Mar 07

Cardiovascular safety of NSAIDs

Journal reference: Trelle S, Reichenbach S, Wandel S, et al. Cardiovascular safety of non-steroidal anti-inflammatory drugs: network meta-analysis. BMJ 2011; 342:c7086 [1]


Published: 11 January 2011

Evidence cookie says...

Non-steroidal anti-inflammatory drugs (NSAIDs) appear to be associated with increased cardiovascular risks.

Naproxen appears to be associated with the least cardiovascular harm.

General practitioners should:

  • avoid NSAIDs in patients with elevated cardiovascular risk
  • avoid long-term use of NSAIDs

Note: there remains substantial uncertainly and the reliability of the evidence is limited. See below for more details.

More details:

Article details

Study design:

systematic review, network meta-analysis of randomised controlled trials

Study aim:

to analyse the cardiovascular safety of non-steroidal anti-inflammatory drugs (NSAIDs)

Methods summary:

  • search:
    • bibliographic databases
    • conference proceedings
    • study registers
    • FDA website
    • reference lists of relevant articles
    • through the Science Citation Index
    • contacted all authors of primary trial reports and manufacturers of NSAIDs for missing outcome data
  • search last updated in July 2009
  • two investigators independently assessed trials for eligibility and extracted data

Trial inclusion criteria:

  • large scale randomised controlled trials (RCTs)
  • comparing any NSAID with other NSAIDs, or paracetamol or placebo for any medical condition
  • at least two arms with 100 patient years of follow-up

Primary outcome:

  • fatal or non-fatal myocardial infarction

Secondary outcomes:

  • stroke
  • cardiovascular death
  • death of unknown causes
  • death from any case
  • composite outcome: non-fatal myocardial infarction, non-fatal stroke, or cardiovascular death (used by the Antiplatelet Triallists’ Collaboration)


  • interventions included for analysis must have had at least 10 patients allocated to the intervention who suffered a myocardial infarction in all eligible trials combined
  • excluded comparisons with zero events in both groups from the relevant analysis
  • whenever possible, used results from intention to treat analysis of the longest follow up available
  • heterogeneity of treatment effects assessed

Results summary:

Primary outcome (myocardial infarction):

  • rate ratio comparing the intervention to placebo (95% credible interval):
    • naproxen: 0.82 (0.37-1.67)
    • ibuprofen: 1.61 (0.50-5.77)
    • diclofenac: 0.82 (0.29-2.20)
    • celecoxib: 1.35 (0.71-2.72)
    • etoricoxib: 0.75 (0.23-2.39)
    • rofecoxib: 2.12 (1.26-3.56)
    • lumiracoxib: 2.00 (0.71-6.21)
    • interpretation note:
      • rate ratio < 1.00 means that the intervention was associated with fewer myocardial infarctions than placebo
      • rate ratio > 1.00 means that the intervention was associated with more myocardial infarctions than placebo
      • credible intervals that include 1.00 in the range means that no statistically significant result was found

Other outcomes (Forest plots based on figure 2 [1]):

Other analyses:

Study conclusion:

Although uncertainty remains, little evidence exists to suggest that any of the investigated drugs are safe in cardiovascular terms. Naproxen seemed least harmful. Cardiovascular risk needs to be taken into account when prescribing any non-steroidal anti-inflammatory drug.


  • included trial characteristics given but grouped participant data not given
  • populations in the included trials included:
    • at risk for Alzheimer’s disease
    • Alzheimer’s disease
    • rheumatoid arthritis
    • adenomatous polyps of the colon
    • osteoarthritis
  • total number of patient years in included trials: 117 218
  • number of myocardial infarctions: 554
  • number of strokes: 377
  • cardiovascular deaths: 312
  • death from all causes: 676

Methodological weaknesses

  • relatively low number of participants and events when looking at individual agents
  • low power; most results with very wide credible intervals
  • rate ratios cannot be easily translated into absolute risks or number needed to treat / number needed to harm
  • the study was not able to examine all NSAIDs including some commonly used agents, e.g., meloxicam (Mobic), piroxicam (Feldene), ketoprofen (Orudis) and indomethacin (Indocid)
  • probable existence of unpublished evidence that was not included (Merck & Co., the manufacturer of rofecoxib and etoricoxib was not willing to share unpublished safety data)
  • study participants took NSAIDs regularly over a long term period which is different to how NSAIDs are often used in the community; this is a threat to the external validity of the results

Methodological strengths

  • comprehensive search strategy
  • although the meta-analysis had wide credible intervals, they are still narrower than previous meta-analyses examining cardiovascular safety of NSAIDs
  • the results of this meta-analysis is less likely to have confounding factors than previous observational studies
  • network meta-analysis allows indirect comparisons to be made (i.e., all interventions as compared to placebo even if the original study did not compare to placebo)

Biases and conflicts of interests

  • funding and grants from the Swiss National Science Foundation
  • no other biases or conflicts of interests declared and none seem obvious on the part of the meta-analysis authors
  • it is likely that there is safety data held by pharmaceutical companies who manufacturer and market NSAIDs that was not released to the study authors

Clinical relevance to primary health care

Non-steroidal anti-inflammatory drugs (NSAIDs) are widely used analgesics in Australia. There has been concern in recent year over the cardiovascular safety of long-term use of NSAIDs.

Rofecoxib (Vioxx) was withdrawn from the Australian market on 1 October 2004, following the worldwide withdraw of the drug by Merck & Co [2]. Of concern, the VIGOR study comparing rofecoxib to naproxen (Naprosyn) found a statistically significant 4- to 5-fold increase in myocardial infarction in the group assigned rofecoxib [3]. The authors speculatively (and dubiously) concluded that the difference was due to a cardioprotective effect of naproxen. This finding was known in 2000, 4 years before the drug was finally taken off the market. In the interim, several other studies demonstrated the increased risk of cardiovascular events with rofecoxib.  By the time rofecoxib was withdrawn, at least 250,000 Australians had taken rofecoxib, many for reasons other than the approved indication [2]. Valdecoxib (Bextra), another COX-2 inhibitor (never released in the Australian market), was withdrawn in the United States in April 2005 due to an increased risk of heart attack and stroke.  Etoricoxib (Arcoxia), also a COX-2 inhibitor (and available on private prescription in Australia) was denied approval by the US FDA, citing an inadequate risk-benefit profile [1]. Since 2004, there have been many observational studies demonstrating an association between use of NSAIDs and risk of myocardial infarction. Moveover, there was no evidence to support a reduction in the risk of myocardial infarction attributable to naproxen [4].

Reliable high level evidence on the cardiovascular safety of NSAIDs remains limited. The only statistically significant result found for the primary outcome, myocardial infarction, was for rofecoxib, which increased the risk two-fold. This is not illuminating given that it has already been removed from the market. Nevertheless, a mix of interesting and statistically significant results for secondary outcomes on current drugs were found, including:

  • ↑ risk of stroke:
  • ↑ risk of cardiovascular death:
    • diclofenac
    • etoricoxib
  • ↑ risk of death from any case:
    • diclofenac

It should be noted that much of the results are limited by wide credible intervals; the numbers of studies and participants with events were relatively small. However, there is a consistent trend towards cardiovascular outcomes for most NSAIDs in the analysis that is arguably clinically significant even if it lacks statistical significance. Naproxen appears to be the NSAID with the least associated cardiovascular harm (no associated increased risk of myocardial infarction or cardiovascular death, but a non-statistically significant increase in the rate of stroke and death from any cause).

It is prudent for Australian general practitioners to follow a precautionary principle:

  • avoid NSAIDs in patients with elevated cardiovascular risk
  • avoid long-term use of NSAIDs
  • where the benefits of long-term NSAIDs are considered to outweigh their risks, naproxen appears to be the NSAID least associated with cardiovascular harm


  1. Trelle S, Reichenbach S, Wandel S, et al. Cardiovascular safety of non-steroidal anti-inflammatory drugs: network meta-analysis. BMJ 2011; 342:c7086
  2. COX-2 chronology. Four Corners, Australian Broadcasting Corporation. 2005. Retrieved 7 February 2011:
  3. Bombardier C, Laine L, Reicin A, et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. N Engl J Med 2000; 343:1520-1528
  4. Hippisley-Cox J, Coupland C. Risk of myocardial infarction in patients taking cyclo-oxygenase-2 inhibitors or conventional non-steroidal anti-inflammatory drugs: population based nested case-control analysis. BMJ 2005; 330: 1366
Print Friendly, PDF & Email

Leave a Reply

%d bloggers like this: