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

expert reaction to study looking at NSAID painkillers (non-steroidal anti-inflammatory drugs) and heart attack risk

A new observational study published in the BMJ investigates the association between painkillers and myocardial infarction risk.


Dr Mike Knapton, Associate Medical Director at the British Heart Foundation, said:

“The authors of this study use data from Canadian and European individual patient databases and concerned themselves with 5 drugs – celecoxib, the three main traditional NSAIDs (diclofenac, ibuprofen, and naproxen), and rofecoxib.  This involved 450 000 individuals, of which 61 000 had a heart attack – it is the largest investigation of its type.  They found an increased risk of acute myocardial infarction with taking NSAIDs.

“We already know this from previous research – i.e. an association between taking NSAIDs and small increase in risk of thrombotic events, such as stroke and heart attack.

“What is new here is that this study shows that (a) the risk of myocardial infarction associated with NSAID use increases immediately with exposure (less than 7 day duration of treatment); (b) there was a relationship between increasing NSAID daily dose and risk of acute myocardial infarction; and (c) a longer duration of treatment generally does not seem to be associated with greater probability of increased risk of myocardial infarction.

“Implications appear to be that, as we already know, there is an increased risk of cardiac events with the use of NSAIDs – this study reminds and emphasises that patients and doctors need to weigh up the risks and benefits of taking NSAIDs, particularly if the dose is high.  Where possible alternative medication should be used – such as paracetamol – or treatments that do not require medication such as exercise and physiotherapy, where appropriate.  Patients buying over the counter painkillers such as pills containing ibuprofen, should be made aware of the risk in the in pack information leaflet.  I would be particularly cautious in patients who have had an MI or are at high risk.

“A particularly difficult decision will be in patients with inflammatory arthritis – such as rheumatoid arthritis – as they often need NSAIDs to damp down the inflammation to control pain and stiffness in the joints.  But patients with rheumatoid arthritis are at increased risk of CVD as well.  So, doctors advising these patients need to consider both the treatment of the arthritis and CVD risk management at the same time.”


Prof. Stephen Evans, Professor of Pharmacoepidemiology, London School of Hygiene & Tropical Medicine, said:

Is this good quality research?

“This is good quality observational research, though there are limitations.

Are the conclusions backed up by solid data?

“The data are from a few observational studies that do not comprehensively describe the patients, so this is one limitation.

What are the absolute risks?

“This is a key point missing from the paper.  The paper has good evidence that there is some risk of a heart attack for all NSAIDs and suggests that the risk starts immediately on starting them, but is only expressed in relative terms.  There is no clear description of the absolute risk.

How does this work fit with the existing evidence?

“It confirms what has been seen in randomised trials.  The effect of naproxen at high doses has been shown in one trial to be similar to these findings but not in others.

Have the authors accounted for confounders?  Are there important limitations to be aware of?

“A number of lifestyle factors (smoking, body mass index) are not available.  They argue this makes little difference, but it leads to uncertainty.

What are the implications in the real world?

“For people at a high risk of a heart attack, the use of NSAIDs should be minimised.

What does this mean for one-off use of these painkillers for a day or two – should people be worried?

“This study suggests that even a few days’ use is associated with an increased risk, but it may not be as clear as the authors suggest.  The two main issues here are that the risks are relatively small, and for most people who are not at high risk of a heart attack, these findings have minimal implications.

“The authors’ finding that the risk begins immediately is also compatible with the possibility that those who use NSAIDs are, on average, at higher risk than non-users at the start for some other reason apart from the drugs and the study has inadequately adjusted for such differences.  The most likely mechanisms for action of the drugs would be expected to show a low risk at the start and only have an effect on heart attacks after longer usage.  That this wasn’t the case casts some doubt on the findings of an immediate increase in risk.

“All effective medicines have unwanted effects, and NSAIDs, although easily available, are not without some risks, but this study is no reason to induce anxiety in most users of these drugs.”


Dr Amitava Banerjee, Senior Clinical Lecturer in Clinical Data Science and Honorary Consultant Cardiologist, UCL, said:

“In a high-quality meta-analysis, which combines individual level data from nearly 450 000 people, non-steroidal anti-inflammatory drugs (NSAIDs) were associated with an increased risk of heart attack.  Previous studies, including trials, have shown an increased risk of heart attack with particular NSAIDs, such as rofecoxib, a so-called COX-2 inhibitor.  This research suggests that short-term use (8-30 days) at a high daily dose of any NSAID (including diclofenac, ibuprofen and naproxen) is associated with the greatest risk of heart attack, with no further increase beyond the initial period.

“This is the largest study of its kind, but it is still observational data based on prescription or dispensing information, rather than whether people were actually taking their medication.  Although these data reflect real world use of NSAIDs, it is impossible to control for all the factors which may lead to confounding or bias.  It is important to note that not all the studies which could be included were available for analysis, so it is possible that there is a bias in the selection of studies.

“The increased risk of heart attack with NSAIDs, regardless of which one, means that both health professionals and the public should weigh up the harm and the benefit when prescribing these medications, especially for more than a day or two.  Despite the over-the-counter availability of the traditional NSAIDs, this caution is still required.  The mechanism of this increased risk of heart attack is not at all clear from existing studies.”


Prof. John Martin, Professor of Cardiovascular Medicine, UCL, said:

“This is an important study, looking at a large group of patients taking non-steroidal anti-inflammatory drugs (NSAIDs).  It adds to our understanding of possible effects of these drugs on the heart and blood vessels.  However, as this study is limited by the fact that it is an observational study, I see this study as a basis for designing larger randomised stratified clinical studies that will have the ability to examine what sort of patients may be vulnerable to the effects of NSAIDs, and to try and dissect out the possible causes involved.  At the moment doctors prescribing these drugs have to balance the risk: benefit ratio as many patients get important pain relief from these drugs.”


Prof. Jane Mitchell, Head of Cardiothoracic Pharmacology, Imperial College London, said:

“This is an observational study so it cannot say whether these painkillers actually cause heart attacks, but it does give more information about an association we knew about from previous studies.  We do not know what the potential underlying mechanism could be, but other research has suggested it might involve these drugs blocking a hormone in the body called prostacyclin which protects our blood vessels, protects the kidney and thins the blood.  But we need more research to know whether this is indeed the case – finding the mechanisms is vital so that we can devise tests that can identify those people at greatest risk of heart attacks so that the majority patients can take their medication with relative confidence.

“These painkillers include some of the most commonly taken drugs worldwide and although the increased risk of heart attack might be low, because of the scale of their use it is seen as an important problem.  Patients, doctors and drug companies are worried – concern over an increased risk of heart attacks associated with drugs like celecoxib or ibuprofen has slowed drug development in this area to a virtual stand-still.  We should also remember that there is some evidence some of these drugs may help prevent some cancers1 but they are not used because of concerns over side effects.”

  1. Arber N, Eagle CJ, Spicak J, Racz I, Dite P, Hajer J, Zavoral M, Lechuga MJ, Gerletti P, Tang J, Rosenstein RB, Macdonald K, Bhadra P, Fowler R, Wittes J, Zauber AG, Solomon SD, Levin B and Pre SAPTI. Celecoxib for the prevention of colorectal adenomatous polyps. New England Journal of Medicine. 2006;355:885-95.


Prof. Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“This is an interesting study, and I think it does throw some light on possible relationships between NSAID painkillers and heart attacks. This new study has helped persuade me that there is probably a real association between taking these painkillers and heart attacks. But, despite the large numbers of patients involved, some aspects do still remain pretty unclear. It remains possible that the painkillers aren’t actually the cause of the extra heart attacks, and there’s still a lot of uncertainty about the strength of the relationships and about how they vary with dose and with the timing of taking the painkillers. And we’ve got to remember that all drugs have side effects, and that people aren’t prescribed these painkillers for fun, but to deal with a real pain problem.

“The press release carefully, and correctly, points out that this is an observational study, and that therefore conclusions cannot be made about cause and effect. That’s because the researchers couldn’t take into account all the factors that might influence both the prescribing of painkillers of this class and the chance of a heart attack, usually because they did not have adequate data on them. The research report mentions several such factors – smoking, obesity, income, education, and taking painkillers bought over the counter (and so not recorded in the data). So, for instance, perhaps people who smoke are more likely to be prescribed painkillers, and are separately more likely to have a heart attack. Or, putting it crudely, if someone is prescribed a high dose of a painkiller because of severe pain, and then has a heart attack in the following week, it’s pretty hard to tell whether the heart attack was caused by the painkiller or by whatever was the reason for prescribing the pain killer (or indeed by something else entirely). The researchers did allow statistically for other medical conditions where they had data, but the data were far from complete. Other researchers have argued that, overall, this so-called confounding would make the relationship between painkillers and heart attacks look weaker rather than stronger than it really is – but still one has to be cautious about concluding that the painkillers might be causing extra heart attacks.

“A difficulty in making sense of the results of the study is that they are almost all about odds ratios, that is, about comparing the chances of a heart attack in people who took particular doses of NSAID painkillers, with the chances in people who hadn’t taken these painkillers. That sort of information is difficult to interpret on its own. If the risk is increased by, say, 30%, that might not be important if the original risk was very low. 30% more than next to nothing is still next to nothing. As far as I can see, the research paper itself gives no information on the absolute risks of heart attacks. (The Abstract says that one of the outcome measures was the (posterior) probability of a heart attack, but I can’t find those results either in the research paper or in the very extensive supplementary results that have been made available.) The press release says that “the risk of heart attack due to NSAIDs is on average about 1% annually”, but they do not give the basis for this statement – it is not in the research paper – and I do not recognise it. 1% of what? In the UK, according to the British Heart Foundation, there were about 190,000 hospital visits due to heart attacks in 2015. Just considering people aged 45 or over, there are about 28 million in the UK. So the risk, per year, of a heart attack in people 45 or over is considerably less than 1%. How can the extra risk due to NSAIDs be 1%. So are these reported 20% to 50% increases in risk really important? Hard to say on the basis of this study.

“A key aspect of the statistics used in the paper is that Bayesian methods were used, and this does allow the researchers to say useful things about the chances of various levels of increased risk. But one has to be careful. It’s true that the researchers conclude that there’s a chance of over 90% that each of the painkillers is associated with an increased heart attack risk. But that says nothing about the size of the increase. A very tiny increase in risk probably doesn’t matter. It also means that there is a chance, albeit a fairly small one (less than 10%), that painkiller use actually decreases the heart attack risk. More usefully, the Bayesian approach allows the researchers to estimate the probability that there is an increase in risk of a given size. Table 3 in the research paper also looks at the chance of a 50% increase in risk, and for most of the drugs for most doses, that chance is not particularly high.

“It’s important not to let the use of Bayesian statistics make a complete change to the way these results are interpreted. The right hand side of Figure 2 in the research paper shows ‘95% credible intervals’ for the odds ratios. These intervals have a subtly different interpretation to the confidence interval you might often see in a research report, but the difference is quite subtle. If a confidence interval for an odds ratio included the number 1, which means that there’s no difference in risk between those taking the NSAID and those not, the result is said not to be statistically significant, and might plausibly be due to chance variability. Most of these credible intervals do in fact include 1, so it’s arguably misleading to say (as the researchers do) that current use of all these NSAIDS “is associated with a significantly increased risk” of a heart attack. It depends on the dose and timing, and on the drug involved. Even with all these patients and all the heart attacks involved, the picture remains somewhat unclear, except for rofecoxib (Vioxx) – but that was withdrawn in 2004 because of concerns about heart attack and stroke.”


* ‘Risk of acute myocardial infarction with NSAIDs in real world use: bayesian meta-analysis of individual patient data’ by Michèle Bally et al. published in the BMJ on Tuesday 9 May 2017.


Declared interests

Dr Amitava Banerjee: “No conflicts of interest.”

Prof. Jane Mitchell: “I’m on the advisory board of Antibe Therapeutics.  I have acted as expert witness in cases relating to drugs such as celecoxib.  I have acted as consultant to companies with interests in drugs such as celecoxib and prostacyclin.  I have had grant funding from charities to investigate pathways related to prostacyclin and celecoxib.”

Prof. Kevin McConway: “I have no relevant interests to declare.”

None others received.

in this section

filter RoundUps by year

search by tag