Pain Relievers vs Heart Disease: What's the Link?

The World Health Organization (WHO) recommends a stepped approach to pain relief, consisting of acetaminophen and non-steroidal anti-inflammatory drugs (NSAIDs) as the pharmacologic mainstay.  However, the downside to acetaminophen is its ubiquity, despite being the number one cause of hepatotoxicity (liver failure) in the US, enough so that the FDA will soon be limiting the amount of acetaminophen in combination narcotic medications.

But I digress. I wanted to review some of the safety concerns about NSAIDs.  It's been known for quite some time that traditional NSAIDs, such as ibuprofen, can increase one's risk for stomach ulcers and kidney damage.  That led to the development of cox-2 inhibitors, a subclass of NSAIDs that was supposed to have a lower risk of stomach ulcers.  However, as we found out w/Vioxx (rofecoxib) and Bextra (valdecoxib), which were pulled off the market due to an increase risk of heart disease, there's no free lunch.  This then led to a re-examination of all NSAIDs to determine each drug's individual risk of heart disease.

In a study just published in the British Medical Journal, the authors analyzed data from 31 trials of 116,42 patients followed for 115,000 patient-years and concluded that most, if not all, NSAIDs increased one's risk for heart attacks, stroke, death from heart disease, and death from any cause, with rofecoxib the worst offender and naproxen the least offensive.  I wouldn't be surprised if these findings hit the lay press shortly w/some rather inflammatory sound bites, such as "Pain Killers Cause Heart Attacks!"

Mind you I'm not here to defend the pharmaceutical industry, but while the statistics do support pulling rofecoxib from the market, the numbers are not statistically significant for most NSAIDs.  To understand this, we have to review the principle behind the 95% confidence interval (or credibility interval as used in the BMJ).  What this range tells you is that if you repeated the analysis 100 times, 95 out of that 100 times you'd have results within this range.  As long as both the upper & lower limits are on the same side of 1, your results are statistically valid.  However, as in the case for a number of the NSAIDs, when the upper limit is greater than 1 and the lower limit is less than 1, the implication is that 95 times out of 100, you might be better or worse than your comparator.  That's no better than flipping a coin!

So while the sound bite might report a doubling of one's risk for heart attack from lumiracoxib (not sold in the States), the fact that the 95% confidence interval ranges from 0.71 to 6.21 implies that the results are not statistically significant.  It says nothing about the risk of heart attack, just that I wouldn't want to mislead my patients.