Thursday, March 24, 2005

Sugar Pills for the Dying

A Wall Street Journal editorial asks why terminal cancer patients must be given a placebo, rather than a drug that may save their lives, in order to satisfy FDA bureaucrats. Link.
The problem here is the FDA's unethical -- and let us stress, unscientific -- insistence on gathering information about drugs by way of "blinded" placebo-controlled trials, in which a subset of study patients are knowingly denied the new treatment and in some cases denied access to any active treatment at all. This may be moral with an antihistamine; it's certainly not with treatments for a terminal disease. What's more, it's entirely unnecessary. We already know what happens to most cancer patients who don't get treated. They die. We generally know, on average, how long that will take.

So placebo groups are entirely unnecessary to prove significant anti-cancer activity, as the yet-unnamed Bayer (BAY 43-9006) and Pfizer (SU 11248) compounds have already done. Yet the FDA is mandating an unethical placebo trial for the Bayer drug. (The Pfizer drug is at least being tested against another form of care, albeit one that's already all but certain not to work as well.) A deadly follow-on effect of the placebo fetish is that it gives companies a disincentive to run compassionate use programs for unapproved drugs. That's because companies won't be able to satisfy FDA demands to enroll patients in placebo trials if patients know they can get the drug for sure (instead of running the risk of getting a sugar pill) through compassionate use. (...)

In almost all recent cases of FDA dawdling, the drugs are proving to be far more beneficial in practice than even the supposed "gold-standard" of placebo trials would have ever suggested. So could someone explain, again, what the benefit is of doing such trials? We're not suggesting Dr. Pazdur is some kind of ogre. But he seems to be more worried about letting drug companies get away with a so-called "race to the bottom" on trial design than he is with getting good drugs to patients. And it's obvious that he can't (or won't) be educated in modern scientific and statistical methods that would allow drugs to be released sooner. (...)

Instead of restricted-access placebo trials, drug researchers could be using large, open access trials in which everyone who wants the new drug can get it. They could then take advantage of advanced statistical methods to figure out whether the drug is working. Wall Street traders use these kinds of math tools all the time, and so do economists. So-called Bayesian statistics are already used in medical device regulation, where even the FDA recognizes that randomizing people into sham surgeries is simply beyond the pale.

For all practical purposes, it is impossible to reform an existing federal agency. It is a better idea to abolish the agency and replace it with a new one that is designed the way you want it. In a January 26 post, I argued that the FDA should be returned to its original purpose of approving only the safety of drugs, not their efficacy. Nearly every medical specialty has an organization that certifies doctors in that specialty. I would rather trust those organizations to decide on the efficacy of drugs than federal bureaucrats. They are far more likely than bureaucrats to keep up with the latest advances in statistical testing, and could identify efficacious drugs without murdering large numbers of Americans in order to satisfy their obsolete standards of proof.


Post a Comment

<< Home