In “Is your patient still using rosiglitazone?” (J Fam Pract. 2011;60:282-284), Drs. Gov-Ari and Stevermer question the continued use of rosiglitazone for type 2 diabetes in view of a reported increase in the risk of acute myocardial infarction (MI) associated with it. As a physician who continues to prescribe rosiglitazone, I would like to explain my rationale.
The meta-analysis upon which the various FDA advisories are based, like all meta-analyses, is “exploratory” or “hypothesis generating.” It is not gold-standard evidence from randomized controlled trials (RCTs), on which evidence-based medicine must be based. Neither of the 2 RCTs that have been done showed an increase in acute MI associated with the use of this drug.1,2
No patient in my practice who is taking rosiglitazone has sustained an acute MI. I have shown that the prediction of the population at risk of atherothrombotic disease, which includes acute MI, is independent of blood sugar levels.3
Hence, after appropriate counseling, I continue to prescribe rosiglitazone for patients who are willing to take it. After all, pioglitazone costs about $100/month more than rosiglitazone, and for many of my patients the increased cost would be devastating.
W.E. Feeman Jr, MD
Bowling Green, Ohio
1. Home PD, et al. Rosiglitazone evaluated for cardiovascular outcomes in oral agent combination therapy for type 2 diabetes (RECORD): Lancet. 2009;373:2125-2135.
2. The BARI 2D Study Group. A randomized trial of therapies for type 2 diabetes and coronary artery disease. N Engl J Med. 2009;360:2503-2515.
3. Feeman WE Jr. Diabetes mellitus is not a coronary heart disease equivalent. Am J Cardiol. 2010;105:754-755.
The authors respond:
We appreciate Dr. Feeman’s response to our PURL, which was based on a recent meta-analysis of RCTs of rosiglitazone.1 We are happy that none of his patients has been harmed by rosiglitazone, but note that the increase in MIs would be an uncommon event in most primary care practices. Over a 5-year period of rosiglitazone use, only one in 37 to 52 people would have suffered an MI. In a single practice, this rate of events would be difficult to detect, given the already elevated baseline risk in a diabetic population.
While we also agree that a well-done RCT may be superior to a meta-analysis, we disagree with the statement that evidence-based medicine should not be based on systematic reviews or meta-analyses. The problem is that the appropriate RCTs are often not done, in many cases because of logistic and financial issues. Almost none of the RCTs of rosiglitazone had the power to detect the increase in MIs. Few, if any, of the 56 trials included in the meta-analysis found a statistically significant difference in MI rate; however, the overall meta-analysis found an important increase (OR=1.28; 95% CI, 1.02-1.63). One of the advantages of meta-analyses is their ability to detect uncommon events distributed over a series of different studies.2 Safety evaluations are a particularly important use of meta-analyses, since drugs are often approved without any single RCT large enough to detect an important increase in serious adverse events.
The FDA has added restrictions to the use of rosiglitazone in the form of a Risk Evaluation and Mitigation Strategy (REMS), a program used to manage serious risks of marketed drugs.3 Only patients already successfully treated with rosiglitazone can enroll in the program, unless their physician does not wish to use pioglitazone when other antidiabetic agents have failed to provide adequate control. Prescribers and patients must enroll in the REMS program to be able to prescribe and receive the medication, which will be available only through specially certified pharmacies and dispensed only by mail. This additional step reflects the FDA’s significant concerns about MIs associated with rosiglitazone. We also note that at a popular Internet pharmacy (www.drugstore.com), the difference in cost between rosiglitazone and pioglitazone is less than $90 for a 90-day supply.
James J. Stevermer, MD, MSPH
Hanna K. Gov-Ari, MD
Columbia, MO