The CEOs of drug and medical device companies must all share one big fear: that fellow scientists or journalists will reanalyze the data they've published and come to different conclusions. It happened to GlaxoSmithKline
An article in today's Wall Street Journal suggests that the company used a flawed method -- called the Wald equation -- to figure out whether its Taxus Liberte drug-eluting stent is different than Boston Scientific's already-approved Taxus Express. Boston Scientific didn't need to prove that the Librete was better than the Express, just that it wasn't inferior. Other statistical methods used by the paper showed that the data wasn’t strong enough to show that the Liberte wasn't inferior to the Express, failing accepted standards of statistical significance.
The paper makes a decent case for why the Wald equation shouldn't be used, but I doubt it's going to sway the Food and Drug Administration's view of Boston Scientific's stent.
The reason is pretty simple: The FDA accepted the use of the Wald equation in Boston Scientific's discussions with the agency about how to set up the trial, so the agency is likely to agree to the conclusions that the new stent isn't inferior. In fact, it would be bad scientific practice to switch statistical measurements after the data is known. You can't come up with a hypothesis and then shop around for a statistical method that helps you prove your case; the statistical plan needs to be in place before the trial begins.
Boston Scientific really needs a U.S. approval of the Liberte, since the drug-eluting stent market has gone from a duopoly between it and Johnson & Johnson
Anything can happen in this FDA regulatory environment -- just ask Cardiome