Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

The trouble with aspirin

Systematic review
Results
Comment

Bandolier is always interested to revisit a topic when some new evidence, new analysis, or new thoughts make it relevant. Low dose aspirin (LDA) is an important topic, and worth revisiting for a new look at the data.

We know that it does good in people at high cardiovascular risk. We also know that it does some harm in a variety of ways. A new meta-analysis [1] provides a better insight into some of the harm.

Systematic review


This followed a fairly standard path of searching, and was able to draw on many previous meta-analyses in this therapeutic area. Studies for inclusion were those comparing aspirin with placebo for primary or secondary prevention, or prophylaxis of deep venous thrombosis. Aspirin had to be low dose (75 to 325 mg daily).

Studies had to provide information on bleeding events, non-cardiovascular deaths, or discontinuations or symptoms for other than bleeding or cardiovascular events. They had to be randomised, have a duration of two months or longer, and have 100 patients or more in each treatment arm.

A series of outcomes were extracted from the trials, but the outcomes of primary interest were any major bleeding, major gastrointestinal bleeding, and intracranial bleeding. When not otherwise described as major, those needing transfusion were so defined.

Results


The basis of the analysis was 14 randomised trials with 57,000 participants, about 53,000 of whom were in studies lasting 12 months or longer. Annualised event rates for the three primary outcomes, with calculated numbers needed to harm for LDA compared with placebo, are shown in Table 1. Those taking LDA have an additional risk of any major bleed or major gastrointestinal bleed of about one person in 800 every year.


Table 1: Meta-analysis of bleeding events in about 57,000 patients taking low dose aspirin, showing the absolute annual event rates with low dose aspirin (LDA) and placebo, and relative risk and number needed to harm



Annual event rate (%) with
Bleeding event
LDA
Placebo
Relative risk
(95% CI)
NNH
(95% CI)
Any major bleeding
0.31
0.18
1.7 (1.4 to 2.1)
769 (500 to 1200)
Major gastrointestinal bleeding
0.24
0.12
2.1 (1.6 to 2.7)
833 (530 to 1400)
Intracranial bleeding
0.08
0.05
1.7 (1.1 to 2.4)
3300 (1250 to 10,000)



Comment


In longer-term trials in people at high risk of cardiovascular problems (previous heart attack, stroke, or other high risk causes), there are clear benefits from using LDA in reducing fatal or nonfatal heart attacks or strokes, or vascular deaths. Table 2 shows the benefits for LDA and placebo in high risk patients in an annualised form calculated from the Antithrombotic Trialists' Collaboration (Bandolier 108), alongside the annual risks of all major bleeding events.


Table 2: Some calculations on the annual benefits and harms with placebo and low dose aspirin (LDA), and the risk difference to demonstrate additional annual risk



Annual rate with placebo
Annual rate with LDA
Annual rate difference
Event
Percent
Risk
Percent
Risk
Percent
Risk
Fatal or non-fatal heart attack or stroke, or vascular death
6.9
1 in 14
5.6
1 in 18
1.4
1 in 71
will benefit
Fatal or non-fatal major bleeding event
0.18
1 in 560
0.31
1 in 320
0.13
1 in 770
harmed



Benefits outweigh the risks, though there are probably other risks, so this will overstate the benefit:risk balance. It is possible to present the information in a number of ways, both as a percentage rate, or as a risk or odds, and for the actual rates or the difference.

In people who do not have high levels of cardiovascular risk, the benefits will fall, but the potential for gastrointestinal bleeding almost certainly remains the same. And yet our newspapers, and the tone of the media in general, is that taking a small amount of aspirin every day is beneficial for everyone. Not stated, but implied, is that it harms no one. It might be a useful example to use when explaining that all drugs are also poisons, and that safety is relative. For high risk patients the balance is easy to remember: good outcome 1 in 70, bad outcome 1 in 770.

Reference:


1 KR McQuaid, L Laine. Systematic review and meta-analysis of adverse events of low-dose aspirin and clopidogrel in randomized controlled trials. American Journal of Medicine 2006 119: 624-638.