The prevention of systemic embolism and stroke in patients with nonrheumatic atrial fibrillation (AF) is a difficult problem. However, the results of randomised controlled trials addressing this issue are now available. In all the placebocontrolled trials, warfarin decreased stroke risk. Findings with aspirin were conflicting.

This consistent benefit of warfarin will allow the treatment to be confidently adopted by clinicians. However, a caveat is that patients in clinical trials represent 'ideal cases', particularly when there is the potential for complications such as bleeding. In the trials described, 58-93% of patients were not included for reasons which included increased bleeding risk and concerns relating to compliance.

The relative benefits of warfarin and aspirin may depend upon the mechanisms underlying thrombosis. Thrombosis associated with stasis in an enlarged left atrium or hypocontractile left ventricle probably mainly relates to fibrin formation. When thrombosis occurs on an ulcerated atheromatous plaque, both fibrin formation and platelet activation occur. In the latter case, aspirin may be expected to be beneficial. The mechanisms of thrombosis may explain the absence of significant benefit with aspirin compared with placebo in the AFASAK study (Reference 7 in the accompanying article), whereas aspirin was protective in the first SPAF study (Reference 9). Patients in the AFASAK study had a mean age of 74.2 years (mean age in the first SPAF study was 67.1 years) and about twice the incidence of previous myocardial infarction and congestive heart failure as those in the first SPAF study.

In the second SPAF study, SPAF II (Reference 14), there were no significant differences in cerebral events in patients on warfarin (with a relatively high target INR) or aspirin, when analysis was according to the allocated treatment (intention to treat analysis). Patients taking warfarin had fewer emboli, but more haemorrhagic strokes. However, a significant number of patients randomised to aspirin in the SPAF II study commenced warfarin during follow up. When the study was analysed according to treatment at the time of cerebral events (on treatment analysis), which is perhaps more analogous to everyday practice, warfarin approximately halved cerebral events.

In the first SPAF study, a high risk was indicated by a history of previous embolism or hypertension and recent congestive heart failure. The European secondary prevention study (Reference 12) confirmed that the benefits of warfarin were even greater in patients who had already had a cerebral ischaemic event. It is important to notice that in SPAF II, in patients <75 years with none of the 3 clinical risk factors above, the rate of cerebral ischaemic events in patients taking aspirin was only 0.5% per year. This is less than the usual accepted risk of bleeding with warfarin given in conventional doses (1% per year for intracranial haemorrhage and 2.5% per year for other major haemorrhage). It seems very appropriate at this time to recommend that warfarin is unnecessary in younger patients (perhaps <70-75 years) with no clinical risk factors.

In the first SPAF study, the echocardiographic markers of high risk, left atrial enlargement and left ventricular dysfunction, were even more important than the clinical risk factors. In practice, most patients should be referred for echocardiography and a specialist opinion, because another clinical decision to be made is whether or not it is appropriate to attempt cardioversion to sinus rhythm. The role of transoesophageal echocardiography still needs further clarification. Most patients with nonrheumatic AF are elderly, and they frequently have a number of potential causes for cerebral events. Conditions such as left atrial thrombus and spontaneous echo contrast, left ventricular thrombus and aortic atheroma are much better seen on transoesophageal examination. This should not be routine, but may be suggested in selected patients if the indication for warfarin is doubtful.

In the trials, the benefits of warfarin were maintained even when the target INR was relatively low. This is important as it minimises the risk of bleeding. A personal recommendation is that when warfarin is given to patients with nonrheumatic AF, the target INR should be 1.5-2.5.

Recent relatively simple guidelines for commencing warfarin are shown below. Table 1 shows the use of warfarin in Australia.

Day of therapy INR Level Warfarin dose
1 - 5 mg
2 - 5 mg
3 >5.0 omit
3.5-5.0 2.5 mg
3.0-3.5 4 mg
2.0-3.0 5 mg
1.5-2.0 6 mg

Table 1
Warfarin ­ estimated total community
(nonhospital) use


Defined daily dose/thousand population/day
1990 506,662 1.418
1991 510,494 1.401
1992 639,859 1.724
1993 790,898 2.064
1994 921,413 2.360

The defined daily dose for warfarin from the 1995 WHO ATC Classification Index is 7.5 mg.

Note: Figures supplied by the Drug Utilization Sub Committee of the Pharmaceutical Benefits Advisory Committee.

In summary:
1. The risk of systemic embolism and stroke in patients with nonrheumatic AF should not be underestimated. The embolic rate is probably up to 5% per year in those with left atrial enlargement and/or left ventricular dysfunction, and up to 20% per year in those who have already had systemic embolism.

2. Management decisions must be individualised. Clinical markers of increased risk include previous embolism, recent congestive heart failure, hypertension and diabetes mellitus. The presence of risk factors, particularly structural heart disease, should lead to consideration of warfarin, even if nonrheumatic AF is only a marker of associated abnormalities. If there are no risk factors in a younger patient, aspirin is appropriate.

3. Transthoracic echocardiography improves risk stratification.

4. When warfarin is given, the target INR should usually be relatively low, in the range 1.5-2.5.

Andrew M. Tonkin

Director of Cardiology, Austin and Repatriation Medical Centre (Austin Campus), Melbourne