatelet agents among patients with a whether or not they "planned" to add anticoagulants to the trial
treatments for that particular patient, their plan could be altered if
history of previous MI found a 25% reduction in "serious
vascular events" (reinfarction, stroke, or vascular death), some contraindication developed.
and two trials in unstable angina produced similar results.5
There was, however, little direct evidence on antiplatelet Eligibility
therapy in acute MI, for the only randomised trial6 was Patients were eligible if they were thought to be within 24 hours of
small, and involved just one single aspirin tablet, with no the onset of symptoms of suspected MI, and to have no clear
further treatment. Aspirin is the most convenient and indication for, or contraindication to, streptokinase or aspirin.
widely tested antiplatelet agent, and irreversibly inhibits (Absolute contraindications at the start of ISIS-2 were any history
of stroke or of gastrointestinal haemorrhage or ulcer, although in
cyclo-oxygenase-dependent platelet aggregation.Although
retrospect this may have been too restrictive. Possible
40 mg/day will eventually achieve virtually complete contraindications included recent arterial puncture, recent severe
inhibition, larger doses (such as the 160 mg dose used in trauma, severe persistent hypertension, allergy to streptokinase or
ISIS-2) are needed for a rapid effect8 on the first day of aspirin, low risk of cardiac death, or some other life-threatening
treatment. Much larger doses may have little or no disease.) The fundamental criterion for entry was that the
additional antithrombotic effect, and are more gastrotoxic.9 responsible physician was uncertain whether, for a particular
ISIS-2 trial tablets were continued for only the first month, patient, treatment with streptokinase or with aspirin was indicated.
since previous trials had already shown that long-term ECG changes at entry were not a requirement.
antiplatelet therapy started after the first few weeks is Randomisation
beneficial.
Entry to the study was by a 24-hour telephone service, based in
Streptokinase and Aspirin Berlin for Germany, Gent and Bruxelles for Belgium, Valencia for
Spain, Bellinzona for Austria and Switzerland, Lyon for France,
After fibrinolytic therapy, recanalised coronary arteries10 and Oxford for all other countries. Before randomisation some
might be particularly prone to reocclusion. So, in principle, details were recorded (directly on a computer in Oxford and Lyon,
any early benefits might not persist unless reocclusion can be or first on computer-generated randomisation lists elsewhere),

avoided,4,11 perhaps by anticoagulants, by antiplatelet including patient identifiers, age, systolic blood pressure, hours
agents, or by angioplasty. But, both in the previous trials of from onset of the episode of pain that led to admission, aspirin use
prolonged fibrinolytic therapy1 and in the recent GISSI trial during the week before entry, and "planned" treatment in hospital
of high-dose intravenous streptokinase,3fibrinolytic therapy (ie, whether non-trial treatment was likely to include any aspirin,
appeared to reduce mortality even if anticoagulants were not anticoagulants, or intravenous beta-blockers). If any pre-
randomisation details were incomplete then randomisation was not
used. Hence, no fixed rules about anticoagulation were to be issued, and such patients were not part of the trial. In Oxford
made in ISIS-2-instead, collaborators were merely asked and Lyon, the computer allocated treatment using a
at randomisation of each patient whether or not they planned "minimisation" algorithm" to help avoid any chance differences
to use anticoagulants. This does not assess the effects of between the treatment groups in prognostic features recorded at
anticoagulation, but merely determines whether entry. On Jan 24, 1986, a programming error led to more of the
streptokinase is equally effective whether or not patients randomised at Oxford during a period of about 2 months
anticoagulants are used. The "factorial" design adopted in being allocated to placebo infusion and placebo tablets. Correction
ISIS-2 (see Methods) does, however, allow separate of this programming error restored exact balance by August, 1986.
assessment of the effects of streptokinase and of aspirin (and
(The apparent effects of streptokinase and of aspirin among patients
randomised in Oxford between Jan 24 and Aug 31, 1986, were,
can determine whether much of the increased risk of
however, the same as in all other patients.) After allocation of a
reinfarction observed3,12 following fibrinolytic therapy could specific treatment pack containing active or placebo trial treatments,
be avoided by aspirin). the patient was irrevocably in the trial. Whether or not the treatment
in the pack was actually given, patients remained in their originally
allocated treatment group for an "intention-to-treat" analysis, and
Patients and Methods the few second entries were disregarded.
To encourage recruitment, the trial procedures were as simple as
possible13-randomisation involved only a telephone call and no
Discharge
forms, the use of ancillary treatments was not restricted, and At discharge, a prerandomisation electrocardiogram (ECG) and a
follow-up after discharge involved only mortality, through simple single-sided form were returned to the trial office. This
 351

"discharge form" provided further identifiers to assist central Follow-up

mortality follow-up after discharge, as well as brief details of The present report is of outcome by allocated treatment among
compliance with study treatments in hospital, other drug use in all randomised patients, except those 206 (102 active SK vs 104
hospital and at discharge, any apparent side-effects of treatment,
and major events in hospital (bleeding, cardiac rupture, placebo infusion; 95 active aspirin vs 111 placebo tablets) for whom
reinfarction, cardiac arrest, stroke, and death). Each ECG was read discharge forms had not yet been obtained by July, 1988. Discharge
was at a median of 10 days, and mortality follow-up was for a
"blind" of treatment by three observers (with adjudication by the
trial coordinator if no two agreed). If bundle branch block was maximum of 34 and a median of 15 months. The completeness of
present (6% of ECGs) this alone was noted, otherwise the ECG follow-up is 99% to discharge, 97% to week 5, and 96% to Jan 1,
1988. (About nine-tenths of all deaths in the first 5 weeks occur in
categories (as in ISIS-116) were:
hospital, so it is probable that more than 98% of the 5-week deaths
Inferior ST elevation (29%).&ge;23mm in the sum of among the 17 187 randomised patients are included in the present
II + III +aVF; analysis.)
Anterior ST elevation (25%).-(a) &ge;6mm in the sum of All deaths were reviewed blind of treatment allocation by the trial
V1 + V2 + V3 and/or (b) &ge; 6 mm in V4 + V5 + V6 and/or (c) &ge; 2 coordinator. Causes of death were subdivided into "non-vascular"
mm in I + aVL; (ie, definitely non-vascular) and "vascular" (ie, defmitely or
Inferior and anterior ST elevation (2%) .-Both of above; possibly vascular). The latter, as specified in the original protocol,
ST depression 8%).&mdash;None of the above, but with ST includes all deaths attributed to cardiac, cerebral, haemorrhagic,
depression as extreme as the ST elevations required above; other vascular, or unknown causes (ie, 9th International
Other abnormality (27% ).-None of the above, but with (a) Classification of Disease categories 390-459, 530-535, or 797-799).
pathological Q-waves (16%) 22 mm in any lead other than Those finally classified as non-vascular were not adversely
aVR or V p or (b) no Q-waves but T-wave inversion (11 %) in associated with treatment (table i), so their addition to the analyses
any lead other than aVR or Vl, or (c) any conduction defect (eg, of vascular mortality to yield analyses of all-cause mortality would
atrioventricular block) or any arrhythmia (eg, atrial fibrillation, not weaken the apparent effects of treatment. For all reports of

supraventricular tachycardia); stroke on the discharge forms, further clinical details (including any
"Normal" ( 2 % J .-Any remaining electrocardiograms. relevant investigations, such as computerised tomographic [CT]

Fig 1-Cumulative vascular mortality in days 0-35.

(a) All patients allocated active streptokinase vs all allocated a placebo infusion; (b) all patients allocated active aspirin vs all allocated placebo tablets; and (c) all
patients allocated both active treatments vs all allocated neither. (Statistical tests to day 35--observed number of vascular deaths in active treatment group minus
expected number, and the standard deviation of this difference: (a) - 118-8SD 20-2, (b) - 105 3 SD 20-2, (c) - 112-1SD 14 3.)

Fig 2-Life-table estimates of 12-month and 24-month survival.

Median follow-up 15 months, maximum 34 months. (Logrank17 tests to Jan1,1988:(a) -133.5 SD26.7,(b) -88.5 SD 26.7,(c) -110 7 SD 19’Oj
 352

scans and necropsy reports) were requested. Confirmation or

refutation of the stroke and its aetiology was based on "blind"
review of these records by a neurologist. All strokes not confirmed
by CT scan or necropsy as cerebral haemorrhages are described as
being of "ischaemic or unknown" aetiology. For, while most
probably were ischaemic, a few may have been due to undiagnosed
haemorrhage. For each stroke, the "likely residual disability" was
classified on the discharge form as non-significant (modified
Rankin9 grade 0-1), moderate (2-3), or severe (4-5); in the present
report, the latter two categories are merged. The few reports of
possible anaphylactic shock were also reviewed blind of treatment
allocation.

Statistical Methods
The protocol specified three main analyses. These involved
assessment of the effects of intravenous streptokinase on vascular
mortality during (a) the first 5 weeks and (b) the entire period up to
Jan 1,1988 (ie, a median follow-up of 15 months), and of the effects
of 1 month of oral aspirin on vascular mortality during the first 5
weeks (ie, days 0-35, the period of antiplatelet cover). It also
specified one "most important subsidiary analysis", involving
separate assessment of the effects of streptokinase on 5-week
vascular mortality among patients randomised 0-4,5-12, and 13-24
hours after the onset of symptoms, and six other subsidiary analyses
(see Results).
Comparisons of survival to Jan 1, 1988, involve time-to-death
analyses using logrank17 methods (with exact variance calculations),
but for events in hospital and deaths during the first 5 weeks
comparisons involve simple analyses5 of total numbers affected.
Most results are presented in terms of changes in the odds of death:
for example, a change from 10% dead (odds 10/90) to 8% dead
(odds 8/92) involves an odds ratio of (8/92) - (10/90), or 0-78, and is
therefore described as a 22% reduction in the odds of death5,18
(rather than as a 20 % reduction in the risk of death). Odds ratios and
reductions in odds are generally given with either 95% confidence
intervals or &plusmn; 1 standard deviation (SD). Two-sided p-values (2p)
are cited, even in subgroup analyses where one-sided p-values (1 p)
Fig 3-Reductions in odds of vascular death in days 0-35, subdivided
might be more appropriate.13.17 by time from pain onset.
During recruitment, interim results on events in hospital were
reviewed about twice a year by an independent data monitoring The sizes of the squares in figs 3, 5, and 6 indicate how much information is
committee. In the light of these interim analyses, and of any other available,s and 95 % confidence intervals are denoted by lines or by diamonds.
evidence or advice they wished to seek, the data monitoring
committee were to advise the chairman of the steering committee if,
in their view, there was at any time proof beyond reasonable doubt
continued their trial tablets throughout the hospital stay,
that for all, or for some, types of patient either treatment was clearly
indicated or clearly contraindicated. Otherwise, the steering
when nine-tenths of 5-week mortality occurred (so, among
committee, collaborators, and administrative staff (except those those dying before week 5, compliance must have been
who produced the confidential analyses) would remain ignorant of 90-95%).
the interim results. In January, 1987, the data monitoring
committee reported that, among patients entered 0-4 hours after Comparison of the 8592 Patients Allocated Intravenous
pain onset, there was clear evidence that mortality in hospital had Streptokinase with the 8595 Allocated Placebo Infusion
been reduced by streptokinase.19 Randomisation continued for
those patients where, in the view of the physician actually Effect on vascular mortality in first 5 weeks (figs 1 a, 3a) and
responsible for the patient, substantial uncertainty remained about later (fig 2a).-During the first 5 weeks there were 791
the appropriateness of treatment: this was still often the case not (9-2%) vascular deaths in the streptokinase-allocated group
only for patients more than 4 hours after pain onset (for whom compared with 1029 (120%) in the placebo group. This
interim results were not made available during the trial), but also for 25% SD 4 reduction in the odds of death in the
some presenting earlier.
streptokinase group is highly significant (2p < 0-00001) with
95 % confidence interval ranging from 18 % to 32%. Several
Results
years of further follow-up will be needed to see how long
17 187 patients were randomised between March 5,1985, these early gains persist. Thus far, with median follow-up 15
and Dec 31, 1987. This large size (and the use of months, there has been a slight and non-significant further
"minimisation") ensured good balance between the divergence after day 35 (fig 2a). Hence, the overall
treatment groups for the main prerandomisation prognostic difference, including both early and late vascular mortality,
features that were measured (see subgroup analyses below), remains highly significant (2p< 000001). Non-vascular
and should do likewise for those that were not. Among those deaths were evenly distributed (32 vs 32: table 1), so all-cause
discharged alive, the infusion was completed in 98% of mortality was also highly significantly reduced
patients allocated placebo and in 92% allocated (2p< 00001), both in the first 5 weeks and in the entire
streptokinase. Since the fibrinolytic effect of an infusion that study period to Jan 1, 1988.
has been interrupted (usually because of some side-effect) Influence of delay from pain onset to randomisation on size of
may be substantial, effective compliance with streptokinase effect of streptokinase on 5-week vascular mortality (fig
allocation is likely to be well over 92%. For aspirin, 94% of 3a).-Among patients randomised 0-4, 5-12, and 13-24
those discharged alive in both active and placebo groups hours after the onset of pain, the reductions in the odds of
 353

TABLE I-EFFECTS OF ALLOCATED TREATMENT ON CLINICAL EVENTS IN HOSPITAL AND ON NON-VASCULAR MORTALITY

excluded (2
*Transient ischaemic attacks lasting 24 h or less were not recorded routinely, and the 6 strokes reclassified on review as TIA are SK only, 2 aspirin only, 1
both, 1 neither). In addition to the 7 haemorrhages "confirmed" (by CT scan or necropsy), 3 "possibles" (2 SK only, 1 SK & aspirin) were reported. All 10 occurred on
days 0-1. 9 died in hospital, and the 1 survivor (SK & aspirin, with CT confirmation) was severely disabled.

vascular death were 35% SD 6 (2p < 0-00001), 16% SD 7 and 3-6% SD 0-3 among those not). Bruising or "minor"
(2p=0-02), and 21% SD 12 (2p=0-08). The reduction is bleeding (for example, oozing from puncture sites,
greatest in those randomised within 4 hours, but it is still microscopic haematuria, blood-streaked vomit or sputum)
significant (17% SD 6, 2p=0-004) in those randomised was reported more commonly among streptokinase-
within 5-24 hours. (If p-values are to be used at all to help allocated patients (3-5% vs1.0%), and there was also a small
decide whether, in a clearly positive trial, particular excess (0-3% SD 01) of bleeds requiring transfusion

subgroups are also positive then 1-tailed p-values may be ("major" bleeds: 05% vs 0-2%; 2p < 0-001). This excess
preferred: these are, however, simply half the cited "2p" appeared similar whether or not streptokinase was used with
values, eg, lp = 0-002 instead of 2p = 0-004.) Within the aspirin, and even the excess of minor bleeds caused by
early period, there was no evidence that benefit was streptokinase appeared to be only slightly greater in the
substantially greater among patients randomised within 1 presence of aspirin (2-8% SD 0-3) than in its absence (2-3%
hour than among those randomised after 2-4 hours. SD 0-3). The excess of any bleeds caused by streptokinase
Detailed subdivision of the effects of delay are provided in did, however, depend on whether intravenous,
fig 3a, but the confidence intervals are too wide for the subcutaneous, or no heparin use was planned (absolute
pattern to be clear without also using evidence from other excesses 5-3%, 2-6%, or 1-5%: table 11), as did the excess of
trials (see Discussion). major bleeds (0-7%, 0-4%, or 0-0%: but, the lack of any
Effect of streptokinase on other clinical events in hospital excess of major bleeds in the absence of heparin involved

(’tables I and 11) .-Hypotension and bradycardia were only 8/2903 streptokinase vs 8/2907 control patients). There
reported far more commonly in the streptokinase group was no unusual excess of bleeds or of hypotension among the
than in its control (100% vs 2-0%), generally during the 401 who were aged 80 or over, or among the 178 with a
60-minute infusion or very shortly after. The excess was not systolic blood pressure of 200 mm Hg or more.
related to the initial blood pressure-indeed, even among There was a significant excess (0-1% SD 0-03) of
the 631 with a systolic blood pressure below 100 mm Hg it confirmed cerebral haemorrhage with streptokinase (7 vs 0;
was no greater than average. Reports of allergic reactions 2p < 0-02), all occurring on the day of randomisation or the
(44% vs 0-9%) were in most cases confined to shivering, following day, and all followed by death (6 cases) or severe
pyrexia, or rashes, also generally during or just after the disability (1 case). During this early period there was also a
streptokinase infusion; on blind review, however, no reports slight but non-significant excess of "other" strokes (20 vs
of anaphylactic shock were confirmed. 22% of patients 13), some of which may also have involved cerebral
received prophylactic steroids but, as predicted on haemorrhage. But, although streptokinase was associated
theoretical grounds,2O this strategy did not seem to alter the with an excess of early strokes, there were fewer strokes in
reported rate of allergic reactions (33% SD 0-5excess with the streptokinase group after this early period (34 vs 54,
streptokinase among patients given prophylactic steroids 2p < 005). Consequently, there was no increase in the
 354

TABLE 11-REPORTED SIDE-EFFECTS ASSOCIATED WITH

STREPTOKINASE AND WITH PLACEBO INFUSION

_ _

*The bleeding caused by SK depended on whether the planned treatment

involved iv heparin, sc heparin only, or no heparin (absolute excesses 53%
SD 0-27, 26% SD 0-12, or 15% SD 0-14 respectively; 2p<00001 for
trend).

overall risk of stroke (0-7% streptokinase vs 0-8% control;

NS) or in the risk of disabling/fatal stroke (05% vs 0-6%; Days from randomisation
NS). Fig 4--Cumulative vascular mortality in days 0-35.
Cardiac arrest was significantly less common among
Patients allocated (i) active streptokinase only, (ii) active aspirin only, (iii)
patients allocated streptokinase, but this represented a both active treatments, and (iv) neither.
reduction only in arrest with death in hospital. Hence, it
does not represent any additional benefit over and above the
reduction in total vascular mortality. reinfarction (2p < 000001), in cardiac arrest (2p < 0-01), and
Reinfarctions in hospital were reported slightly more in stroke (2p<0’01), but there is a considerable overlap
commonly among those allocated streptokinase (28% vs between early deaths and these events reported in hospital
2-4%; NS), the difference being in reinfarction with since some patients had both. To avoid double counting in
discharge alive, not in reinfarction with death in hospital. the assessment of any effects of aspirin over and above those
This overall excess of reinfarction was observed only if on mortality, attention may be restricted to patients

streptokinase was given in the absence of aspirin (3-8% discharged alive. When this was done, aspirin was still
among patients allocated streptokinase alone vs 2-9% of associated with significant reductions in reinfarction (1-0%
those allocated neither active drug), whereas among patients vs 2-0%) and in stroke (0-3 vs 0-6%).

allocated the combination of streptokinase and aspirin there

was no such excess (1-8% combination vs 1-9% aspirin Comparison of the 4292 Patients Allocated the Combination of
alone). Streptokinase and Aspirin with the 4300 Allocated Neither
Effect on vascular mortality in first 5 weeks (figs 1 c, 3c, 4)
Comparison of the 8587 Patients Allocated Oral Aspirin with and later (fig 2c).-Mortality among patients allocated the
the 8600 Allocated Placebo Tablets combination of both agents was significantly less than that
Effect on vascular mortality in first 5 weeks (figs 1 b, 3b)and among patients allocated both placebos (42% SD 5
later (fig 2b).-During the first 5 weeks there were 804 reduction by the combination, with 95% confidence limits
(9-4%) vascular deaths in the aspirin-allocated group 34%-50%; 2p < 0-00001). Indeed, the combination was
compared with 1016 (11 8%) in the placebo control group. significantly (2p < 00001) better than either active
This 23% SD 4 reduction in the odds of death in the aspirin treatment on its own. After the first 5 weeks there was no
group is highly significant (2p<0-00001), with 95% indication of any further convergence or divergence (fig 2c).
confidence interval ranging from 15% to 30%. As was the Hence, the overall difference in vascular mortality,
case for streptokinase, several years of further follow-up will including both early and late deaths, remains highly
be needed to see how long these early gains persist. Thus far, significant (2p < 0-0001). Non-vascular deaths were slightly
during a median follow-up of 15 months there has been a fewer among treated than among control patients, so
slight and non-significant convergence after day 35 (fig 2b), all-cause mortality was also highly significantly reduced
but the overall difference, including both early and late (2p < 0-0001), both in the first 5 weeks and in the entire
vascular mortality, remains highly significant (2p < 0-001). study period to Jan 1, 1988.
There were fewer non-vascular deaths among aspirin- The beneficial effects of streptokinase and of aspirin on
allocated patients (25 vs 39; NS: table I), so all-cause mortality appear to be largely independent of each other (fig
mortality was also highly significantly reduced (2p < 0-001), 4). Streptokinase significantly reduced 5-week vascular
both in the first 5 weeks and in the entire study period to Jan mortality both among patients allocated aspirin and among
1,1988. those allocated placebo tablets (reductions in the odds of
Effect of aspirin on other clinical events in hospital (table death by streptokinase: 28% SD 6 and 23% SD 6
I)-Bleeds requiring transfusion ("major" bleeds) were respectively; each 2p< 00001). Similarly, aspirin reduced
reported with a similar frequency in both groups (0-4% mortality irrespective of whether patients were allocated
aspirin vs 0-4% placebo). Apart from a small absolute excess streptokinase or placebo infusions (reductions in the odds of
(0-6% SD 0-2; 2p < 0-01) of "minor" bleeds, there were no death by aspirin: 25% SD 6 and 21 % SD 6, respectively;
significant adverse effects of the low-dose aspirin regimen each 2p < 0-001).
tested. In particular, there was no significant excess of Aspirin produced similar-sized reductions in mortality
cerebral haemorrhage. There were significant reductions in among patients treated early and those treated late after the
355
356

TABLE III-PERCENT REDUCTIONS IN VASCULAR EVENTS OBSERVED

IN TRIALS OF MEDIUM AND LONG-TERM ANTIPLATELET THERAPY
FOR SUSPECTED OR DEFINITE MYOCARDIAL INFARCTION

*With survival to wk 5 in ISIS-2 and to the end of the scheduled treatment

period in the 10 long-term trials.
tIn the ISIS-2 pilot trial12 there were 27/313 (8-6%) vascular deaths during
the first 5 wk among patients allocated 4 wk of aspirin compared with 36/306
(11-8%) among those allocated placebo tablets. Among those discharged
alive, reinfarctions in hospital were 7 vs 10 and strokes were 1 vs 2. The only
other randomised trial in acute MI6 involved 300 mg aspirin once only, with
no further trial treatment after day 0, and yielded only a very slightly
favourable result (total deaths by day 28: 159/1249 vs 172/1281).

Fig 6-Reductions in the odds of early death, subdivided by time

from pain onset: overview of the six randomised trials of 1.5 MU
intravenous streptokinase (and of late treatment in other trials’). 13 increase; NS), while for patients born under all other
The infusion duration was 60 min in 5 trials3,1223 32 and 30 min in White et astrological signs there was a strikingly beneficial effect
aL.33 Square sizes and 95% confidence intervals are as in fig 3. The difference (28% SD 5 reduction; 2p < 0-00001). It is, of course, clear
between the 31 % reduction 0-3 hours after pain onset and the 14% reduction that the best estimate of the real size of the treatment effect in
6-24 hours after pain onset is 17% SD 7, 2p < 0 02. each astrological subgroup is given not by the results in that
subgroup alone but by the overall results in all subgroups
combined. In different biological subgroups, however, the
onset of symptoms (odds reduction at 0-4, 5-12, and sizes of the effects of treatment may well really be somewhat
13-24 hours: 25% SD 7, 21 % SD 7, and 21 % SD 12, different, but still the directions of the effects in the different
respectively), while the effects of streptokinase appeared subgroups may well all be the same, as long as patients with
greatest among those treated earliest. So, when patients clear contraindications to treatment or with negligible risk of
allocated the combination of both streptokinase and aspirin early death from infarction are excluded. If so, then the best
are compared with those allocated neither active drug, the estimate of the direction (but only of the approximate size) of
odds of death were significantly reduced among patients the real effect of treatment in each different biological
randomised 0-4 hours (53% SD 8 reduction; 2p < 0-00001), subgroup may be that suggested by the proportional risk
5-12 hours (32 % SD 9 reduction; 2p < 0-0001), and 13-24 reduction in all subgroups combined. Clearly significant
hours (38% SD 15 reduction; 2p = 0-01) after pain onset. overall results may therefore provide strong indirect
Effects of the combination of both agents on other clinical evidence of benefit in subgroups where the results,
events in hospital (table I).&mdash;As for streptokinase alone, the considered in isolation, are not conventionally significant (or
combination of streptokinase and aspirin was associated even, perhaps, slightly adverse).
with a 0-3% SD 0-1 excess of "major" bleeds when "Lack of evidence of benefit" just in one particular
compared with patients receiving neither active treatment. subgroup is not good "evidence of lack of benefit". Overall,
A 0-1% excess of cerebral haemorrhage was still observed, for example, GISSI provided strong evidence that
but this was offset by shortfalls of 03 % in other strokes with streptokinase reduces mortality,3 but inevitably the results
death in hospital, of0.1 % in disabling strokes, and of 0 -2 % were not conventionally significant in particular subgroups
in other strokes with discharge alive. Overall, therefore, the (eg, patients with inferior infarcts, those aged over 65 years,
combination of streptokinase and aspirin was associated those with a previous infarct, those presenting more than 6
with 0-5% SD 0-2 fewer strokes (06% vs 1 .1 %; 2p 0-02). =
hours after pain onset, and so on). However, the overall
It was also associated with fewer reinfarctions, cardiac GISSI result provided strong indirect evidence that
ruptures, and cardiac arrests in hospital, but most of these fibrinolytic therapy has at least some effect in these
differences were among patients who died in hospital and so subgroups, and ISIS-2 confirms this (figs 5a and 6). As
have already been accounted for in the mortality analyses. would be expected if subgroup analyses were not
particularly informative, some are concordant in different
studies (eg, no apparent benefit from streptokinase in
Subgroup Analyses of the Effects of Streptokinase and of GISSI or ISIS-2 among patients with ST depression) and
Aspirin on 5-week Vascular Mortality (fig 5):Results, with some are discordant (eg, in ISIS-2 streptokinase reduced
Discussion
mortality significantly [2p < 0-001 in patients with a history
Even in a trial as large as IS I S-2, reliable identification of of MI, while in GISSI it did not). Of the aspirin subgroup
subgroups of patients among whom treatment is particularly analyses, two did indicate a degree of heterogeneity just
advantageous (or among whom it is ineffective) is unlikely to about as great as that for astrology, but neither is supported
be possible. When in a trial with a clearly positive overall by other evidence. First, the lack of apparent effect among
result many subgroup analyses are considered, false diabetics is not conventionally significant, and a non-
negative results in some particular subgroups must be significant interaction in another trial has led to the opposite
expected. For example, subdivision of the patients in ISIS-2 suggestion.21 Second, the lack of apparent effect of aspirin in
with respect to their astrological birth signs appears to people with a previous infarct is implausible, since in the
indicate that for patients born under Gemini or Libra there present study aspirin significantly reduces reinfarction
was a slightly adverse effect of aspirin on mortality (9% SD among such patients (38/1454 vs 66/1483, 2p < 001), and
 357

long-term aspirin use after MI reduces both reinfarction doubt that fibrinolytic therapy reduces mortality-indeed,
and death (table III). All these subgroup analyses should, even the lower 95% confidence limits for the estimated

perhaps, be taken less as evidence about who benefits than as mortality reductions are well away from zero. In contrast,
evidence that such analyses are potentially misleading. although the observed mortality reduction is statistically
(Statistically valid methods for estimating the true effect of significant (2p = 0 02) among patients treated more than 6
treatment in an "outlying" subgroup do exist,22 but do not hours after pain onset, it is still consistent with there being
include analysis just of that one subgroup on its own; only a small benefit. In view of the pattern of gradual
instead, where there is little evidence of any real diminution of benefit with delay, however, and in view of
heterogeneity, as in the present study, they involve giving the benefit from streptokinase after 6 hours in the previous
more weight to the overall result than to the data in the trials (fig 6), it is likely that fibrinolytic therapy does produce
subgroup of interest.) some worthwhile benefit even when used 7-24 hours after

Discussion pain onset. But, no analyses of ISIS-2 alone are likely to

establish reliably which such patients can expect the greatest
The size of this trial ensures that the reductions in proportional benefit. (For example, the reduction in 5-week
vascular mortality (or in all-cause mortality) that have been vascular mortality by streptokinase in patients treated 7-24
demonstrated for intravenous streptokinase and for oral hours after pain onset was 22% SD10 [2p<0’05] among
aspirin are both definite. The benefits appeared to be largely those with ST elevation recorded on their pre-
independent of each other-that is, streptokinase reduced randomisation ECG, which is similar to the 18% SD 7
mortality by a roughly similar additional amount [2p 002] mortality reduction among all patients entered
=

irrespective of whether aspirin was used or not, and vice late.)

versa. Consequently, the combination of both active drugs When deciding whether to use fibrinolytic therapy in
produced a reduction in the odds of death at 5 weeks that was patients with suspected acute MI, particularly those
far larger (42% SD 5 for both agents versus neither) than presenting late, it is necessary to consider not only the
that produced by either drug alone, with such a narrow 95 % possible benefits but also the possible risks. At least for
confidence interval for this reduction (34%-50%) that the streptokinase, however, the risks are small: side-effects such
possibility of there being little real benefit is absolutely as hypotension, allergic reactions, or minor bleeding may be
excluded. This does not, of course, guarantee that the risk worrying but do not usually cause patients managed in
reductions will be exactly the same size in each specific hospital any serious problem, and the recent trials (ISIS-2,
category of future patients, but it does provide strong ISIS-2 pilot, GISSI, ISAM) have demonstrated that
indirect evidence that these treatments will generally reduce serious side-effects of the rapid high-dose iv streptokinase
coronary mortality substantially. Generalising too far may regimen are rare--only about 3 bleeds requiring transfusion
lead to patients being treated inappropriately, for if and about 1 or 2 cerebral haemorrhages per 1000 patients,
streptokinase is used where it can confer no benefit then it with no significant increase in the total number of strokes
may cause 1 or 2 unnecessary deaths per thousand. But so (ISIS-2 61 SK vs 67 placebo, GISSI 49 vs 40, ISAM 5 vs 0,
too may failing to generalise far enough, for if aspirin and ISIS-2 pilot 2/413 vs 5/206 [ -1/206 vs 5], other trials 1 vs 0:
streptokinase are not used where they could confer benefit total 117 vs 112). On the other hand, the risk of coronary
then this may lead to about 50 avoidable deaths per death in patients with suspected acute MI may be high, even
thousand. among those without pronounced ST elevation on their
ECG or among those admitted late after symptom onset. A
Fibrinolytic Therapy reduction in mortality of "only" 14%, as suggested by the
The principal findings were that streptokinase improves trial results for those presenting after 6 hours, would
5-week survival, that this benefit persists for some years, and typically avoid about 15-20 early deaths among every 1000
that (although the effects seem greatest among patients patients, which compares favourably with the small number
treated most promptly) fibrinolytic therapy appears to of serious complications that might be expected.
produce some benefit even among those treated up to 24
hours from pain onset.
The hypothesis that fibrinolytic therapy might be Antiplatelet Therapy
beneficial more than 6 hours after pain onset was generated Reinfarction, stroke, and mortality are reduced by
by an overview1 of previous trials (fig 6). Among such antiplatelet therapy used long-term in the months or years
patients in ISIS-2 streptokinase produced a reduction in after a myocardial infarction or in patients with unstable
vascular mortality that was significant at week 5 (2p 0 02)
=
angina (table III). But before ISIS-2 and its pilot, daily
and that has persisted well after week 5, with a median of 15 aspirin had not been tested in the acute phase of MI. The
months of follow-up thus far (2p 0-004 for logrank17 test of
=
present trial indicates that one month of low-dose aspirin,
effect of streptokinase on overall mortality among 6477 started immediately in 1000 patients with suspected acute
patients entered 7-24 hours after pain). To obtain the most MI, would typically avoid about 25 deaths and 10-15
reliable test of this hypothesis, the results of ISIS-2 need to non-fatal reinfarctions or strokes, and that much of this
be considered along with those from other recent benefit persists well beyond this short treatment period.
unconfounded randomised trials of rapid high-dose (1’5 Continuation of antiplatelet therapy for 2-3 years in 1000
MU within 1 hour) iv streptokinase regimens. Only two post-MI patients would, moreover, typically prevent about
such trials3,12 have included patients presenting more than 6 afurther 20 deaths and afurther 30 non-fatal events.5
hours from pain onset, and these, along with ISIS-2 and The optimum dose, and frequency of dosing, of aspirin
ISAM,23 are the four largest trials of a rapid high-dose remains uncertain. If the chief mechanism is inhibition of
regimen. There were no significant differences between cyclo-oxygenase-dependent platelet aggregation, then any
their findings, and an overview of these results (fig 6) daily dose from about 40 mg upwards may suffice-as
provides significant confirmation of benefit among patients indeed, may less frequent doses. The definite reductions in
treated after 6 hours. For patients treated within 0-3 hours vascular events observed with 160 mg/day in ISIS-2
and 3-6 hours of the onset of pain, the data now available resemble those observed with higher doses (300-1500 mg
from ISIS-2, GISSI, and the other trials should leave little daily) in the long-term trials (table III). Higher doses of
 358

aspirin are several times more gastrotoxic9 than lower doses, 5-10 times asexpensive as streptokinase, and are now being
but do not seem to be any effective.5 At present,
more tested, though only within 6 hours of pain onset. For
therefore, when long-term antiplatelet therapy is to be used neither, however, have the trials been large enough for
after MI, unstable angina, transient cerebral ischaemia, or indirect comparison even of an overview of them with the
stroke, a dose of about 160 mg/day may be preferred. streptokinase results in hours 0-6 only (fig 6) to yield
statistically reliable information (and, of course, selected
The Combination of Streptokinase and Aspirin comparisons just of single trial results with each other would
be even less statistically informative). Hence, it is not known
This combination appears to have serious side-effects that
which, if any, of the fibrinolytic agents is most effective at
are no more frequent than those of streptokinase alone. Yet
the combination reduces the risk of disabling or fatal stroke averting cardiac death and which, if any, carries the greatest
risk of cerebral haemorrhage or other serious side-effects.3o
and of reinfarction, as well as reducing mortality much more
GISSI-2 and ISIS-3 will therefore involve direct large-
substantially than either agent does alone. For patients in scale randomised comparison between different fibrinolytic
ISIS-2 admitted early after the onset of pain, the
combination seemed to reduce 5-week mortality by about agents.
one-half, and even when used after a delay of several hours it
For antiplatelet therapy, the drug costs may be
seemed to reduce mortality by about one-third. For negligible-for example, perhaps just a few tens of pounds
per life saved by aspirin-but, now that such treatments
example, among those entered 13-24 hours after the onset of have been proved to avert death, some really large
pain the mortality reduction was 38% SD 15 (45/608 randomised comparisons are needed between different
vs71/614), with apparently equal contributions from aspirin antithrombotic regimens (for example, one antiplatelet
and from streptokinase (fig 3). Overall, therefore, among
patients treated up to 24 hours after pain onset, the regimen versus another or aspirin alone versus aspirin plus
combination reduced the odds of vascular death by 42 % SD anticoagulation).
5, with a lower confidence limit of 34%. The apparent effect
of streptokinase is a little better in IS I S-2 than in some other Clinical Implications
trials (fig 6), but still the combination of streptokinase and The initial diagnosis of MI, and decisions about acute
aspirin is likely to reduce 5-week mortality by more than treatment, depend largely on the physician’s judgment of
one-third for a wide range of patients, avoiding about 50 the clinical history and of the admission ECG. Such
deaths (or more, if antiplatelet therapy is continued for a
decisions cannot, of course, take into account information
longer period) among every 1000 treated. available only later from cardiac enzyme estimation or
Platelet activity is increased in acute Ml, and is increased
coronary angiography. ISIS-2 is directly relevant to acute
still further by fibrinolytic therapy. But this increase can be
treatment, for the principal entry criterion was that the
avoided by the addition of aspirin.24 In ISIS-2, the increase
in reinfarction-and hence, presumably, in reocclusion-
responsible physician suspected acute MI on the basis of the
clinical presentation alone (although in practice 98% of
produced by streptokinase alone was avoided if patients had some ECG abnormality). The absolute
streptokinase was given in combination with aspirin. If mortality reductions appear to be greatest for patients at
fibrinolytic and antiplatelet therapies are combined, then it greatest risk of death (for example, women, older patients,
is not yet clear whether additional interventions to avoid
hypotensive patients, patients with a previous MI or with an
reocclusion (eg, anticoagulants or angioplasty) will confer anterior infarct). Patients with systolic blood pressures
additional benefit. But it is clear that streptokinase and below 100 mm Hg are at particular risk, and among them
aspirin will reduce both reinfarction and death, even if other streptokinase appeared to produce a particularly large
interventions are not used. This finding substantially absolute benefit (while the incidence of hypotension due to
simplifies the routine use of fibrinolytic therapy. streptokinase was no larger in them than in normotensives).
Research
Similarly, among the 3411 patients aged over 70 (fig 5a)
Implications streptokinase was associated with a significant reduction in
These results indicate what large simple randomised mortality, and in absolute terms the 5-week survival benefit
trials13 can offer, not only in acute MI but also in many other was somewhat greater among them than among younger
conditions. They also show what overviews of randomised patients (with no evidence that the risks of treatment, in
trials can offer, for ISIS-2 was undertaken because of the particular cerebral haemorrhage or other bleeding, were
benefits suggested by two overviews (oneS of which also related to age). But even though the absolute benefits
engendered two studies of aspirin for primary prevention,2s produced by streptokinase and by aspirin may be largest
and the other1 of which also engendered the GIS S 13 study of among high-risk patients, they may still also be worthwhile
streptokinase). Moreover, the fibrinolytic overview in many categories of patients at below-average risk of
generated the hypothesis that treatment could be effective cardiac death. Furthermore, the benefits of streptokinase
even if given to patients presenting late after pain onset. This may also outweigh the risks even among many patients who
influenced the design of ISIS-2, and the results for such have some relative contraindication (eg, recent use of
patients could double the number of patients who benefit streptokinase or an old history of stroke or of gastrointestinal
from treatment. haemorrhage or ulcer).
ISIS-2 has shown that fibrinolytic therapy and For streptokinase, ISIS-2 does not support the
antiplatelet therapy both reduce mortality in many suggestion3 that the benefit among those treated within the
categories of patient. For any patients where uncertainty first hour is much greater than among those treated slightly
remains about the benefits of fibrinolytic therapy, this later (fig 3a); but, a small decrease in the median time of
should encourage further randomisation between treatment (eg, from 5 hours to 4 hours3l) does produce a
fibrinolytic and control treatment, as in EMERA26 and small improvement in the mortality reduction (eg, from
ISIS-3. Streptokinase is the most extensively studied 20% to 23%; fig 6). Worthwhile improvements might
fibrinolytic agent, and the present results indicate that the therefore be achieved by simple measures, such as
cost per life saved by it is only a few thousand pounds. encouragement of prompt hospital admission and starting
Newer fibrinolytic agents such as tPA27,28 and APSAC29 are treatment in the emergency room before transfer to
 359

Burlington: Carling L, Beare L ; Cambridge Memorial: Cooper j., Wilkinson D , Pubhc

coronary care. might, however, be best to delay
It
General, Chatharn Hamarine C , Pfaff S.; St Joseph’s, Chatham: Willston M . ; Queensway
fibrinolytic treatment until hospital admission unless the General, Etobicoke: Semtt B., Daniels L., Wassmer S Chedoke, Hamilton Finkelstem L.,
patient’s condition can be monitored during and after Cooper M.; Henderson General, HamiltonSealey B., Olde J. ; .McMaster, Hamilton’Fallen
E , Mmard J. ; St Joseph’s, Hanrrlton: Sullivan M., Tanser P Kennedy D., Seager S St
infusion, for streptokinase produces some side-effects (eg, Joseph’s, Lortdon: Goddard M., Janssen J.; University Hospital, London: Jablonsky G.,
sudden profound hypotension) that, although generally easy Bradley R., Victoria, London: Fmme K., Adkms B.; Oakville- Trafalgar, Oakvillle:
Zawadowski A., Graham J., Oshawa General: Bhargava R., Taylor S, Harper G.;
to manage in hospital, do require prompt medical attention.
Peterborough Civic: Hughes W., Jamieson S., St Joseph’s, Peterborough Nawaz S.,
Aspirin does not require particularly careful monitoring, Lanovic A , Reesor J., Young J.; Verona., Renfrew: Appleyard C, Riddell R.; Plurrvner
and it might well be appropriate to start it as soon as possible Memorial, Sault Ste Marie: Stanton E., Rivet L; Norfolk General, Srrneoe Chiu S.,
McCrea C. ; Sudbury Memorial: Juma Z., Ballance S., Turcotte J., St Mary’s, Timmins:
(in the home, ambulance, or emergency room) provided Cunningham H , Wilson L ; Woodstock General. WIlkey W., Parclak S.
there are no clear contraindications. Short-term antiplatelet Denmark ( 339 patients.-Rigshospitalet, Copenhagen Pedersen F.*, Sandoe E. *, Kyst
therapy is likely to be applicable to almost all patients with Madsen J.* ; Braedstrup: Amtoft A ; KAS Gentofte, Copenhagen Saunan1aeki K,
Svendsen J ; Skt Lukas Stlftelsen, Copenhagen: Tidstroem B.; Fakse: Hansen J.; Farso:
suspected MI, since the side-effects of low-dose aspirin Lmdskov J ; Frederica. Homum 1, Give: Dedmg A, Kalundborg: Lmdbjerg 1.;
seem negligible and the drug costs are small. Aspirin could Kjellerup: Scheibel M., Maribo-- Hilden M. ; Ringsted Marcussen H ; Silkeborg Central:
be used widely not only in developed countries but also in Jensen T , Soenderborg: OllendorfP.
countries with limited medical resources. If one month of Fmland (433 pat2ents).-Maria, Helsinki: Kala R.*; University Central, Helsinki:
Heikkila J.*, Irnatra District Itkonen A., Havakka E,;Fokilaakso District: Nyyssonen S.,
low-dose aspirin were to be given to just one million new Lahtinen K.; Kemijaru D2strzct- Sako S., Tauriainen P., Keskl-Pohjanmaa Central:
patients a year-which is only a fraction of the worldwide Halkosaan M Salonen P ; Kotka Central: Salmmen K., Si)vennoinen M , Lappi Central:
Eloranta M , Autio R.; Lohja District: Kolvunen L., Kaarlola A. ; Lozinats-Hdine District:
total with acute MI-then a few tens of thousands of deaths, Koskelainen J., Ryhta 1.; Malrrtska District: Wmgren P-E , Sodo C.; Mantta Dzstnct:
reinfarctions, and strokes could be avoided or substantially Valhttu H , Aikas E.;,Wikkeh Central: Tarssanen L., Virkki L ; Porvoo District Harkonen
delayed (and these benefits could be doubled if low-dose M., Rask K ; Rarsio District. Karmakoski J., Lattu M., Selkamert District - Tuunanen V,
Jaatmen R, Vaasa Central: Lmdroos M, Kivisalo M. Ahtan District: Maattanen J
aspirin were continued for at least a few more years5). Mursula R
Although further research may eventually identify some France (827 patients).-Hospital Cardzologzgzte, Lyon Boissel J-P.*, Leizorovlcz A.*,
fibrinolytic or antithrombotic regimens more effective than Glllet J. ; Hospital Sud, Amlens’ Qmret J., HermIda J-S.; Centre Hospitaller, Annec-v:
those tested in ISIS-2, streptokinase and aspirin are Dupont J., H,5pital Satnt- rlndre, Bordeaux: Blanchot P Lemetayer P.; Centre Hospualzer
General, Dunkerque. Demarq J. ; H6pztaZ Andre Mzgnot, Le (Ghesnay., Normand ].,
practicable,and are of demonstrated value and safety. If Schwob J. ; Hi5tel Dleu, Le Creztsot: Ravisy J. ; Hospital Emile Roux, Le Puy. Viallet M.;
both are used widely then they should avoid several tens of Centre Hospitaller Irztercorrtrnurtal, Longjumeau: Roche P Centre Hospitalier, Macon:
Cavallaro J , Auberger R., Centre Medical d’Evecquemont, Vfeulan: Abastado M Centre
thousands of deaths each year. Hospitalier, Roanne: Gh&egrave;re R., Centre HosPltalier, Romans: Michelon G.; Centre
Hospitaller, Strasbourg: Finckler L Muller J ; Centre Hospitaller, Tourcoirig Beuscart C ,
The most important acknowledgment is to the 17 187 patients who agreed Leroy 0., Centre Hospitaller, Valence. Grand A., Centre Medzcal, Vaux sur Seme: Canny
toparticipate, and to the thousands of doctors and nurses in 417 participating M, Centre Hospitalier Luczen Hussel, Vienne. Veyre B.
hospitals who collaborated with the national coordinators in each country. Germany ( 1884 patients).-Klznzkum Steghtz, Berlin: Schroder R. *, Schafer H Amtz
The coronary care unit nurses in Oxford, Berlin, Bruxelles, Gent, Valencia, R., Schroder R ; Herz-Krelslauf-Klimk, Bad Bevensen. Noetges A, ,WolfR ; Med Klinik,
and Bellinzona, the computer department in Lyon, and the CTSU staff in Bad Homburg Raisig S Bodem G. ; St Gertrauden-KH, Berlin: Ramdohr B., Grupp H.;
Oxford provided the 24-hour randomisation service. The steering and data KH am Urban, Berlin Dissmann W., Topp H. ; Augusta-Krankenanstalten, Bochurn.
Kniemann H , Bohle E Knappschafts-KH, Bottrop Harbarth P ,Schtlling T St Joseph,
monitoring committees supervised the study and the IS IS trial office staff and Bremahm’en. Martin K., Stadtische Klmzken, Darmstadt. Frederkmg H Knappschafts-
the national coordinators collected and checked the data. The coordinator was
KH, Dortmund. Wagner P., von der Ecken B.; St Johannes, Duzsbi4rg: Beckmann M., St
supported by the British Heart Foundation. Sterling Drugs donated aspirin Vincenz-KH, Essen. Toker Y, , Franztskus-KH, Essen. Dorwald R., Burgerhospital,
and its placebo, but otherwise the entire study was financed by Behringwerke, Frankfurt: Gottstem U., Sedhneyer L, St Josef, Gelsenk:zrchen. Callsen H., Klzmk am
a subsidiary of Hoechst, the manufacturers of ’Streptase’. The study was, Elchert, Goppingen GIesler H ; Allgemelnes KH, Hagen: Rox J.; St S,xtus, Haltern:
however, designed, conducted, analysed, and interpreted independently of Beythien R , Padel T ; Allgernenzes KH Bagedorf, Hamburg Seevers H ; Evangehsches
the companies. The following centres and investigators collaborated KH, Hamm Mosseler U. ; KH St Marienberg, Helrnstedt. Strauch A, Walther C.; Med
Klinik II, Koln-,Herhezm Gnebenow R; Med Klfnzk I Krankenanstaltert, KonstanN:
(*steering committee member, and tdata monitoring committee): Aschoff G ; Stadt KHKorbach. Engelsmg B, Nasher D; KHLandsberg. Perl R., Gatz
Australia (1995 patients).-Royal Melbourne: Hunt D.*, Varigos J.,* Carlisle C,
Falconer P.;Albury Base: Landy T., Smedley A.; Ballarat Base: O’Sullivan D., Lems F.;
S Luzsen-KH,Lmdenfels’ Huep W.; St L’tncenz-u Elisabeth,Mainz Schwartz B , Weber
M ;.Behrzng, Marburg. Bernhardt W, Carls C, Muller J, Weidmann E; Krets-KH
Bendigo Base: Gault J , Tumer ; Blacktown: Sadick N , Patterson D.; Box Hill: Ryan Mechernich. Neuhaus J Mohr P. ; Klznzkzipn Minden: Ammer W.; Krezs-KH Mosbach.
W., Chong J.; Redcliffe, Brisbane: Heazlewood R., Jansen J. ; Royal Brisbane: Bett N., Heller A., Evangelisches KH, Mulhezrn,’’.Ruhr Kotter V. ; Med Klinik II u. III,
Perry B.; Cazrns Base: Rolan P., Balodis M.; Woden Valley, Canberra: Coles D., McGill Grosshadern, ,Munchen Engelhardt D Riess H , KH des 3 Ordens, Munchen Wetdmger
D. ; Pnnce Charles, Chermside: Neilson G , Fung P.; Concord Repatrzation: Noble N., P, Lindner D., Stadtlsches KH, Nettetal. Appenrodt H., KH Neu-Ulrrt. Spiker A.; St
Sullivan M.; Fremantle: Lane G., Sparks C.; Gippsland Base: Ziffer R., Edwards A.; Gold
Coast General: Aroney G., Hicks P.; Goulburn Valley Base. Buncle A., Pogue W.;
Josef, Oberhausen: Schulze-Wethmar H., Kremer G.; Sr JosefSterkrade, Oberhausen:
Muting H., Nienhaus T;Prosper, Recklmf:hausen Becker E, Sodomann C.;
Repatnatzon General, Greenslopes: Colquhoun D., Lmdsay C.; Gnffith Base: Negus R., Knappschafts-KH, Recklinghausen’ Meyer W. ; ,Med Klmlk I, Remschend Lollgen H. ; Med
Henderson J. ; Heidelberg Repatriation General’ Hamer A, Dowhng R.; Blue Mountains
Klunk 11, Rarrtschetd: Zack W, Lang A; ; Dtakonze-KH, Rotenburg: Bott)er H.; Khrnken
District, Karoomba. England J., Harwyn J ; St George’s Kew. O’Sullivan E., Firman M., der Stadt, Saarbrucken. Zmrner K.; Stadt-KH, Salzgitter-Bad: Meister G ; Krels-KH,
Launceston General Smgh B., Parker J , Lzsrnore: Tsicalas M., Pulsford N. ; Mackay Base
Gmessler A, Kochevatkin B ; Maryborough Base. Ekin F., Jensen M., St Vincent,s,
Schleswtg. Schoormans W; Med Klznzk B Kantonsspual, St Gallen: Huber B.; KH vom
Roten Kreuz, Srzirtgart Kapune W., Rupp E. ; Knappschaftsklinik Sulzbach Konrad A.;
Melbourne Hale G, Elmore T.; Western General, Melbourne: Newman R, Kilmier C ;
St Johannes-KH, Tr&ugrave;lsd&ugrave;rjSle"lar. Strannann P.; Krets-KH Uelzen Borm P,
.Lltldura Base Dowty B , Perez B., Royal Newcastle. Basnan B , Carpenter K., Orange
Stamatelatos M ; ; Klnukurn der Landeshauptstadt (x’ze.sbaden: PIper C, von EgIdy H ;
Base. Huxtable B., Sox E. ; Rockhampton Base. Ewart A , Lee R; Ryde: Young A , Kelly
Remhard-Nieter-KH, !17z1helmshaven: Frerichs H.
Manning River, Taree: Newman R., Kmght M; Townsville General. Gunawardane
K, Cosgrove M., It:’agga Wagga Base: Childs C., Mantell D-, Wangaratta Fiddes P., Ireland 831 patients) Beaummlt Hospital, Dubh7i: Horgan *, O’Callaghan D.,
Tonmi F ; Warrnambool: Morphett B., Kenadh H. ; Newcastle Mater Misericordiae, Ohman M., Boyle E , Crowe B. ; Castlebar General : Lucey C , Roche M. ; Cavart General,
Waratah Curteis P., Casey C, Westmead. Young A, Dunn L. ; Princess Alexandra, Co Cavan: Farrelly A., Fay A., County Hospital, Barrtry, Co Cork McCoy D., Kmgston
Woolloongabba Woodhouse S., Gow J J.; Mallo General, Co Cork. O’Sullivan P., O’Bnen E ; Regional, Wilton, Co Cork: Erwin
Austria 117 patients-Umv Klzntk, Innsbruck Dlenstl F *, Lechleitner P.; KH
J., Fennell W., Palmer C ; Lettarkenny General, Co Donegal. Bannon L, Doyle H.; St
Bludenz: Schobel B, KH Bregenz. Erhart K., KH Eisenstadt. Silberbauer K., KH
Michael’s, Dun Laoghatra, Dublin. Bamiville H , Mason H ; Naas General, Co Kildare -

Crowe M., Power J , Boland T ; St Luke’s, Kilkenny - Mahon J., Leahy J , Barringtons,
Kufsten:Stuehlinger W ;KH Muerzzuschlag ForenbacherH KH Schwaz Ciresa M.; Lzrrtertck: Holmes R , Geary A, , Limerick RegIOnal, Co Ltrrtertck Peirce T., Hayes J.; Our
KH St Johann (Tirol ; Baumgartl P, KH Zams Pall H., KHZella7ii See Erd W
Lady of Lourdes Drogheda, Co Louth. Muldoon B , O’NeIll R., Costello T. ; Mullingar
Belgzurn (693 patients).-Universataar Ziekenhuis, Gent: De Backer G.*, Buylaert W., General. Quinlan C, Wem A., Hyland M., Tullamore General, Co Offalv: Taafe J., Quinn
Michem N., Clement D., Rubbens L.; Erasme, Bruxelles : Komitzer M.*, Degre S., A., Shgo General, Co Shgo Collns D -,%4ollhan A St Joseph’s, Clonmel, Co Tzpperary
Barthels M., van der Valk C.; St Vmcentzus, Antwerpen. Goethals M.; AZ St Fan, Brugge: O’Regan P., Sr Columba.; Ardkeen, Waterford’ FItzgerald G, Kennedy A
Vincke J.;IMC d’lxelles, Bruxelles: Schoenfeld Ph., Poppe M.; Joseph Bracops, Bruxelles:
Italv GISSI liaisoti, Mario Negri, Milano Tognom G.*, Franzosl M-G.*
Depaepe A., Roslet M.; AZ VUB, Bruxelles.; De Wilde Ph.; Civil de Charleroz: Ptcart N.,
Roosbeck E.; IMC A Gailly, Charleron: Pmon F., Visee A.; St Vincentzus, Deznze: De New Zealand, 281 patients, I -Princess Margaret, Christchurch.- Fitzpatrick M., White
Clerck D.; Stedelzjk Zzekenhuzs, Dendermonde: Zenner R ; St Vmcentius, Gent: Van M; ; Waikato, Hamilton: Freldlander D.; Hutt Hospital. Logan R., Dewar J. ; Palmerston
Durme J. ; Volksklimek, Gent: Jordaens L., St Fozefkliniek, Gentbrugge: Kluyskens Y.; North: Campbell-Macdonald J , Hope N.
Notre Dame de Grace, Gosselies. Sottiaux T., Charlot Chr. ; AZ Hoge Beuken: Bruyneel K.;
Czznl de Fumet: Balthazar E., Dehavay M ; OLV Ziekenhuis, Mechelen: Beys J.; IMC, Norway (1067 patients) -Baerum - Kjekshus J.* ; Sentral-sjukehuset, Forde. Reikvam
Tournai: Dereume G, Vasseur N.; CGTR Le Rayon de Solezl, Vesale: Henuzet C., A.* ; Sentral Sykehuset for Ostfold, Fredenksrad: Kahrs J. *; Haukeland, Bergen - Ronnevik
Charlot N.; St Elisabethklimek, Zottegem: Bladt D. P. ; Farsund Bjenng E; Fylkessjukehuset Floro. Smdre R. ; Harrzmerfest: Johansen J.;
Rmgerike, Honefass: Tenstad 0.; Fylkessjukhuset Knstiansund: Rongved G. ; Kongsvinger:
Canada 675 patients)-Hamilton General Cairns J.*, Turpie A.*, Fraser K., Webb Anker E.; Lillehammer Fylkessykehus: Dahle M.; Stensby, Mznnesund- Koss A.; Vefsn,
J: : Bellville General. Grant F, Vanslyke D, Taylor S., Pedlev S; Joseph Braru, Mosjoen: Haugnes T.; Moss: Moum B. ; Nordfjord: Klykken B; ; Fylkessukehztset Odda