select search filters
roundups & rapid reactions
factsheets & briefing notes
before the headlines
Fiona fox's blog

aspirin use and cardiovascular/bleeding events

A review and meta-analysis published in JAMA demonstrates that the use of aspirin in individuals without cardiovascular disease was associated with a lower risk of cardiovascular events and an increased risk of major bleeding.

Round Up accompanied this Before the Headlines.

Title, Date of Publication & Journal

‘Association of aspirin use for primary prevention with cardiovascular events and bleeding events: A systematic review and meta-analysis’ by Sean L. Zheng et al. was published in JAMA at 16:00 UK time on Tuesday 22 January 2019.

Study’s main claims – and are they supported by the data

This paper is a systematic review and meta-analysis of aspirin for primary prevention of cardiovascular events.

The study’s primary claim is that aspirin used for primary prevention of cardiovascular events leads to a decrease in the risk of cardiovascular events (a positive outcome), but also a comparable increase in the risk of major bleeding (a negative outcome). It is a well conducted study, and the data and analyses strongly support this claim. 

The press release from King’s College London makes some statements that could be misinterpreted:

In referring to the results, the press release presents the relative risk for patients given aspirin vs placebo/no aspirin and does not give the absolute risks given in the study. Without the absolute risk the relative risk could be interpreted as meaning the impact on bleeding is stronger than the impact on cardiovascular events (when the absolute risk differences are actually comparable). The reason that the relative risks are so different, whilst the absolute risks are similar, is that the risk of experiencing a cardiovascular event is generally higher than the risk of experiencing a major bleed.

The press release title, “Frequent use of aspirin can lead to increased bleeding”, is not incorrect but does not include the equally important finding that it can lead to reduced risk of cardiovascular events. The word “frequent” might be interpreted as meaning that individuals who sometimes take aspirin for other purposes (e.g. to relieve pain) may be at increased risk of major bleeds, whereas the trials included in the study are looking specifically at those given aspirin for the primary prevention of cardiovascular events.

The purpose is quite specific and the regularity of dosing is also specific (included trials that have given patients aspirin daily over a period of several years), which may be important in terms of the risk of cardiovascular events or bleeding.

The fourth paragraph of the press release states that the study looked at outcomes of more than 1000 participants. This could be taken as meaning this study included participants directly, when in fact the study included previously-published trials and performed a pooled analysis of the individual trial results. The characteristics that are referred to in the fourth paragraph of the press release are characteristics determined by the authors that decided whether or not trials would be included in their pooled analysis.



The study is well conducted and describes clearly the steps taken in the data collection and analysis, making it easily reproducible. 

Results are provided both in terms of relative risk and absolute risk (as well as number needed to treat/harm), preventing claims from being overstated.

The authors provide a protocol that clearly states the outcomes they are interested in, written in advance of performing the study. This limits the potential for selectively reporting ‘desirable’ results.

The authors performed a range of sensitivity analyses to check how robust their conclusions were.

The study excludes studies with less than 1000 participants. Smaller studies are typically likely to be at higher risk of bias.

The study accounts for a number of factors that might be expected to affect the risk of cardiovascular events (e.g. small-study bias, changes in International Disease Classification of Disease codes).

Results are generally very consistent across the different included trials for all outcomes.


The authors clearly state several limitations of the study, and how sensitivity analyses were used to assess their impact on the conclusions:

– availability and quantity of reported data

– changes in definitions of outcomes between studies

– range of different aspirin doses used.

However, there are a couple of limitations that were not addressed. Firstly, it only included studies published in the English language. This could limit the generalisability of results to other populations.

Secondly, it seems as though the authors have assumed that the risk of different outcomes would remain the same over a 10-year period. In reality this is unlikely to be the case, as risk is likely to increase over time (since the risk of cardiovascular events and bleeding will increase as individuals get older). Therefore, the absolute risk reductions/increases calculated are likely to be slight underestimates of what we might expect the ‘true’ absolute risk reduction/increase to be. However, this will be the case for both risk of cardiovascular events and risk of major bleeds, so even though the risk of both will be higher, they are still likely to be reasonably similar.


Number needed to treat/harm: The number of patients that would need to be given aspirin for one of them to experience a single outcome. The treat/harm depends on whether the outcome is positive or negative.

Absolute risk reduction/increase: The difference in the risk of an event between the group of patients given aspirin and the group of patients given no aspirin/placebo.

Relative risk reduction/increase: The relative change in risk of an event between the group of patients given aspirin and the group of patients given no aspirin/placebo.

Cardiovascular event: A negative clinical outcome involving the blood vessels (including stroke) or the heart. Can be fatal or non-fatal.

Any specific expertise relevant to studied paper (beyond statistical)?


in this section

filter Headlines by year

search by tag