Aspirin for everyone older than 50?: AGAINSTBMJ 2005; 330 doi: https://doi.org/10.1136/bmj.330.7505.1442 (Published 16 June 2005) Cite this as: BMJ 2005;330:1442
- Colin Baigent (), reader in clinical epidemiology1
- Accepted 11 May 2005
Current population screening for vascular disease is neither efficient nor effective. Peter Elwood and colleagues believe we should have a public information strategy highlighting the benefits (and risks) of aspirin for older people, but Colin Baigent argues that the evidence of benefit is not yet strong enough
An age threshold approach to aspirin prophylaxis in people without known vascular disease has two important problems. The balance of benefits and risks of aspirin in people aged 70 or over has not been clearly defined in randomised trials, and the benefits do not clearly exceed the risks in younger people without vascular disease. Consequently, it would be unwise to adopt such a policy, whatever age threshold is chosen, until we are sure that older patients will derive net benefit from it.
What studies show
Among high risk patients with known occlusive arterial disease and a greater than 3% annual risk of a vascular event (defined as non-fatal myocardial infarction, non-fatal stroke, or vascular death), the benefits of aspirin substantially outweigh the risks of bleeding. A recent meta-analysis of randomised trials of antiplatelet drugs versus control showed that for every 1000 such patients treated for a year, aspirin would be expected to prevent about 10-20 vascular events and cause one or two major gastrointestinal bleeds.1 In a wide range of high risk patients, aspirin reduces the risk of a vascular event by around one quarter, with reductions of one third in non-fatal myocardial infarction, one quarter in non-fatal stroke, and one sixth in vascular death. In a patient with occlusive arterial disease, therefore, the expected benefit can be quantified as about a quarter of that person's baseline risk of a vascular event. For anyone with a definite history of arterial disease, the expected benefits far exceed the hazards, and so all should be considered for aspirin (or, if appropriate, another antiplatelet drug) unless there is a clear contraindication.1
Unfortunately, predicting the benefits and hazards of aspirin in someone without known arterial disease is far less straightforward. To date, six primary prevention trials comparing aspirin versus control have reported their findings.2–7 A meta-analysis of the first five of these trials indicates that aspirin reduces myocardial infarction by about one third, but (in contrast to patients with known vascular disease) has little or no effect on stroke or death from vascular causes.8 More recently, data from the women's health study suggested that aspirin may protect against stroke in healthy women, but myocardial infarction was not significantly reduced.7 Although the 95% confidence intervals for the effect on myocardial infarction in this study were wide, and compatible with a modest benefit, these findings add to uncertainty about the net effects of aspirin in women.
Aside from this uncertainty, we can perhaps be moderately confident of preventing myocardial infarction in people without vascular disease, but it would be prudent to assume that aspirin would not necessarily reduce the risk of stroke or vascular death. (Nor indeed, is there reliable evidence from randomised trials that aspirin can prevent cancer or dementia, so these potential benefits should not be assumed.)
Risk of vascular events and bleeding
Estimating the current annual risks of myocardial infarction in the UK is complicated by the fall in event rates over the past two decades, when most cohort studies were conducted. Consequently, the rates calculated from the Caerphilly study may have overestimated current UK annual risks of a vascular event. However, based on Elwood and colleagues' calculations from Caerphilly data,9 and assuming that a myocardial infarction is the first vascular event in around half of individuals, the annual risk of myocardial infarction for the age group 55-59 is 7 per 1000 population (from their table 2: 7.0% over 5 years, divided by 5, multiplied by one half). If aspirin reduces this risk by one third, then the absolute benefit (which may be an overestimate), is around two first myocardial infarctions avoided per 1000 population each year. How does this compare with the expected risks of aspirin? The first thing to note is that the absolute excess risk of haemorrhagic stroke attributable to aspirin is small (around 0.1/1000 a year). However, the risks of major gastrointestinal bleeding (that is, bleeds that are fatal or require transfusion) also need considering. To assess these risks for unselected individuals we have to turn to observational studies because people with risk factors for bleeding have been excluded from the primary prevention trials. A review of observational studies suggests that the background risk of major gastrointestinal complications rises steeply from 1-2/1000 a year at age 60 to around 7/1000 above age 80,10 and since aspirin roughly doubles the risk of gastrointestinal bleeding, the attributable excess risks at each age are likewise 1-2/1000 a year at age 60, rising to 7/1000 a year at age 80. Among unselected people younger than 60, therefore, the expected benefit on myocardial infarction (2/1000/year avoided) does not clearly exceed the expected risk of a major gastrointestinal bleed (1-2/1000/year).
Given this lack of a clear benefit at age 60, we might consider raising the age threshold to 65 or 70. But, whereas the benefits of aspirin on myocardial infarction are reasonably well defined in middle age, little is known of its effects in older people. In particular, primary prevention trials have included relatively small numbers of people over 70, for whom the potential benefit may be largest. Given that the observational studies strongly suggest that the risks of bleeding might increase substantially in older people (as does fatality from a bleed), we need decisive evidence of benefit in this age group before exposing large numbers of healthy people to potential harm. In my view, therefore, we should not contemplate an age threshold approach to primary prevention with aspirin until we have much better evidence of its benefits in older people. We therefore need further randomised trials comparing low dose aspirin with placebo, such as the aspirin in reducing events in the elderly (ASPREE) study,11 which aims to randomise 15 000 people aged 70 or over to aspirin 100 mg daily versus placebo. A recommendation that aspirin be used for primary prevention of vascular disease in unselected people over a certain age could result in net harm, and we must have very good evidence to the contrary before instituting such a policy.—Colin Baigent
Contributors and sources CB coordinates the Antithrombotic Trialists' Collaboration, which is currently reviewing the evidence from the primary prevention trials of aspirin versus placebo.
Conflict of interest None declared.