Recently there was an electronic pre-publication ahead of print by Lancet, the British medical magazine. The article (and the accompanying editorial) discussed the result of a meta-analysis (and see my prior blogs for the errors inherent in meta-analyses) of 639 randomized trials of Cox-inhibitors (such as Celebrex) and NSAIDs (ibuprofen or Motrin/Advil, naproxen or Aleve, and diclofenac or Voltaren) to see if their use (a) caused increased upper GI bleeding and/or (b) increased vascular events (heart attack, heart failure, stroke). You may recall that Vioxx, a cox-inhibitor, was taken off the market because of an increased incidence of cardiac events.
The results were first announced as an increase in relative risk, which always looks worse than the increase in absolute risk. The studies looked at the effect of either a standard dose of a coxib, or a high dose (150mg diclofenac, 2400mg ibuprofen, 1000mg naproxen). The studies looked at major vascular events (heart attack or stroke), heart failure, overall mortality, and upper GI complications.
There was an increase in major vascular events of 33% with the use of Celebrex and Voltaren. There was also an increased risk with Motrin, but it did not reach clinical significance. There was no increased risk with Aleve. OTOH, the use of Aleve was associated with the greatest risk of upper GI bleeding compared with the other three drugs. And only coxibs showed a statistically significant rise in all-cause mortality.
In the case of vascular events and upper GI bleeding, the greatest risk factors were age, and conditions (diabetes, hypertension, GERD) that pre-dispose to a problem.
Now for the absolute numbers, which illustrate the real risk. Using a coxib or diclofenac was associated with 3 additional vascular events per 1,000 person years, one of which was fatal. If a patient had a 2% annual risk of a vascular event, then a coxib would add 7 additional events per 1000 patient years (2 of which would be fatal), Voltaren would add 8 (2 fatal), and Advil would add 9 (3 fatal).
In patients with a 0.5% annual risk of upper GI complications, coxibs and diclofenac would add 4 more events per 1000 patient years, ibuprofen 15 more events, and naproxen 16 more.
All four drugs showed a statistically significant increase in hospitalizations due to heart failure.
Once again, we see that the absolute risk seems less serious than the relative risk, as is almost always the case. Also, no studies were done at a lower dose of the drugs, and no studies were done to see if the addition of a PPI or other acid inhibitor lessened the risk of upper GI complications.
The data thus shows a small but statistically significant risk with some of these drugs to cause problems.
These problems occurred more frequently in high risk groups, and you should discuss with your doctor which group you fall into. To my mind, the data is barely significant, and the use of these drugs is a judgment call, and one has to compare quality of life to statistical risk.