In chasing down the perfect p-value (p<0.05), we can sometimes forget the overall objective of performing a clinical trial. You can have the best-designed study in the world with all the statistical issues carefully considered, and the trial can succeed, but if your trial result isn’t _important_, or _clinically significant_, then the statistical significance means _nothing_.
Genta, Inc. unfortunately got bit by this fact as the FDA’s ODAC(Oncologic Drugs Advisory Committee) “voted against their Genasense”:http://factsandcomparisons.com/News/ArticlePage.aspx?id=7361 product on the basis that their statistically significant difference wasn’t large enough to be clinically significant. (Apparently their drug also increases the toxicity of chemotherapy.)
Of course the FDA is not bound to the opinion of the ODAC(Oncologic Drugs Advisory Committee), but this certainly isn’t good news for Genta. It should be enough to give pause to anyone developing a drug whose effect isn’t huge.
Filed under: Drug industry, Lying with statistics
