Skip navigation

Clinical effectiveness of anticoagulation for stroke prevention in AF


Clinical bottom line

Stroke and haemorrage rates in patients on warfarin for primary prevention of stroke were similar in a clinical population to those found in randomised trials.


L Kalra et al. Prospective cohort study to determine if trial efficacy of anticoagulation for stroke prevention in atrial fibrillation translates into clinical effectiveness. BMJ 2000 320:1236-1239.


From general medical clinics in a general hospital were recrited patients with chronic non-valcular atrial fibrillation and less than 90 years old, but not already on warfarin or with a previous stroke. Diagnosis of atrial fibrillation was by ECG and echocardiography. Those without contarindications and with a high risk of stroke were eligible for anticoagulation with warfarin. The target range was an INR of 2.0 to 3.0.

The patients were cared for by their usual physicians. They were followed up for two years and reviewed every six months by a physician unaware of their INR. they were assessed for episodes of bleeding, other illness, neurological status, or changes in drugs that might alter anticoagulation control. Records were consulted for admissions or events not recalled by the patients. patients not attending were followed up by the research team, and there was consequently a complete follow up for all patients.


There were 2457 patients screened, with chronic atrial fibrillation in 344. Of these 286 had a high risk of stroke. Of these 248 had no major contraindication to anticoagulation. Of these 76 were already on warfarin and five refused anticoagulation. Thus 167 patients entered the study.

Their mean age was 77 years, and 58% were women. That meant that the study patients were on average seven years older than those in the RCTs. They were more likely to have had a previous stroke and less likely to have ischaemic heart disease.

During the two years of follow up, INR was ithin target range for 61% of the time, below it for 26% of the time, and above it for 13% of the time. Thirty patients (18%) stopped warfarin. These figures were similar to those from a pooled analysis of RCTs, though the proportion of the time within the target range of 2.0 to 3,.0 was higher at 66% in the RCTs.

The number and annual rate of events in clinical practice and annual rates from the pooled analysis of RCTs is shown in Table 1. There was no statistical difference between the rate of events occurring in RCTs of warfarin treatment and the rate of events occurring in clinical practice.

Table 1: Clinical practice and RCTs compared


Number of events in clinical practice

Annual rate in clinical practice

Annual rate from pooled RCTs


Percent (95% CI)

Ischaemic stroke


2.0 (0.7 to 4.4)

1.4 (0.8 to 2.3)

Transient ischaemic attack


1.0 (0.1 to 4.0)

0.7 (0.3 to 1.4)

Death from non-cerebral causes


2.4 (0.6 to 6.1)

3.6 (2.7 to 4.9)

Intracranial haemorrhage


0.3 (0.03 to 2.7)

0.3 (0.06 to 0.7)

major bleeding


1.4 (0.2 to 4.6)

1.3 (0.7 to 2.1)

Minor bleeding


5.4 (2.4 to 10.2)

9.2 (3.7 to 12)


Derspite the study population in this prospective cohort study being older and spensing less time in the target INR range, the results onbtained in clinical practice were similar to those from RCTs. It is always helpful to know that good evidence from quality randomised trials translates to the real world.