Tuesday, February 4, 2014

All NSAIDs are Created Equally . . . Right?

A guest post by Anand Swaminathan (@EMSwami) of EM Lyceum and Essentials of EM fame.

It’s been more than 10 years since Merck made headlines regarding the increased risk of atherothrombotic events with it’s drug rofecoxib (Vioxx) and 3 years since the almost $1 billion settlement regarding the drug. Over this time period, physicians have questioned the use of all non-steroidal anti-inflammatory drugs (NSAIDs) in patients with increased risk of coronary artery disease. Rofecoxib and its brethren were developed in the 1990’s as alternatives to the traditional NSAIDs (tNSAIDs). These newer drugs only inhibited COX-2 and theoretically would lead to less gastrointestinal bleeding since inhibition of COX-1 is believed to lead to this side effect. Based on this pathophysiologic theory and strong marketing, the drugs became widely prescribed. Unfortunately for our patients, saving the gut may compromise the heart. Or is it actually that all NSAIDs, regardless of which COX they inhibit, raises the risk?

This research group performed a meta-analysis of a large number of trials looking at both NSAIDs vs. placebo (280 trials) and one NSAID vs. another NSAID (474 trials). All of the studies look at length of use of at least four weeks but many had longer treatment durations. So, what did they find? When compared to placebo, coxibs (COX-2 inhibitors), diclofenac and ibuprofen all increased the risk of major coronary events but naproxen did not:

Coxibs RR (CI)
Diclofenac RR (CI)
Ibuprofen RR (CI)
Naproxen RR (CI)
Major Vascular Events
1.37 (1.14 – 1.66)
1.41 (1.12 – 1.78)
1.44 (0.89 – 2.33)
0.93 (0.69-1.27)
Major Coronary Events
1.76 (1.31-2.37)
1.70 1.19-2.41)
2.22 (1.10 – 4.48)
0.84 (0.52 – 1.35)
Gastrointestinal Complications
1.81 (1.17-2.81)
1.89 (1.16-3.09)
3.97 (2.22-7.10)
4.22 (2.71-6.56)

Additionally, it didn’t matter which coxib they looked at: all of them increased major coronary events. The coxibs did, however, do what they were supposed to do; the risk of gastrointestinal bleeding was lower in comparison to tNSAIDs like ibuprofen and naproxen.

Does this change what we should do? Can we give NSAIDs to older patients without fearing a major coronary event or death from an MI? The numbers here are small but important. For every 1000 patients allocated to a coxib, three more had major vascular events and one of these events would be fatal. That’s an absolute increase of 0.1% in comparison to placebo. Small increased risk but avoidable. Why not give naproxen instead? In fact, NSAIDs like ibuprofen interfere with the binding of aspirin negating its cardioprotective effects if the two are taken together. Naproxen doesn’t have this issue.

The relative safety in terms of major vascular events associated with naproxen versus ibuprofen was recently written about by the FDA and reported by the Associated Press here -


I think the bottom line here isn’t much of a surprise. If you can, avoid NSAIDs in patients with cardiac risk factors. Ibuprofen isn’t a safe alternative. If you need to give an NSAID for whatever reason, use naproxen and use short courses of therapy.

"Vascular and upper gastrointestinal effects of non-steroidal anti-inflammatory drugs: meta-analyses of individual participant data from randomized trials"


  1. Wondering about that table and the meta analysis, I know of a previous meta-analysis that showed ibuprofen to be safest for side effects, but thought to be due to relatively lower doses of ibuprofen compared to the others. Does the meta-analysis from which this table is derived account for the relative doses of each drug? My understanding is that the adverse reactions for NSAIDs increase rapidly as the dose is increased.

  2. This is a complicated area.

    First we should be comparing iso-effective doses of NSAIDs.

    I think we have no idea of what they are.
    What if you need huge doses of naproxen to match the efect of 150 mg diclofenac per day ? Does the preference still hold ? ANd all other comparisons are possible.

    But if some one knows please tell !

    ANd theoretically the % inhibition of Cox 2 1 is important. Not obvious.


    Where are the absolute risks ?
    Where are the figures we can understand ?
    I once calculated from some papers the difference between good and ugly NSAIDs was in the range of 0.3-0.5 events (cardiovascular) per 100 years/patient.

    Are we certain such tiny differences seen in trials are the pure truth and can be "measured"? I'd follow Nortin Hadler in thinking they can't .

    It is thus politically correct to prescribe naproxen which some algologists say works poorly, and hurts the epigastrium more ... I abide by the correctness, but keep my doubts. Without any conflict of interest. Pure as a baby.

    ANother problem I see is that this is an additional area of lousy research where monocular shortsighted people look at the landscape with a telescope and then publish a "meta-analysis" That ignores much of what is very important. They study the cardiovascular endpoints and don't tell us about the others.

    Mortality from all causes would be a nice one to start with since we were once told 16000 US citizens died from an NSAID induced digestive haemorrhage per year . Plus those not registered !

  3. Thank you for the comments. All excellent points. One of the major problems/limitations with all of the NSAID study comes back to dose. It's unclear what the optimal doses are. It seems unlikely that there's a difference in terms of pain relief between 400 mg and 800 mg of ibuprofen but the higher dose likely has more side effects. We see this with ketorlac (not addressed in this study) as well. The recommendation to go with naproxen instead of ibuprofen is really based on the best available information. The interference of ibuprofen with the action of aspirin is central to any decision as well.

  4. Hi Anand.
    And thanks for your papers , here and at Lyceum.

    The interference of ibuprofen with the action of aspirin isn't such a concern to me for 2 reasons:
    - I don't prescribe NSAIDs to aspirin treated patients. Rarely if ever.
    - It's an interesting theoretical point, but if it were important it should translate into the cardiovascular accident figures. And we sees the RR intervals overlap. We need absolute risks. The ones I remember are tiny tiny . Do you have any down-to-earth figures ?

  5. Interesting. I posted on this topic this morning and made the point that cardiorenal risks are at least as important as cardiovascular. In that respect naproxen is no exception. In the meta-analysis it was associated with a marked increase in heart failure. And what about non-acetylated salicylates as an alternative for patients who need NSAIDs?


  6. Great points. Agree that ibuprofen shouldn't be given to those on ASA but I see this happen frequently.
    As far as heart failure risk, I think this speaks to prolonged therapy as well. We should keep it short. Many times, the default in an EMR is a months worth of pills with multiple refills. We should be careful to avoid this.

  7. Nice post. For sure heart failure can occur early in therapy. All NSAIDs can reduce glomerular perfusion pressure, leading to Na+ and H20 retention. This is particularly likely in people who have systolic LV dysfunction or are 'prenal' for some other reason. Very much agree with your point about limiting duration. True of opioids too, but for different reasons.


Comments on posts > 10 days old will be moderated; blame spammers.