Intended for healthcare professionals

Practice Uncertainties Page

Should more patients with acute ischaemic stroke receive thrombolytic treatment?

BMJ 2009; 339 doi: (Published 11 November 2009) Cite this as: BMJ 2009;339:b4584
  1. Joanna M Wardlaw, professor of applied neuroimaging 1,
  2. Peter A G Sandercock, professor of neurology1,
  3. Veronica Murray, senior scientist2
  1. 1Division of Clinical Neurosciences, University of Edinburgh, Western General Hospital, Edinburgh EH4 2XU
  2. 2Division of Medicine, Danderyd Hospital, Karolinska Institutet, SE-182 88 Stockholm, Sweden
  1. Correspondence to: J M Wardlaw joanna.wardlaw{at}
  • Accepted 24 June 2009

In developed countries, stroke is the third most common cause of death and the most common cause of dependency in adults. Thrombolysis with recombinant tissue plasminogen activator (rt-PA) was licensed for use in highly selected patients within three hours of acute ischaemic stroke in the United States in 1996 on the basis of the National Institutes of Neurological Disorders and Stroke (NINDS) randomised controlled trial (n=624),1 which showed substantially lower combined rates of death or dependency with this treatment (140/1000 fewer) despite an excess of symptomatic intracranial haemorrhage (60/1000 more). The Cochrane review of all data from randomised trials of rt-PA and a meta-analysis of individual patient data agreed with these findings.2 3 In Europe, a conditional licence for use within three hours in highly selected patients was granted in 2002 on the basis of the NINDS trial plus data from two European randomised controlled trials (n=930).4 5

However, six years after European and 10 years after US licensing, fewer than 10% of eligible patients receive thrombolysis—1-7% in the US,6 7 3% in Canada,8 3% in Germany,9 and 3.3% in Sweden (—and use of rt-PA varies greatly between European countries.10 Clinicians and managers are uncertain about how widely to use rt-PA in routine clinical practice.11 The fact that treatment licenses are based on data from fewer than 1000 randomised patients with narrow entry criteria, the substantial excess of symptomatic intracranial haemorrhage, plus the restricted licence conditions may worry many clinicians.

What is the evidence of the uncertainty?

In 2009, a systematic review of all randomised trials comparing thrombolysis with control (mostly placebo) in patients with acute ischaemic stroke analysed 26 trials of various thrombolytic drugs, 11 of which tested rt-PA up to six hours after stroke (n=3977) (figure).12 Compared with control, rt-PA given up to six hours after stroke increased the risk of symptomatic intracranial haemorrhage about threefold (odds ratio 3.28, 95% confidence interval 2.48 to 4.33; P=0.00001), non-significantly increased the risk of death at the end of follow-up (1.14, 0.95 to 1.38; P=0.16), and significantly reduced the proportion of patients with a poor outcome (death or dependence: 0.78, 0.68 to 0.88; P=0.0001). Significant heterogeneity existed between trials in the estimate of effect of rt-PA on poor outcome but not on symptomatic intracranial haemorrhage or death (table). Treatment within three hours resulted in a similar sized increase in symptomatic intracranial haemorrhage, with no clear effect on death, but a greater reduction in poor outcome. Treatment between three and six hours also seemed to be beneficial—the upper confidence interval at six hours was similar to the effect of treatment in a stroke unit.13 Restricting the analyses to within separate time strata (zero to three and three to six hours) removed the significant heterogeneity for the effect on poor functional outcome (table). The randomisation methods used caused some imbalances in prognostic variables (such as stroke severity) at baseline between the treatment groups, which may have influenced the apparent effect of rt-PA. However, aggregate data indicate that for every 1000 patients treated within six hours, about 60 will avoid a poor functional outcome (110/1000 if treated within three hours; table).


Systematic review of all randomised trials comparing thrombolysis with control in patients with acute ischaemic stroke.12 Effect of any thrombolytic agent v control on death or dependency at the end of follow-up

Outcome for all trials of recombinant tissue plasminogen activator in acute ischaemic stroke (up to October 2008)

View this table:

These findings are consistent with all the data on any thrombolytic drug versus control (fig 1), which showed that thrombolysis might be beneficial up to nine hours after stroke and for all drugs tested—by the intravenous or intra-arterial route—but with heterogeneity in the estimate of effect on poor outcomes and substantial uncertainties similar to those for rt-PA alone (figure).12

These data give rise to three areas of uncertainty. Firstly, the heterogeneity for the estimate of effect of rt-PA treatment on the softer end point of poor outcome (dependency is more difficult to assess than death) makes the interpretation of this key result less clear. This justifies the need for additional randomised data to obtain a more precise estimate of benefit (especially at later times) and determine more reliably who will benefit. In contrast, there is no heterogeneity (and hence little doubt) about the size of the relative risk of symptomatic intracranial haemorrhage, although reliably identifying those patients most at risk is difficult.

Secondly, because of the strict inclusion-exclusion criteria of the trials so far, the results apply to a small proportion of patients. All trials excluded patients over 80, so no reliable randomised evidence is available for these patients. In the United Kingdom each year, about 30 000 people over 80 have an acute ischaemic stroke, and in Sweden 43% of patients with acute ischaemic stroke are over 80, yet they cannot be treated within the licence because of their age. Furthermore, patients with early infarct signs on computed tomography, a history of stroke in the past three months, diabetes, or high blood pressure were also excluded from the trials. Although not excluded from some previous trials, we have little randomised evidence on rt-PA in lacunar stroke, posterior circulation stroke, and patients on aspirin. Hence, reliable randomised evidence is needed to inform clinical decision making in many patients with acute ischaemic stroke (box).

Uncertainties about thrombolysis in acute ischaemic stroke

  • What is the size of the reduction in death or dependency when treatment is given at different times after stroke onset?

  • What is the latest time for worthwhile benefit, and could it be more than six hours?

  • What is the effect on death?

  • Is the current upper age limit for treatment of 80 years justified?

  • What key clinical and radiological features identify patients most (or least) likely to benefit?

  • Should antithrombotics (an antiplatelet or anticoagulant) be coadministered to reduce the risk of early reocclusion after initial successful reperfusion?

Thirdly, the latest time for safe and effective treatment is unclear. Three hours may be too late for some, whereas treatment could be effective well beyond six hours in others. Although the Third European Cooperative Acute Stroke Study (ECASS 3)—a randomised controlled trial of rt-PA between three and 4.5 hours after stroke in 800 patients—increased confidence that treatment is effective beyond three hours, all patients had to meet the restrictive criteria imposed by the licensing authorities.14 Ischaemic stroke has many causes, dynamic pathophysiology, and variable severity, all of which affect the length of time that cerebral tissue is “rescuable” by reperfusion therapy and influence the risk-benefit balance. Post hoc analyses of two trials suggest that the benefit of thrombolysis varies with stroke severity, with severe strokes seeming to gain less benefit than mild to moderate ones. Patients with severe strokes reach hospital more quickly than those with mild strokes, further confounding the association between time and treatment effect. Thus we still know relatively little about how to predict which patients will benefit and which are at high risk of adverse events.

Is ongoing research likely to provide relevant evidence?

Unfortunately, the publication of large observational non-randomised registries does not provide reliable evidence on the risk-benefit balance in many key categories of patients.10 The most pressing need is for studies that will enable safe and optimally effective delivery of rt-PA to more patients of all ages, within three hours and beyond. The only relevant ongoing trial is the Third International Stroke Trial (IST-3) of intravenous rt-PA versus control up to six hours after ischaemic stroke ( With over 1800 patients recruited so far, it is already more than double the size of the largest existing trial and aims to recruit 3100 patients by mid-2011. It includes patients over 80 years (580 so far) and those with lacunar strokes (153 so far), thereby already increasing 10-fold and threefold the world evidence on rt-PA in these key subgroups. The median time to randomisation is four hours, and by including patients treated up to six hours, IST-3 will show whether the benefit extends to six hours, and if so, which patients benefit most. By including patients with previous stroke, who are on aspirin already, with radiological features of early ischaemic change, and leucoaraiosis, and by minimising the randomisation on key prognostic variables, IST-3 will answer many unanswered questions (box) and help change practice. With 3100 patients, IST-3 could detect a 4.7% absolute difference in poor outcomes (close to the 4% difference seen in the systematic review; table); and with 6000 patients, mostly treated three to six hours after onset, the trial could detect a 3% absolute difference in poor outcomes. IST-3 does not preclude further explanatory trials, such as ones that test extended time windows using imaging based selection criteria or different drugs. However, these trials necessarily have restrictive entry criteria and a small sample size. Greater statistical power is needed to deal with the heterogeneity in patients’ characteristics, which may influence the effects of rt-PA and hence the risk-benefit ratio in individual patients.

What should we do in the light of the uncertainty?

The current approved use of rt-PA within three hours is based on evidence from about 930 randomised patients. For treatment within six hours, the randomised evidence (3977 patients) is heterogeneous and small. We need additional randomised evidence from large trials to provide more precise point estimates of key outcomes and inform treatment decisions in individual patients.

Recommendation for future research

  • Population: patients who fall outside existing licence criteria and have no contraindication to recombinant tissue plasminogen activator (rt-PA), especially those over 80, with vascular disease, who reach hospital and can be treated within six hours

  • Intervention and comparison: intravenous rt-PA compared with control (no rt-PA)

  • Outcome: symptomatic intracranial haemorrhage and death within the first seven days of stroke; death or dependency at three months or later after stroke; quality of life

In the meantime, patients under 80 who reach hospital, meet strict existing licence criteria, have no contraindication to rt-PA, and can be treated within three hours should be treated within the licence. If the licence is extended to 4.5 hours, then similar patients could be treated up to that time. All other patients who reach hospital and could be treated at any time up to six hours, with no clear contraindication to rt-PA, should be considered for inclusion in a randomised trial, such as IST-3. Given that rt-PA works for stroke and has a licence, it is unfortunate that fewer than 10% of patients receive it and that we do not know who is at greatest risk of hazard and who has the most chance of benefit. Patients are entitled to have individual risk-benefit analyses for all therapeutic measures, not least for treatments with such huge potential benefit as rt-PA.


Cite this as: BMJ 2009;339:b4584


  • This is a series of occasional articles that highlights areas of practice where management lacks convincing supporting evidence. The series advisers are David Tovey, editor in chief, the Cochrane Library, and Charles Young, editor of BMJ Clinical Evidence and editor in chief, BMJ Point of Care. We welcome any suggestions for future articles (uncertainties.bmj{at}

  • Contributors: JMW conceived the article, analysed the data, and drafted the paper with VM; all three authors interpreted the data, revised the manuscript critically, and approved the final version. Heather Goodare, lay member of the IST-3 steering committee, and Richard Lindley, co-chief investigator of IST-3, also read the manuscript and provided lay and scientific comment. JMW is guarantor.

  • Funding: JMW is part funded by the Scottish Funding Council through the SINAPSE Collaboration (Scottish Imaging Network, A Platform for Scientific Excellence, VM is in part funded by the Swedish Heart-Lungfund, AFA Insurances.

  • Competing interests: The authors are on the IST-3 trial steering committee.

  • Provenance and peer review: Not commissioned; externally peer reviewed.