Evidence based stroke

medicine. Evaluating
treatments for acute
ischaemic stroke -what
works and what doesnt?
Professor Peter Sandercock
 85% of strokes are ischaemic, and related to
blockage of an artery by a blood clot, so potential
treatments to improve the circulation might be:
Thrombolytic (clot-dissolving): eg
Streptokinase, TPA. Breaks up clot by
splitting fibrin
Anticoagulant (Clot preventing): prevents
formation of fibrin, prevents spreading of clot
& formation of new clot
Antiplatelet (clot preventing): prevents
platelets sticking together prevents spreading
of clot & formation of new clot.
 Does treatment X do more
good than harm?
Benefits of the
treatment (eg
reduced
disability)?
Risks (eg fatal
?
bleeding)?
What is the balance of RISK
and BENEFIT?

RISK BENEFIT
 Getting reliable evidence on new
treatments (in animals and patients)
Minimise selection bias - RANDOM
ALLOCATION
Minimise observer bias - BLIND
ASSESSMENT of outcome
Minimise random error - LARGE
SAMPLE
Choose appropriate measure of
outcome - use COMMON SENSE
 Effects of treatment X on
death in a small trial
Experimental Control
(n=100) (n=100)
10/100 12/100
(10%) (12%)

Difference = 2% or 20 deaths avoided per 1000 treated,

but not statistically significant. Would need a trial with over
10,000 subjects to confirm or refute a benefit of this size.
OK, how to design appropriate
randomised trial?
Large, simple
What to use as primary measure of outcome?
vessel patency
volume of brain damaged on scan?
disability
handicap?
Quality of life
death?
what about side effects (eg bleeding?)
Highly selected patients or a broad range?
Thrombolysis in Acute MI: accumulating the evidence

Antman & Lau. JAMA 1992; 268: 240-8

 By 1980, what did we know about
thrombolysis for heart attacks?
It unblocks arteries in dogs with experimental
heart attacks:
23 small randomised trials completed, including a
total of about 6000 patients, showing:
It can unblock arteries after heart attacks
bleeding could be a side-effect (sometimes fatal)
trials were too small reliably to answer the question
does thrombolysis save lives in patients with heart
attacks? , but suggested the benefit might be about 20
lives saved per 1000 patients treated
Doctors not convinced: treatment not used
Publication of large-scale MI trials followed
by increased use in UK Trent region
(Ketley and Woods Lancet 1993: 342: 891-4)
 What is a systematic
review?
Defined aim
Defined methods
defined search strategy
criteria to include studies
type of study (eg strictly randomised)
type of patient
type of outcome
Estimate of overall treatment effect
 What is the best sort of
evidence?
BEST
Up-to date systematic review of all
relevant randomised trials
Narrative review article
Textbook
WORST
 Where can you find
systematic reviews and
RCTs?
MEDLINE: but only about 50% of
relevant RCTs can be found here
Cochrane Library: in many hospital
libraries on CD ROM, or Internet
(Abstracts at: http://www.update-
software.com/ccweb/cochrane/revabstr/
abidx.htm)
What is in the April 2000 edition
of the Cochrane Library?

Cochrane Database of Systematic

Reviews: 795 Reviews
Database of Abstracts of Reviews of
Effectiveness: 2433 Review abstracts
Cochrane Controlled Trials Register:
264,000 References to RCTs and CCTs
Cochrane systematic review of the evidence
for thrombolytic therapy in acute ischaemic
stroke
Joanna Wardlaw
abstract available free at:
www.dcn.ed.ac.uk/csrg
or on CDROM
The Cochrane Library
Fatal intracerebral haemorrhage (ICH)
with thrombolysis for ischaemic stroke
 ICH after thrombolysis for ischaemic
stroke
Fatal ICH increased from 1% to 5%.
five fold increase (p<0.00001)
50 extra fatal ICH per 1000 treated
Fatal or non-fatal symptomatic ICH
increased from 3% to 10%.
three fold increase (p<0.00001)
70 extra haemorrhages per 1000 treated
Similar proportional increase in ICH with
different thrombolytic agents
Thrombolysis for ischaemic stroke: long-term outcome
 Overall, thrombolytic treatment within 6
hours of onset of acute ischaemic stroke
significantly improved long-term outcome

At final follow-up, patients treated with

thrombolysis had a highly significant 24%
reduction in the relative odds of a bad outcome
(2p <0.0001) (= 10% relative risk reduction).
For every 1000 patients treated < 6hrs, 65
avoided death or dependence.
Need to treat 16 to prevent one poor outcome
 Implications for research. A large
randomised trial of thrombolysis with
rt-PA in acute ischaemic stroke would
help answer:

how wide is the time window beyond

3hours (6,9, 12 hours?)
which patients benefit most?
which patients most likely to be
harmed?
is the risk of death reduced or not?
 Suggested trial design:
Randomised controlled trial of rT-PA
versus placebo in 5,000 patients.
< 6 hours symptom onset with CT scan
Eligibility based on the uncertainty
principle.
Clear indication to treat: TREAT
Clear contraindication: DONT TREAT
Uncertain: RANDOMISE
Careful characterisation at baseline
 Cochrane systematic review of the
randomised trials of anticoagulants in acute
ischaemic stroke (Gubitz et al CDSR 1999)
Anticoagulant Trials Patients Deaths

Unfractionated heparin sc 6 20 048 4349

Heparinoid iv/sc 3 1413 98
LMW heparin 6 1321 292
Thrombin Inhibitor 2 294 10
Unfractionated heparin iv 2 270 27
Oral anticoagulant 2 81 28

TOTAL 21 23 427* 4804

* of whom 95% had CT head scans
Recurrent ischaemic stroke or intracranial
haemorrhage during treatment period
5%
4.1% 4.1% NS
4%

3% 3.6% p<0.0001
2.7%
2%

1%
1.4% 0.5% p<0.001
0%
Anticoagulant Control
Recurrent Ischaemic Stroke
Symptomatic Intracranial Haemorrhage
Outcome at end of follow-up
70%
NS
60%

50%

40%

30% NS 59.7% 60.1%

20%

10% 21.4% 20.6%

0%
Dead Dead or dependent

Anticoagulant Control
 Summary of effects of anticoagulants
(mainly heparin) in acute ischaemic stroke
No net short- or long-term benefit.
No subgroup of patient or anticoagulant
regimen associated with clear net benefit.
Significant bleeding risk: 9 extra symptomatic
intracranial and 9 major extracranial
haemorrhages per 1000 patients treated.
Bleeding risk dose-related: High > Low > Nil
It gives no overall benefit, causes bleeds, is a
pain in the leg/arm/abdomen for the patient,
costs money (and nurses time); why use it?
 Indications for early aspirin use in
acute ischaemic stroke: a combined
analysis of over 40,000 randomised
patients from CAST and IST

Sandercock PAG, Chen ZM, on

behalf of IST and CAST
collaborative groups
Design features of CAST & IST
Randomised
Acute ischaemic stroke <48 hours
CT before entry where possible, or soon
after
Aspirin dose (scheduled treatment
period):
IST: 300mg daily vs open control (2 weeks)
CAST: 160mg daily vs placebo control (4
weeks)
 Patients included in the trials

CAST IST
No. randomised 20,655 19,435
CT scan
- Before entry 87% 68%
- Total 97% 96%
Heparin allocation 0 50%
Recurrent ischaemic stroke or intracranial
haemorrhage during treatment period
5 %
3.5% 4.0% 2p < 0.0001

3
2p < 0.0001
2 2.5 % 3.2%

1
1% 0.8% 2p = 0.07
0

Aspirin Control
Intracranial haemorrhage Recurrent ischaemic stroke
 Summary of aspirin benefit
For every 1000 patients started < 48 hrs of onset:
< 14 days, 7 avoid recurrent ischaemic stroke
at 6 months, 12 avoid death or dependency, &
an extra 10 make a complete recovery
The risk of cerebral haemorrhage is low (1-2 per
1000) and is completely outweighed by the
benefits
Early aspirin is of net benefit for a wide range of
patients, so prompt treatment should be
considered for almost all patients presenting with
suspected acute ischaemic stroke.
 Worldwide benefit each year of a
policy of 'give aspirin without delay' in
acute stroke
8 million patients with acute stroke
5 million with acute ischaemic stroke
1 million reach medical attention and get
aspirin
10,000 avoid a poor outcome, extra
10,000 make a complete recovery
Lesson: a small benefit in a large
number of people adds up to a
worthwhile benefit to mankind
What have we learned about thrombolysis,
anticoagulants and aspirin?
Thrombolysis: promising, but applicable to 1% of all
ischaemic strokes? Need much larger-scale trials.
Anticoagulants/heparin: benefits balanced by bleeding
risk. No net benefit.
Aspirin. Modest benefits, but applicable to almost all
patients. Like thrombolysis for AMI, It needed 40,000
randomised patients to prove it and persuade
clinicians to change
Effort and audit needed to ensure ALL patients with
acute ischaemic stroke get aspirin
CT has excluded haemorrhage?
patient able to swallow safely? -> oral aspirin
not able to swalow? -> rectally or via NG tube