An authoritative meta-analysis [1] looks at the efficacy of these treatments. Although our prejudice is that they are effective, having this confirmed, and being able to quantify benefits and risks, is comforting.
Search
A comprehensive strategy looked for randomised trials for nonvalvular atrial fibrillation. Trials looking at prosthetic valves or mitral stenosis were not considered. Blind and non-blinded trials were included, as were studies of primary and secondary prevention.
Outcomes
Because ischaemic and haemorrhagic strokes were not reliably distinguished in many of the trials, 'all strokes' was used as the main outcome. The number of intracranial and extracranial bleeds, and all-cause mortality data was also collected.
Table 1: Results of randomised trials of antithrombotic interventions to prevent
all-cause stroke in atrial fibrillation
| Strokes on: | ||||||||
| Warfarin | Aspirin | Placebo | ||||||
| Number | Percent per year | Number | Percent per year | Number | Percent per year | Duration (years) | Relative risk (95%CI) | NNT (95%CI) |
| 53/1450 | 2.3 | 133/1450 | 5.8 | 1.6 | 0.4 (0.3 to 0.6) | 18 (14 to 27) | ||
| 159/1624 | 6.5 | 190/1601 | 7.9 | 1.5 | 0.8 (0.7 to 0.98) | 48 (23 to >1000) | ||
| 82/1416 | 2.6 | 123/1421 | 4 | 2.2 | 0.7 (0.5 to 0.9) | 35 (21 to 104) | ||
| Modelling events in 1000 patients with non-valvular atrial fibrillation: benefits and harms with antithrombotic treatment | |||||
| All strokes | Major extracranial bleed | ||||
| Primary prevention risk: | Secondary prevention | ||||
| Low | Moderate | High | |||
| Events per year with: | |||||
| No therapy | 10 | 35 | 60 | 120 | 6 |
| Adjusted-dose warfarin | 4 | 14 | 24 | 48 | 9 |
| Aspirin (low dose) | 8 | 28 | 48 | 96 | 7 |
| Illustration of the magnitude of treatment in one year effects from clinical trials, assuming different initial risk rates, and based on 1000 patients who have not had a stroke and with low moderate and high risk of stroke, and for 1000 patients who have had a stroke. Shaded area indicates harm from major extracranial bleeds. | |||||
Aspirin versus placeboThere were six trials with 3225 patients and 349 strokes. The mean duration was 1.5 years, the daily aspirin dose was 325 mg/day or below in all but 21 patients, the mean age was 70 years and 40% had a previous stroke or transient ischaemic attack. The rate of stroke on placebo was 5.2% per year for those with no previous stroke, and 13% a year for those patients who had had a previous stroke. BenefitThere was an overall reduction in strokes of 20%, from a rate of 7.9% per year with placebo to 6.5% a year with aspirin. The NNT was 48 (23 to >1000). This means that 48 patients have to be treated with aspirin for 1.5 years to prevent one stroke that would have occurred without treatment. The review [1] quotes one-year NNTs to prevent one stroke of 67 for primary prevention and 40 for secondary prevention for aspirin compared with placebo. All-cause mortality was not significantly reduced by aspirin. HarmThere were four intracranial haemorrhages in patients receiving aspirin (annual rate 0.2%) and three in those with placebo (annual rate 0.1%). The annual rate of major extracranial haemorrhage was 0.5% for aspirin and 0.6% for placebo. Adjusted-dose warfarin versus aspirinThere were six trials with 2837 patients and 205 strokes. The mean duration was 2.2 years, the mean INR achieved was 2.2 to 3.1, the mean age was 71 years and 21% had a previous stroke or transient ischaemic attack. The rate of stroke on aspirin was 2.7% per year for those with no previous stroke, and 11% a year for those patients who had had a previous stroke. BenefitThere was an overall reduction in strokes of 35%, from a rate of 4.0% per year with aspirin to 2.6% a year with adjusted-dose warfarin. The NNT was 35 (21 to 104). This means that 35 patients have to be treated with adjusted-dose warfarin for 2.2 years to prevent one stroke that would have occurred with aspirin. The paper quotes one-year NNTs to prevent one stroke of 167 for primary prevention and 14 for secondary prevention for warfarin compared to aspirin, but with some data excluded for this calculation. All-cause mortality was similar for both treatments. HarmThere were 17 intracranial haemorrhages in patients receiving warfarin (annual rate 0.5%) and seven in those with aspirin (annual rate 0.2%). The annual rate of major extracranial haemorrhage was 0.2% for higher for warfarin compared with aspirin, a relative risk of 2.0 (1.2 to 3.4). CommentThis is a review written by people involved in some of the main trials in this area, and with a view to examining benefit and harm. The reviews will also be available separately on the Cochrane Library in early 2000. The authors modelled the effects of treatment - the benefit in terms of reductions in all-cause strokes, and the harm in terms of major extracranial bleeds. The effects based on cohorts of 1000 patients over one year, at various baseline risk of stroke and for primary and secondary prevention are shown in Table 2 . Benefit outweighs harm. Given that atrial fibrillation is often aggressively treated in primary care, and is relatively common (three million warfarin prescriptions in primary care in England in 1998), this conclusion is comforting. The review did not deal with people with structural heart disease, in whom the risk of stroke may be higher. It is a worthwhile read. Reference:
|