Dr. Clifford Rosen speaks candidly in the NEJM about the multi-faceted situation regarding the issue of cardiovascular risk of Avandia and how hard it is to pluck out the "truth" from often conflicting and incomplete data.
His full text comments are available here from the NEJM.
One of the important points he makes is that is more than just very hard to discern small increases in risk between groups when the disease or event of interest is fairly common such as myocardial infarction. I have written before about how statistics works well to find big difference but struggles to differentiate between small things and nothing at all.
Dr.Rosen provides a good overview of the various data sets that were analyzed in the FDA panel's effort to reach a decision about what to do with Avandia. I believe at the end of the day public sentiment and physician assessments of the risk has for the most part lead to the end of the road for Avandia.
He also speaks of several guiding principles to apply in regard to drug approval.
Two of which are :
the pathogenesis must be "fully understood"
treatment options should be clarified through an evidence based system.
Worthy aspirations to be sure, but I doubt we can wait for the pathogenesis to be "fully understood" to make judgments. What diseases can you think of in which we fully understand their pathogenesis, certainly not hypertension, heart failure,diabetes, depression for which many drugs have received FDA approval.
The Avandia saga and the panel's deliberations illustrate how difficult it can be and usually is to discern the truth from conflicting and incomplete data. At the end of the day someone has to decide something. It is commonly said of trials and court decisions that they do not produce justice so much as they settle the issue and the participants can move on. FDA's decisions are sort of like that.