Vitamin K antagonists versus antiplatelet therapy after transient ischaemic attack or minor ischaemic stroke of presumed arterial origin

Els LLM De Schryver; Ale Algra; L Jaap Kappelle; Jan van Gijn; Peter J Koudstaal

doi:10.1002/14651858.CD001342.pub3

. 2012 Sep 12;2012(9):CD001342. doi: 10.1002/14651858.CD001342.pub3

Vitamin K antagonists versus antiplatelet therapy after transient ischaemic attack or minor ischaemic stroke of presumed arterial origin

Els LLM De Schryver ¹, Ale Algra ^2,^✉, L Jaap Kappelle ³, Jan van Gijn ³, Peter J Koudstaal ⁴

Editor: Cochrane Stroke Group

PMCID: PMC7055052 PMID: 22972051

Abstract

Background

People who have had a transient ischaemic attack (TIA) or non‐disabling ischaemic stroke have an annual risk of major vascular events of between 4% and 11%. Aspirin reduces this risk by 20% at most. Secondary prevention trials after myocardial infarction indicate that treatment with vitamin K antagonists is associated with a risk reduction approximately twice that of treatment with antiplatelet therapy.

Objectives

To compare the efficacy and safety of vitamin K antagonists and antiplatelet therapy in the secondary prevention of vascular events after cerebral ischaemia of presumed arterial origin.

Search methods

We searched the Cochrane Stroke Group Trials Register (last searched 15 September 2011), the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2011, Issue 3), MEDLINE (2008 to September 2011) and EMBASE (2008 to September 2011). In an effort to identify further relevant trials we searched ongoing trials registers and reference lists. We also contacted authors of published trials for further information and unpublished data.

Selection criteria

Randomised trials of oral anticoagulant therapy with vitamin K antagonists (warfarin, phenprocoumon or acenocoumarol) versus antiplatelet therapy for long‐term secondary prevention after recent transient ischaemic attack or minor ischaemic stroke of presumed arterial origin.

Data collection and analysis

Two review authors independently selected trials, assessed trial quality and extracted data.

Main results

We included eight trials with a total of 5762 participants. The data showed that anticoagulants (in any intensity) are not more efficacious in the prevention of recurrent ischaemic stroke than antiplatelet therapy (medium intensity anticoagulation: relative risk (RR) 0.80, 95% confidence interval (CI) 0.56 to 1.14; high intensity anticoagulation: RR 1.02, 95% CI 0.49 to 2.13).

There is no evidence that treatment with low intensity anticoagulation gives a higher bleeding risk than treatment with antiplatelet agents: RR 1.27 (95% CI 0.79 to 2.03). However, it was clear that medium and high intensity anticoagulation with vitamin K antagonists, with an INR of 2.0 to 4.5, were not safe because they yielded a higher risk of major bleeding complications (medium intensity anticoagulation: RR 1.93, 95% CI 1.27 to 2.94; high intensity anticoagulation: RR 9.0, 95% CI 3.9 to 21).

Authors' conclusions

For the secondary prevention of recurrent ischemic stroke after TIA or minor stroke of presumed arterial origin, there is sufficient evidence to conclude that vitamin K antagonists in any dose are not more efficacious than antiplatelet therapy and that medium and high intensity anticoagulation leads to a significant increase in major bleeding complications.

Keywords: Humans; Administration, Oral; Anticoagulants; Anticoagulants/adverse effects; Anticoagulants/therapeutic use; Cause of Death; Hemorrhage; Hemorrhage/chemically induced; International Normalized Ratio; Ischemic Attack, Transient; Ischemic Attack, Transient/drug therapy; Platelet Aggregation Inhibitors; Platelet Aggregation Inhibitors/adverse effects; Platelet Aggregation Inhibitors/therapeutic use; Randomized Controlled Trials as Topic; Secondary Prevention; Stroke; Stroke/prevention & control; Vitamin K; Vitamin K/antagonists & inhibitors

Plain language summary

Vitamin K antagonists versus antiplatelet therapy after transient ischaemic attack or minor ischaemic stroke of presumed arterial origin

People who have a stroke due to a blockage of an artery have a higher risk of having another possibly fatal stroke, or a heart attack. Treatment with antiplatelet drugs (like aspirin) definitely reduces this risk. Blood thinning treatment (anticoagulation by vitamin K antagonists) was believed to provide added protection. We reviewed eight trials involving 5762 participants that compared anticoagulants with antiplatelet agents for preventing recurrent stroke and found no benefit of low intensity anticoagulation over aspirin, and an increased risk of bleeding with high intensity anticoagulation.

Background

Description of the condition

Patients who are entered in clinical trials after a transient ischaemic attack (TIA) or non‐disabling ischaemic stroke have an annual risk of major vascular events (death from all vascular causes, non‐fatal stroke, or non‐fatal myocardial infarction) of between 4% and 11% (Algra 1996; APT I 1994). The corresponding estimate for population‐based studies is 9% per year (Warlow 1992).

Description of the intervention

Secondary prevention trials after myocardial infarction indicate that treatment with vitamin K antagonists is associated with a risk reduction approximately twice that of treatment with aspirin or other antiplatelet drugs (Anand 1999; APT I 1994; ASPECT 1994; EPSIM 1982; Sixty Plus 1980; Smith 1990). Both treatment with aspirin and with vitamin K antagonists increases the risk of bleeding complications (Algra 1996; Hirsh 1991). Few data are available on the efficacy and safety of vitamin K antagonists in patients with cerebral ischaemia (Jonas 1988).

How the intervention might work

A Cochrane review on the efficacy of vitamin K antagonists compared with control found nine old trials, performed before 1980 and of poor methodological quality, and two newer trials conducted before 2000. The review concluded that there is no clear evidence of benefit from long‐term anticoagulation, but that there is evidence of a significant bleeding hazard (Sandercock 2009). In patients with cerebral ischaemia and atrial fibrillation, however, vitamin K antagonists were found to be safe and significantly more effective than aspirin (EAFT 1993; Koudstaal 1996; Saxena 2004a; Saxena 2004b).

Why it is important to do this review

Aspirin, in a daily dose of 30 mg or more, offers only modest protection after cerebral ischaemia; it reduces the incidence of major vascular events by 20% at most (Algra 1996; Algra 1999; APT I 1994). In direct comparisons no differences in efficacy were found between doses of 300 mg and 1200 mg (UK‐TIA Study 1991) and 30 mg and 283 mg (Dutch TIA Trial 1991).

Objectives

To compare the efficacy of vitamin K antagonists and antiplatelet drugs in the secondary prevention of vascular events after cerebral ischaemia of presumed arterial origin.
To compare the safety of vitamin K antagonists and antiplatelet drugs in the secondary prevention of vascular events after cerebral ischaemia of presumed arterial origin.

Methods

Criteria for considering studies for this review

Types of studies

Randomised trials with concealed treatment allocation examining long‐term (more than six months) secondary prevention after recent (less than six months) cerebral ischaemia of presumed arterial origin.

Types of participants

Patients after cerebral ischaemia (TIA or minor ischaemic stroke) of presumed arterial origin, that is without demonstrable cardiac origin. The symptoms in TIAs or minor ischaemic strokes should develop within a few seconds, should not progress from one part of the body to another in an orderly march, and should last at least one minute (Dutch TIA Trial 1988). We excluded acute (onset less than 48 hours ago) ischaemic events as well as those that occurred more than six months previously.

Types of interventions

Vitamin K antagonists of specified intensity (by means of the International Normalised Ratio: INR) with warfarin, phenprocoumon or acenocoumarol.
A single antiplatelet drug in any dose or a combination of antiplatelet drugs.

Types of outcome measures

Efficacy outcome measures

The composite event 'vascular death, non‐fatal stroke, non‐fatal myocardial infarction or major bleeding complication' (primary outcome measure)
All‐cause death
Vascular death
Vascular death or non‐fatal stroke
Vascular death, non‐fatal stroke or non‐fatal myocardial infarction
Recurrent stroke (fatal or non‐fatal; ischaemic or unknown pathologic type)
Recurrent ischaemic stroke or intracranial haemorrhage
Death or dependency at end of follow‐up

Safety outcome measures

Major bleeding complication, i.e. any fatal or non‐fatal intracranial or major extracranial haemorrhage
Fatal intracranial or extracranial haemorrhage
Intracranial haemorrhage, fatal or non‐fatal
Major extracranial haemorrhage, according to the definition of the original investigator

Search methods for identification of studies

See the 'Specialized register' section of the Cochrane Stroke Group module. We searched for trials in all languages and arranged translation of articles published in languages other than English.

Electronic searches

We searched the Cochrane Stroke Group Trials Register, which was last searched by the Managing Editor on 15 September 2011. In addition, we searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2011, Issue 3) (Appendix 1), MEDLINE (2008 to September 2011) (Appendix 2) and EMBASE (2008 to September 2011) (Appendix 3).

The Cochrane Stroke Group Trials Search Co‐ordinator devised the search strategies for the electronic databases and, to avoid duplication of effort, we limited the search of MEDLINE and EMBASE to 2008 onwards. We completed searches of these databases to January 2008 for all stroke trials and all relevant trials have been identified and are included in the Cochrane Stroke Group.

In an effort to identify further published, unpublished and ongoing trials we searched the following ongoing trials registers (last search 10 May 2012):

ClinicalTrials.gov (http://www.clinicaltrials.gov/);
EU Clinical Trials Register (https://www.clinicaltrialsregister.eu);
Stroke Trials Registry (www.strokecenter.org/trials/);
Current Controlled Trials (www.controlled‐trials.com);
WHO International Clinical Trials Registry Platform (http://apps.who.int/trialsearch/).

Searching other resources

We also:

searched the reference lists of relevant studies;
contacted the authors of published trials for further information and unpublished data;
used Science Citation Index Cited Reference Search for forward tracking of important articles.

We have had no explicit contact with manufacturers in the search for unpublished trials.

Data collection and analysis

Selection of studies

Two review authors (AA, EDS) independently selected trials that met the inclusion criteria.

Data extraction and management

The same two review authors extracted details of randomisation methods, blinding of treatments and assessments, whether intention‐to‐treat analysis was possible from the published data, whether treatment groups were comparable with regard to major prognostic risk factors for outcomes, the number of patients excluded or lost to follow‐up, definitions of outcomes, and entry and exclusion criteria. The same two review authors assessed the methodological quality of each trial by using these extracted data. In addition, we recorded target INRs for anticoagulant treatment and the dose and type of antiplatelet treatment, duration of follow‐up and the numbers of defined outcome events. The same two review authors independently extracted and cross‐checked all data; they resolved discrepancies by discussion.

Assessment of risk of bias in included studies

We used the Cochrane tool for assessing risk of bias to assess the methodological quality of the included studies (Higgins 2011). The tool covers the domains of sequence generation, allocation concealment, blinding of participants and personnel, and blinding of outcome assessments. We classified items as 'low risk', 'high risk' or 'unclear risk' of bias. We resolved disagreements with help from a third review author.

Measures of treatment effect

We used risk ratios to measure the effect.

Unit of analysis issues

The outcome measure that occurred first was analysed from every individual randomised patient.

Dealing with missing data

We contacted the authors of included trials for further information and unpublished data.

Assessment of heterogeneity

We used a fixed‐effect model and assessed heterogeneity by means of the I² statistic. We analysed post hoc subgroups between different levels of vitamin K antagonists. We chose the categories according to the anticoagulation levels used in the trials.

Assessment of reporting biases

None

Data synthesis

We analysed the data according to the intention‐to‐treat principle.

Subgroup analysis and investigation of heterogeneity

We planned the following subgroup analyses in advance: age (65 years of age or less versus over 65 years of age), sex, history of ischaemic heart disease and type of cerebral ischaemia (large vessel disease versus small vessel disease). We planned separate analyses on the comparison of vitamin K antagonists and aspirin only. A separate analysis also included studies in which patients with intracranial stenosis were analysed (because of a possible different pathophysiologic mechanism of stroke) and a separate analysis included concealed trials only. If outcome data on randomised patients were lacking, we planned both a best and worst case scenario: the best case scenario (with regards to treatment) assumed that none of the patients excluded from the analysis in the treatment group had the outcome of interest whilst all those excluded from the reference group did, and vice versa for the worst case analysis. We calculated risk ratios using the Cochrane Review Manager software (RevMan 2011). We used a fixed‐effect model and assessed heterogeneity by means of the I² statistic. We analysed post hoc subgroups between different levels of vitamin K antagonists. We chose the categories according to anticoagulation levels used in the trials.

Sensitivity analysis

None

Results

Description of studies

Results of the search

We identified a total of 13 randomised controlled trials (RCTs). Six trials met all our inclusion criteria (AVASIS 2006; ESPRIT 2007; SPIRIT 1997; SWAT 1998; WARSS 2001; WASID 2005). Because the number of selected trials was low, we decided to include two more RCTs which did not specify the intensity of vitamin K antagonists by means of the INR and did not specify the method of randomisation concealment (Garde 1983; Olsson 1980). We did not obtain this information from the study authors: we estimated the therapeutic INR range with the help of a conversion table which reflects the relationship between the thrombotest and INR (Van den Besselaar 1993). This is not an exact estimation since there are differences between the thromboplastin reagents used (ICSH/ICTH 1985).

Included studies

Of the eight included RCTs, which involved a total of 5762 patients, four were performed in Europe, three in North America and one in Europe, Australia and Asia. The mean age of the patients ranged from 60 to 68 years. All eight studies included patients with TIAs or an ischaemic stroke not fully recovered from within 24 hours. Two studies only selected patients with a proven intracranial major artery stenosis as well (AVASIS 2006; WASID 2005). The baseline characteristics were comparable in all included studies. The treatments in the included trials were given in an open, non‐blinded fashion except in WARSS 2001 and WASID 2005, which were blinded.

One of the included studies did not distinguish between cerebral ischaemia caused by atherosclerosis or cardiac embolism, for example atrial fibrillation (Olsson 1980). Two trials did not define exclusion of severe stroke (Garde 1983; SWAT 1998). The mean duration of the follow‐up varied from 12.4 months to 4.6 years. For one trial data were only available from an abstract; the abstract did not specify the duration of follow‐up (SWAT 1998). This information was not supplied after requesting it.

Garde 1983 and Olsson 1980 did not specify the outcome 'non‐fatal myocardial infarction' and SWAT 1998 also did not specify 'vascular death' and whether bleeding complications were intra‐ or extracranial. We used the WHO stroke definition as a focal neurological deficit lasting for more than 24 hours. Therefore, we classified all outcome events described as reversible ischaemic neurological deficit (RIND) or TIA‐incomplete recovery (TIA‐IR) as 'ischaemic stroke'. Six trials gave a definition of a bleeding complication (AVASIS 2006; ESPRIT 2007; Olsson 1980; SPIRIT 1997; WARSS 2001; WASID 2005), which included hospitalisation or a blood transfusion, or both. The definition was not clear in the other included studies and we tried to abstract the data on bleeding by interpretation of their descriptions. We could only analyse data on the outcome 'death or dependent at end of follow‐up' in one trial (SPIRIT 1997) because the authors of this review conducted that study.

One trial (WARSS 2001) studied low intensity anticoagulation (INR 1.4 to 2.8) versus aspirin. However, the primary outcome event 'death from any cause or recurrent ischaemic stroke' did not fit in this meta‐analysis. No further data on other vascular outcome events were published and we have not been able to obtain these data from the authors.

Excluded studies

We excluded five trials: three of these studies did not use concealed treatment allocation (Buren 1981; Eriksson 1985; Olsson 1981) and in two trials the method of randomisation was not specified and we could not obtain this information (Norrving 1983; Ortiz Castellon 1992). Additionally, Norrving 1983 did not specify the intensity of anticoagulation and also did not clarify whether patients with stroke caused by cardiac emboli were excluded. The publication of Ortiz Castellon 1992 did not contain information about the duration of the treatment and the exact number of outcome events. We could not obtain this additional information.

Risk of bias in included studies

In two trials the method of concealment of randomisation was not defined (Garde 1983; Olsson 1980); in the six other trials randomisation was concealed (AVASIS 2006; ESPRIT 2007; SPIRIT 1997; SWAT 1998; WASID 2005). Two trials were blinded for treatment (WARSS 2001; WASID 2005) and four stated blinding of outcome assessment (ESPRIT 2007; SPIRIT 1997; WARSS 2001; WASID 2005).

In Garde 1983 two intensities of anticoagulation were used, which were assumed to be equivalent to each other: a thrombotest activity of 7% to 15% or Simplastin A test value of 10% to 25% which, however, are not the same. The reference they quoted describes a relationship between thrombotest activity 7% to 18% and Simplastin A 14% to 34% and 10% to 25% P&P Human thromboplastin (Korsan‐Bengtsen 1970).

Five trials specifically stated the number of lost patients and the possibility of an intention‐to‐treat analysis (AVASIS 2006; ESPRIT 2007; SPIRIT 1997; WARSS 2001; WASID 2005); we do not have this information for one trial (SWAT 1998). The other trials suggested that none of the patients were lost to follow‐up, but there was no explicit statement about this nor about the possibility of intention‐to‐treat analysis. Moreover, the monitoring of compliance was not very strict if done at all.

Effects of interventions

For the analyses we presumed that the published outcome events were based on intention‐to‐treat analysis if this was not stated specifically and that if no loss to follow‐up was reported and no patients were lost. This assumption was based on the fact that the trials included that did not make any statements about this were relatively small trials with a short period of follow‐up. The bias created by a wrong assumption will therefore not be of major influence. We present the results as risk ratios (RR) with 95% confidence intervals (CI), analysed with a fixed‐effect model. We could not perform the planned subgroup analyses because we could not obtain enough data for adequate analyses. Since few participants were described as lost, we did not perform a worst or best case scenario analysis.

Because the SPIRIT 1997 trial showed that risk of bleeding complications increased sharply with the achieved INR, we decided to separate the trials with lower levels of anticoagulation intensity from those with a higher level. We chose the intensity groups pragmatically according to the levels used in the trials. We defined low intensity anticoagulation as a target INR of 1.4 to 2.8, medium intensity anticoagulation as INR 2.0 to 3.6 and high intensity anticoagulation as INR 3.0 to 4.5 INR. Hence we present the results for the three categories of trials separately.

Vitamin K antagonists versus antiplatelet therapy (Analysis 1)

The composite outcome 'vascular death, non‐fatal stroke, non‐fatal myocardial infarction or major bleeding complication' was reported by two trials (ESPRIT 2007; SPIRIT 1997). The risk of the outcome in the high intensity anticoagulation trial was more than two times higher in the anticoagulated patients than in those using aspirin (RR 2.30, 95% CI 1.58 to 3.35) (Analysis 1.1). The excess was due to a higher bleeding frequency in the anticoagulated patients (see below).

1.1 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 1 The composite vascular death, non‐fatal stroke, non‐fatal myocardial infarction or major bleeding complication.

The numbers of 'deaths from all causes' tended to be higher with the medium intensity anticoagulation trials (RR 1.33, 95% CI 0.94 to 1.88 for concealed trials plus trials with unknown concealment together) (Analysis 1.2). In the low intensity trial the RR was 0.89 (95% CI 0.60 to 1.30). In the one high intensity anticoagulation trial more patients on anticoagulants died than on antiplatelet drugs (RR 2.38, 95% CI 1.31 to 4.32).

1.2 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 2 All death.

In all trials that analysed this outcome event the majority of patients died from a 'vascular cause of death', hence the relative risks for vascular death did not differ importantly from those for all deaths (Analysis 1.3).

1.3 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 3 Vascular death.

For the combination of 'vascular death and stroke', there was no difference between medium intensity anticoagulation and antiplatelet therapy: RR of 0.93 (95% CI 0.71 to 1.22) (Analysis 1.4). Among patients who received high intensity anticoagulation the risk of this outcome event was increased (RR 1.69, 95% CI 1.08 to 2.65).

1.4 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 4 Vascular death or non‐fatal stroke.

Data on the outcome criterion 'vascular death, non‐fatal stroke or non‐fatal myocardial infarction' were available only for two trials (ESPRIT 2007; SPIRIT 1997) for anticoagulation versus antiplatelet therapy: the RR was 0.85 (95% CI 0.65 to 1.12) in the medium intensity group and RR of 1.70 (95% CI 1.12 to 2.59) in the high intensity group (Analysis 1.5).

1.5 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 5 Vascular death, non‐fatal stroke or non‐fatal myocardial infarction.

For the outcome 'recurrent stroke' (fatal or non‐fatal; ischaemic or unknown pathologic type, that is excluding haemorrhage confirmed by autopsy or imaging) there was no evidence of a difference between the different treatments. This observation was regardless of the intensity of anticoagulation. The RR of all trials together was 0.84 (95% CI 0.61 to 1.15) (Analysis 1.6).

1.6 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 6 Recurrent ischaemic stroke.

For 'recurrent ischaemic stroke or intracranial haemorrhage' there was no important difference in the medium intensity anticoagulation trials: RR of 0.91 (95% CI 0.66 to 1.29) for the concealed plus unconcealed trials together. Treatment with anticoagulation INR 3.0 to 4.5 resulted in a higher risk of bleeding complications than with antiplatelet therapy: RR of 2.04 (CI 1.21 to 3.46) (Analysis 1.7).

1.7 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 7 Recurrent ischaemic stroke or intracranial haemorrhage.

In SPIRIT 1997 the RR of anticoagulation versus antiplatelet therapy for the outcome event of 'death or dependency at the end of follow‐up' was 2.30 (95% CI 1.37 to 3.85) (Analysis 1.8).

1.8 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 8 Death or dependent at end of follow‐up.

The RR of a 'major bleeding complication', that is any fatal or non‐fatal intracranial or major extracranial haemorrhage, during treatment with low intensity anticoagulation was 1.27 (95% CI 0.79 to 2.03) and with anticoagulation INR 2.0 to 3.6 the RR was 1.93 (CI 1.27 to 2.94) as compared with antiplatelet therapy. High intensity anticoagulation increased the risk of major bleeding complications importantly compared with antiplatelet therapy: RR of 9.02 (95% CI 3.91 to 20.8) (Analysis 1.9).

1.9 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 9 Major bleeding complication.

A large proportion of the 'intracranial haemorrhages' were fatal. The confidence interval for both fatal and intracranial bleeding was wide in the low anticoagulation trial, with a point estimate RR of 1.40; and also in the medium anticoagulation comparison, with a point estimate RR of 2.18. For anticoagulation with INR 3.0 to 4.5 the RR of a fatal haemorrhage was 17.4 (95% CI 2.32 to 130) (Analysis 1.10) and the RR of an intracranial haemorrhage was 9.2 (95% CI 2.80 to 30.2) (Analysis 1.11).

1.10 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 10 Fatal intracranial or extracranial haemorrhage.

1.11 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 11 Intracranial haemorrhage, fatal or non‐fatal.

For the outcome 'major extracranial haemorrhage', according to the definition of the original investigator, there was an RR of 2.77 (95% CI 1.55 to 4.96) in the medium intensity anticoagulation group and an RR of 8.85 (95% CI 2.69 to 29.1) in the high intensity anticoagulation group (Analysis 1.12).

1.12 — Comparison 1 Vitamin K antagonists versus antiplatelet therapy, Outcome 12 Major extracranial haemorrhage.

Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials) (Analysis 2)

Analyses with only the concealed trials did not show major differences compared with the analyses of all trials (Analysis 2.1; Analysis 2.2; Analysis 2.3; Analysis 2.4; Analysis 2.5; Analysis 2.6; Analysis 2.7; Analysis 2.8; Analysis 2.9; Analysis 2.10; Analysis 2.11; Analysis 2.12).

2.1 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 1 The composite vascular death, non‐fatal stroke, non‐fatal myocardial infarction or major bleeding complication.

2.2 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 2 All death.

2.3 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 3 Vascular death.

2.4 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 4 Vascular death or non‐fatal stroke.

2.5 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 5 Vascular death, non‐fatal stroke or non‐fatal myocardial infarction.

2.6 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 6 Recurrent ischaemic stroke.

2.7 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 7 Recurrent ischaemic stroke or intracranial haemorrhage.

2.8 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 8 Death or dependent at end of follow‐up.

2.9 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 9 Major bleeding complication.

2.10 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 10 Fatal intracranial or extracranial haemorrhage.

2.11 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 11 Intracranial haemorrhage, fatal or non‐fatal.

2.12 — Comparison 2 Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials), Outcome 12 Major extracranial haemorrhage.

Vitamin K antagonists versus aspirin only (Analysis 3)

The analyses of studies comparing anticoagulation versus aspirin alone (ESPRIT 2007; Garde 1983; SPIRIT 1997; SWAT 1998; WARSS 2001) gave similar results as those of anticoagulation versus any antiplatelet therapy (aspirin combined with dipyridamole in one trial (Olsson 1980) was omitted for this analysis) (Analysis 3.1; Analysis 3.2; Analysis 3.3; Analysis 3.4; Analysis 3.5; Analysis 3.6; Analysis 3.7; Analysis 3.8; Analysis 3.9; Analysis 3.10; Analysis 3.11; Analysis 3.12).

3.1 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 1 The composite vascular death, non‐fatal stroke, non‐fatal myocardial infarction or major bleeding complication.

3.2 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 2 All death.

3.3 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 3 Vascular death.

3.4 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 4 Vascular death or non‐fatal stroke.

3.5 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 5 Vascular death, non‐fatal stroke or non‐fatal myocardial infarction.

3.6 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 6 Recurrent ischaemic stroke.

3.7 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 7 Recurrent ischaemic stroke or intracranial haemorrhage.

3.8 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 8 Death or dependent at end of follow‐up.

3.9 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 9 Major bleeding complication.

3.10 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 10 Fatal intracranial or extracranial haemorrhage.

3.11 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 11 Intracranial haemorrhage, fatal or non‐fatal.

3.12 — Comparison 3 Vitamin K antagonists versus aspirin only, Outcome 12 Major extracranial haemorrhage.

The additive effect of dipyridamole to aspirin has been analysed in another Cochrane review (De Schryver 2007).

Vitamin K antagonists versus aspirin, including intracranial artery stenosis (Analysis 4)

Two trials included patients with a TIA or minor ischaemic stroke and a major cerebral artery stenosis of 50% to 99%: AVASIS 2006 studied stenosis of the middle cerebral artery only whereas WASID 2005 studied stenosis of any major intracranial artery. The addition of results of these analyses did not lead to substantially different results from the original ones without the two trials, except for the outcome events:

all deaths, in the medium intensity anticoagulation group there now was significantly more deaths, RR of 1.47 (95% CI 1.08 to 2.00) (Analysis 4.2);
fatal intracranial or extracranial haemorrhage reached statistical significance now to the disadvantage of the medium intensity anticoagulation group, RR of 2.64 (95% CI 1.04 to 6.70) (Analysis 4.10).

4.2 — Comparison 4 Vitamin K antagonists versus aspirin, including intracranial artery stenosis, Outcome 2 All death.

4.10 — Comparison 4 Vitamin K antagonists versus aspirin, including intracranial artery stenosis, Outcome 10 Fatal intracranial or extracranial haemorrhage.

Discussion

Summary of main results

There are two main issues that influence the interpretation of this review, the level of anticoagulation used and the quality of the trials.

Level of anticoagulation

From SPIRIT 1997 it emerged that there was a strong relationship between the intensity of anticoagulation and bleeding risk (Gorter 1999). Thus we decided to present the results of this review according to the level of anticoagulation: low intensity (INR 1.4 to 2.8), medium intensity (INR 2.0 to 3.6) and high intensity (INR 3.0 to 4.5). Analysis of all trials together was not useful since there is only one large trial (SPIRIT 1997) in the high intensity anticoagulation subgroup, with many bleeding complications, outweighing all other smaller trials in the medium anticoagulation group. Because of the high bleeding risk with high intensity anticoagulation, all composite outcome events containing bleeding complications were all to the disadvantage of anticoagulation in SPIRIT 1997. A large proportion of the vascular deaths was also due to bleeding complications in the high intensity anticoagulation group. The intensity of anticoagulation was specified by means of the INR in six trials (AVASIS 2006; ESPRIT 2007; SPIRIT 1997; SWAT 1998; WARSS 2001; WASID 2005). To include more studies in this review we estimated the INR level for two other studies, but this is not an exact approach (ICSH/ICTH 1985). One trial (Garde 1983) provided a confusing description of the level of anticoagulation by reporting two different intensity levels (see Risk of bias in included studies). The method of monitoring treatment was also poorly reported.

Efficacy of vitamin K antagonists compared with antiplatelet treatment

For the outcome events without bleeding complications (that is the ischaemic events) after TIA or minor stroke of presumed arterial origin in the secondary prevention of further vascular events, there were enough data to conclude that oral anticoagulation with vitamin K antagonists in low and medium intensities is not superior to antiplatelet therapy in patients without cardiac emboli.

Safety of vitamin K antagonists compared with antiplatelet treatment

It is clear that high intensity vitamin K antagonists (INR 3.0 to 4.5) are not safe, because they yield a significantly higher risk of major bleeding complications. Treatment at low intensity (INR 1.4 to 2.8) was not associated with a clearly higher risk of bleeding than treatment with antiplatelet therapy. The WARSS trial evaluated low intensity anticoagulation (INR 1.4 to 2.8; median achieved INR 1.9) compared with aspirin. For the combined outcome of death or recurrent ischaemic stroke the hazard ratio was 1.13 (95% CI 0.92 to 1.38) (WARSS 2001). In the same trial, the hazard ratio for major bleeding complication was 1.27 (95% CI 0.79 to 2.03). These results are consistent with the hypothesis that low intensity warfarin is likely to be of comparable safety to aspirin. At present there are no direct randomised comparisons of one intensity of vitamin K antagonists versus another. However, an observational study in Leiden, on the use of vitamin K antagonists among 356 patients observed for a total of 644 patient years, suggested that the risk of ischaemic and haemorrhagic events was lowest in this type of patient when the INR was in the range 2.5 to 3.5, with a nadir of the event rate at INR 3.0 (Torn 2001). This review is now updated with two randomised trials that investigated medium intensity anticoagulation (INR 2.0 to 3.6); there is no evidence that vitamin K antagonists are superior to antiplatelet therapy in preventing ischaemic events, but there is a clear increase in risk for bleeding complications.

Note: the WARSS trial (WARSS 2001) is the largest trial in this review and has published tabulated numbers of outcome events by allocated treatment for death, major haemorrhages and for the combined outcomes of recurrent ischaemic stroke or death recurrent ischaemic stroke or death or major haemorrhage. The trialists have not yet published tabulations of the number of patients for the first occurrence of each individual event by allocated treatment in a way that would permit the inclusion of that outcome data in this review. Furthermore, the trialists have not provided these data to us. When such data are publicly available, we will update the review to include them.

Overall completeness and applicability of evidence

The number of patients lost to follow‐up or with missing data in the included studies is very low and therefore we conclude that there is a good applicability of evidence.

Quality of the evidence

There were important methodological differences between the included studies. The method of concealment was not known in two trials (Garde 1983; Olsson 1980). Analysis without these trials did not show any differences in outcome, and the number of concealed trials was low. We assumed that the authors of the trials listed analysed their data according to the intention‐to‐treat principle, but we were not sure about it as we did not know whether any patients were lost to follow‐up. For this reason we were not able to perform worst and best case scenario analyses. CT scans of the brain were not required in some trials (Olsson 1980; SWAT 1998). Exclusion of atrial fibrillation as the cause of stroke was not specified in one trial (Olsson 1980). Patients with cerebral ischaemia and atrial fibrillation have a different risk profile; for these patients evidence is available that anticoagulation is the first choice in secondary prevention (EAFT 1993). Only two trials were double‐blind studies (WARSS 2001; WASID 2005); the open treatment studies could potentially give a bias, but on the other hand open treatment approaches reflect 'real life' more accurately. Moreover, in the largest two open studies outcome assessment was blind. Finally, the definition of bleeding complications was not always clear and differed between the trials. We had to rely on our own interpretation of a major bleeding complication in some cases. The data were dominated by the SPIRIT 1997 trial, which had a low risk of bias. The overall inclusion or exclusion of the other trials did not materially alter the conclusion of the review.

Potential biases in the review process

Since two review authors extracted the data independently and cross‐checked the results, the risk of potential bias is low. The authors consider it unlikely that they have missed important trials in their search for data, because of the extensive literature searches and the fact that they have been working in this field for over 20 years.

Agreements and disagreements with other studies or reviews

There are no major differences between the included studies and other reviews.

Authors' conclusions

Implications for practice.

There is no evidence to support the routine use of vitamin K antagonists in any dose in the prevention of further vascular events after transient ischaemic attack or stroke of presumed arterial origin.

Implications for research.

Further trials of vitamin K antagonist versus aspirin after transient ischaemic attack or minor stroke of presumed arterial origin are difficult to justify. Given the non‐inferiority of new direct thrombin and factor Xa inhibitors as compared with vitamin K antagonists in patients with atrial fibrillation, and their tendency to result in fewer major haemorrhages, it is conceivable that trials with these new agents will be designed for patients with cerebral ischaemia of arterial origin

Feedback

Summary

Feedback received for this review, and other reviews and protocols of anticoagulants, is available on the Cochrane Editorial Unit website at http://www.editorial‐unit.cochrane.org/anticoagulants‐feedback

What's new

Date	Event	Description
14 January 2013	Amended	Minor amendment to the Abstract

Date	Event	Description
1 February 2012	New search has been performed	We updated the search of the Cochrane Stroke Group Trials Register. We included three new studies (AVASIS 2006; ESPRIT 2007; WASID 2005), bringing the total number of included studies to eight, involving 5762 participants. We have updated the text of the review throughout.
1 February 2012	New citation required and conclusions have changed	The conclusions are now more clearly based.
7 March 2011	Feedback has been incorporated	Link to feedback added.
4 September 2008	Amended	Converted to new review format.
21 February 2005	New search has been performed	The WARSS trial was published 15 November 2001. Data that could be extracted from the publication were incorporated in the review. A minor update in early 2005 revealed only one new ongoing study; the conclusions of the review did not change.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 The composite vascular death, non‐fatal stroke, non‐fatal myocardial infarction or major bleeding complication	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 INR 2.0 ‐ 3.6	1	1068	Risk Ratio (M‐H, Fixed, 95% CI)	1.00 [0.78, 1.29]
1.2 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	2.30 [1.58, 3.35]
2 All death	6		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
2.1 INR 1.4 ‐ 2.8	1	2206	Risk Ratio (M‐H, Fixed, 95% CI)	0.89 [0.60, 1.30]
2.2 INR 2.0 ‐ 3.6	4	1561	Risk Ratio (M‐H, Fixed, 95% CI)	1.33 [0.94, 1.88]
2.3 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	2.38 [1.31, 4.32]
3 Vascular death	4		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.1 INR 2.0 ‐ 3.6	3	1444	Risk Ratio (M‐H, Fixed, 95% CI)	1.35 [0.84, 2.16]
3.2 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	2.23 [1.10, 4.51]
4 Vascular death or non‐fatal stroke	4		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
4.1 INR 2.0 ‐ 3.6	3	1444	Risk Ratio (M‐H, Fixed, 95% CI)	0.93 [0.71, 1.22]
4.2 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	1.69 [1.08, 2.65]
5 Vascular death, non‐fatal stroke or non‐fatal myocardial infarction	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
5.1 INR 2.0 ‐ 3.6	1	1068	Risk Ratio (M‐H, Fixed, 95% CI)	0.85 [0.65, 1.12]
5.2 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	1.70 [1.12, 2.59]
6 Recurrent ischaemic stroke	4	2760	Risk Ratio (M‐H, Fixed, 95% CI)	0.84 [0.61, 1.15]
6.1 INR 2.0 ‐ 3.6	3	1444	Risk Ratio (M‐H, Fixed, 95% CI)	0.80 [0.56, 1.14]
6.2 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	1.02 [0.49, 2.13]
7 Recurrent ischaemic stroke or intracranial haemorrhage	5		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
7.1 INR 2.0 ‐ 3.6	4	1561	Risk Ratio (M‐H, Fixed, 95% CI)	0.91 [0.66, 1.24]
7.2 INR 3. 0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	2.04 [1.21, 3.46]
8 Death or dependent at end of follow‐up	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
8.1 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	2.30 [1.37, 3.85]
9 Major bleeding complication	6		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
9.1 INR 1.4 ‐ 2.8	1	2206	Risk Ratio (M‐H, Fixed, 95% CI)	1.27 [0.79, 2.03]
9.2 INR 2.0 ‐ 3.6	4	1561	Risk Ratio (M‐H, Fixed, 95% CI)	1.93 [1.27, 2.94]
9.3 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	9.02 [3.91, 20.84]
10 Fatal intracranial or extracranial haemorrhage	5		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
10.1 INR 1.4 ‐ 2.8	1	2206	Risk Ratio (M‐H, Fixed, 95% CI)	1.4 [0.45, 4.40]
10.2 INR 2.0 ‐ 3.6	3	1444	Risk Ratio (M‐H, Fixed, 95% CI)	2.18 [0.83, 5.75]
10.3 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	17.37 [2.32, 130.11]
11 Intracranial haemorrhage, fatal or non‐fatal	4		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
11.1 INR 2.0 ‐ 3.6	3	1444	Risk Ratio (M‐H, Fixed, 95% CI)	1.82 [0.88, 3.78]
11.2 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	9.19 [2.80, 30.16]
12 Major extracranial haemorrhage	4		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
12.1 INR 2.0 ‐ 3.6	3	1444	Risk Ratio (M‐H, Fixed, 95% CI)	2.77 [1.55, 4.96]
12.2 INR 3.0 ‐ 4.5	1	1316	Risk Ratio (M‐H, Fixed, 95% CI)	8.85 [2.69, 29.11]

Methods	Randomised trial   Concealment: computer randomised   Blinding: open label, outcome assessment not described   Results available for all patients.   Intention‐to‐treat analysis
Participants	Spain   28 patients   Mean age 67 years   68% male   Patients with TIA or mild stroke in the MCA territory presenting within 90 days, who either have no history of cerebral ischaemia, a history of ischaemia exclusively in the region of the qualifying event, or a history of cerebral ischaemia in any region with indeterminate aetiology, who were angiographically diagnosed with 50% to 99% MCA stenosis within 7 days of the qualifying event, who have 50% or less stenosis of the ipsilateral extracranial or intracranial internal carotid artery, and who have a modified Rankin scale score of 3 or less   Conventional angiography or TCD + MRA/TCD + CTA in all patients   Time since stroke: mean 24.3 days   Comparability of groups: no major differences for prognostic variables
Interventions	Rx: coumarin (INR 2.0 to 3.0)   Monitoring: according to centre, monthly or shorter   Compliance: stopped in 3 patients, from monthly values 67% within target range   Control: aspirin 300 mg daily   Monitoring: with compliance form   Compliance: stopped in 0 patients
Outcomes	Vascular death, non‐fatal ischaemic stroke, non‐fatal myocardial infarction, major bleeding complication, all deaths
Notes	Exclusion criteria: other arterial cause of stroke, (contra) indication for study medication, difficult monitoring   Follow‐up: mean 23.1 months   Trial stopped early (recruitment too slow, aim was 300 patients)   No patients were lost
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer generated
Allocation concealment (selection bias)	Low risk	Adequate. However, unclear whether outcome assessment was blinded
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Open label
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Unclear

Methods	Randomised trial   Concealment by telephone, computer randomised   Blinding: outcome assessment only   Results available for 96.9%   Intention‐to‐treat analysis
Participants	Europe, Australia, Asia   1089 patients   Mean age 61 years   67% male   Cerebral ischaemia of non‐cardiac origin or transient monocular blindness   CT in all but 72 (93.3%)   Time since stroke: less than 6 months   Comparability of groups: no major differences for prognostic variables
Interventions	Rx: phenprocoumon or acenocoumarol or warfarin (INR 2.0 to 3.0)   Monitoring: on regular basis check of INR level   Compliance: specified in table, stopped in 198 patients (37%), 70% of time within target range   Control: aspirin 30 mg (57%), 75 mg (15%), 80 mg (6%), 100 mg (12%), 150 mg (3%), 200 mg (6%)   Monitoring: no monitoring   Compliance: stopped in 84 (15%)
Outcomes	Vacular death, non‐fatal stroke, non‐fatal myocardial infarction, major bleeding complication
Notes	Exclusion criteria: Rankin grade > 3, (contra) indication for study medication   Follow‐up: mean 4.6 years   Trial stopped early (after 55% of planned patients years) because of new standard therapy of aspirin plus dipyridamole   34 patients were lost, 7 had incomplete follow‐up
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer generated
Allocation concealment (selection bias)	Low risk	Adequate
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Open label study
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Blinded outcome assessment

Methods	Randomised trial   Concealment: unknown   Blinding: none   Results available for all patients randomised: unknown   Intention‐to‐treat not defined
Participants	Sweden   241 patients   Mean age 60.0 years   64% male   Carotid symptoms or homonymous hemianopia (104 TIA)   Time from stroke < 14 days   LP and CT or radionuclide scan in all   Comparability of groups: age and history similar
Interventions	Rx: warfarin (thrombotest 7% to 15%)   Monitoring: not defined   Compliance: 77.4% within values, stopped in 12 patients   Control: acetylsalicylic acid (1000 mg/day)   Monitoring: 5 hours after intake of acetylsalicylic acid   Compliance: 97%, stopped in 21 patients
Outcomes	Ischaemic stroke, TIA, vascular death
Notes	Exclusion criteria: atrial fibrillation, carotid endarterectomy   Follow‐up: mean 20.5 months in treated group versus 19.6 months in control group   No lost patients described
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Unclear
Allocation concealment (selection bias)	Unclear risk	Unclear
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Open label study
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Unclear

Methods	Randomised trial   Concealment by sealed envelopes (opaque and sequentially numbered)   Blinding: none   Results availability for all patients: not assessable   Intention‐to‐treat not assessable
Participants	Canada   178 patients   Mean age 68 years   58% male   Non‐cardiogenic TIA or mild stroke   CT not in all   Time since stroke: < 180 days   Comparability of groups not described
Interventions	Rx: warfarin (INR 2.0 to 3.0)   Monitoring and compliance not described   Control: acetylsalicylic acid (1300 mg)   Monitoring and compliance not described   Rx 2: warfarin (INR 2.0 to 3.0) + acetylsalicylic acid (80 mg)
Outcomes	Stroke, death, myocardial infarction, haemorrhage, TIA
Notes	Exclusion criteria: cardiac cause of stroke, carotid stenosis > 70%, (contra) indication for warfarin   Follow‐up: not specified
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	With envelopes
Allocation concealment (selection bias)	Low risk	Adequate
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Open label study
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Unclear

Methods	Randomised trial   Concealment by pre‐arranged assignment and labelled boxes   Blinding: double   Results available for 98.5% patients   Intention‐to‐treat analysis
Participants	USA   2206 patients   Mean age 63 years   59% male   Ischaemic stroke of non‐cardiac origin   Time since stroke: < 30 days   Comparibility of groups: no significant differences in any characteristic
Interventions	Rx: warfarin (INR 1.4 to 2.8)   Monitoring: first month 7 times, thereafter monthly INR measurements compliance: not described, 16.3% below range   Control: aspirin 325 mg daily   Monitoring: idem   Compliance: not described
Outcomes	Death from any cause or recurrent ischaemic stroke
Notes	Exclusion criteria: GOS < 3, TIA, contraindication for study medication   Follow‐up: 2 years   12 patients lost in treated group, 21 patients in control group   Outcomes measured at end of planned follow‐up
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	With labelled boxes
Allocation concealment (selection bias)	Low risk	Adequate
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Double‐blind study

Methods	Randomised trial   Concealment computer randomised   Blinding: double   Results available for 97.7%   Intention‐to‐treat analysis
Participants	North America   569 patients   Mean age 64 years   62% male   Patients with TIA or ischaemic stroke and an angiographically proven, 50% to 99% symptomatic stenosis of a major intracranial artery, Rankin ≤ 3   Time since stroke: < 90 days   Comparibility of groups: no significant differences in any characteristic
Interventions	Rx: warfarin (INR 2.0 to 3.0)   Monitoring: at least monthly compliance: stopped in 28.4%, 63% of time within target range   Control: aspirin 1300 mg daily   Monitoring: idem   Compliance: stopped in 16.4%
Outcomes	Ischaemic and haemorrhagic stroke and death from vascular causes other than stroke, major bleedings, myocardial infarction
Notes	Exclusion criteria: tandem stenosis extracranial carotid, cardiac embolus, limited survival, (contra) indication for study medication   Follow‐up: 1.8 years   8 patients lost in treated group, 5 in control group   Trial stopped earlier because of safety recommendations of the monitoring committee
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer generated
Allocation concealment (selection bias)	Low risk	Adequate
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Double‐blind study
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Double‐blind study

Study	Reason for exclusion
ARCH 2012	After consideration, the review authors agreed that this trial was not suitable for inclusion
Buren 1981	No concealed randomisation (depending whether they were born on odd or even dates)   The intensity of anticoagulation was not specified   No definition of outcomes
Eriksson 1985	Non‐randomised study (the month of admission determined treatment allocation)
Norrving 1983	The cause of stroke is not described (is atrial fibrillation included or not?)   The intensity of anticoagulation is not described and could not be obtained   The method of randomisation is not detailed
Olsson 1981	Treatment allocation according to date of birth, the intensity of anticoagulation is not defined by means of the INR and no maximum aimed anticoagulation level is defined   Atrial fibrillation is not specified as exclusion criterion
Ortiz Castellon 1992	No information on the method of randomisation, the duration of treatment and follow‐up, or number of outcome events   A request for this information was returned by mail as undeliverable

PERMALINK

Vitamin K antagonists versus antiplatelet therapy after transient ischaemic attack or minor ischaemic stroke of presumed arterial origin

Els LLM De Schryver

Ale Algra

L Jaap Kappelle

Jan van Gijn

Peter J Koudstaal

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Efficacy outcome measures

Safety outcome measures

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Results

Description of studies

Results of the search

Included studies

Excluded studies

Risk of bias in included studies

Effects of interventions

Vitamin K antagonists versus antiplatelet therapy (Analysis 1)

1.1. Analysis.

1.2. Analysis.

1.3. Analysis.

1.4. Analysis.

1.5. Analysis.

1.6. Analysis.

1.7. Analysis.

1.8. Analysis.

1.9. Analysis.

1.10. Analysis.

1.11. Analysis.

1.12. Analysis.

Vitamin K antagonists versus antiplatelet therapy (concealed randomised trials) (Analysis 2)

2.1. Analysis.

2.2. Analysis.

2.3. Analysis.

2.4. Analysis.

2.5. Analysis.

2.6. Analysis.

2.7. Analysis.

2.8. Analysis.

2.9. Analysis.

2.10. Analysis.

2.11. Analysis.

2.12. Analysis.

Vitamin K antagonists versus aspirin only (Analysis 3)

3.1. Analysis.