Rosiglitazone (Avandia – GSK), an oral hypoglycemic agent (OHA) used in treating diabetes, has been in the news recently because of concern that it is associated with an excessive number of adverse events in users of the drug.
After a week of media hype with headlines such as “Avandia users experience 30% more heart attacks,” the FDA on July 14 determined that existing warnings regarding the drug are sufficient and that the drug does not need to be removed from the market.
So what’s the truth? Why the hype? Why the confusion?
On June 28, 2010, JAMA published an article by Graham et al. entitled “Risk of Acute Myocardial Infarction, Stroke, Heart Failure, and Death in Elderly Medicare Patients Treated With Rosiglitazone or Pioglitazone” (JAMA. 2010;304(4):(doi:10.1001/jama.2010.920). The authors’ research found Avandia to be more dangerous than a comparison drug pioglitazone (Actos), another OHA.
But what are the facts?
Here are some pertinent points about the study and others leading up to it:
- Included only Medicare beneficiaries age 65 and over, with 70% of the patients being 70 years of age or older.
- Compared Avandia with only one other drug, not with a placebo group who took neither drug.
- The comparison drug, Actos, actually improves outcomes in those using it, so the question is really not which drug is “dangerous,” but which drug provides the desired effect more safely.
- Previous studies on Avandia have shown inconsistent results.
- Previous studies showing a “statistically significant” difference do not necessarily mean that there is a “clinical significance” to the difference, when all mitigating factors are considered.
- The study followed patients for a maximum of 3 years.
So what did Graham and his colleagues actually find and report?
- The “danger” of Avandia is calculated based on 100 person-years of use. (How many diabetics will live another 100 years?)
- Heart attack occurs 1.83 times for Avandia vs. 1.68 times for Actos during 100 person-years of use.
- For stroke, the numbers are 1.27 vs. 0.95
- For heart failure, the numbers are 3.94 vs 3.00
- For combined heart attack, stroke and heart failure mortality, the numbers are 9.10 vs 7.42
- For mortality from all other causes, the numbers are 2.85 vs 2.40
- For combined mortality, the numbers are 9.10 vs. 7.42
This is where the hype begins. Remember that Actos, the comparison drug, actually improves mortality. Thus, presumably so does Avandia, but that argument is beyond the scope of this “Perspective.” Avandia is thus beneficial, but just associated with more potential risks.
Let’s look at relative vs. absolute risk. Suppose my chance of winning the lottery is one in a million. Suppose I develop a “system” that improves my chances of winning the lottery to one in 100,000. Do I now have a 90% better chance of winning the lottery? Yes. Do I now have a 90% chance of winning the lottery? No. My chance of winning increased from 0.000001 to 0.00001, or only 0.000099. The former number is my “relative” increased chance of winning. The number 0.000099 is my “absolute” increased chance of winning the lottery.
Based on this analysis, should I buy a lottery ticket? It depends.
The same holds true for Avandia vs. Actos. For simplicity’s sake, let’s just look at the “all cause mortality” numbers of 9.10 vs. 7.42 per 100 patient-years of use of either drug. The media reports that Avandia is “24% more likely to cause” death than Actos because 9.10 is 24% greater than 7.42. In reality, one’s chance of dying while using Avandia is only 1.78% greater (9.71 – 7.42) than with the other drug. And that’s for a person taking Avandia for 100 years.
To summarize, let’s give each patient in the study a 20 year life expectancy during which they continue to use Avandia. Five patients will make up 100 patient years of use. Accepting the authors’ numbers in this study, during their remaining 20 years, 4 of those 5 patients will be unaffected by their use of Avandia, while only one patient will have a 1.78% increased risk of dying as a result of taking the drug. The “per patient” risk is thus reduced by 80% to 0.356 (1.78 / 5) over their 20 year life expectancy.
Bottom line: Only 1 in 281 [(100 / 0.356] diabetic Medicare patients 65 year of age or older will are affected by taking the drug between age 65 and 85.
When the media blares that Avandia causes a 30% increased chance of death, many patients hear “I have a 30% chance of dying from this drug.” Not true. Nor do they have even a 30% chance of having a worse outcome.
Absolute risk is important. It puts relative risk in its proper perspective.
We have undoubtedly not heard the last of this, but the question is, “Did the FDA make the right decision.” In this author’s mind, the answer is a definite “Yes.”