I had all of 4 minutes on CNBC this week – which I shared with Merrill Goozner -- to answer the question, “Should the FDA be speeding cancer drugs to market?” What a question.
Answer: YES -- if they work, NO -- if they don’t! (Merrill agreed. that left 3 minutes of dead space.)
But there’s the rub. What does it mean for a drug to “work?” Is extending life three months at a cost of ten to twenty thousand dollars per patient “working?” Is a drug that has a 92% chance of not being a placebo effect “working” compared to a drug that has a 95% chance?
This issue came to a head at the recent meeting of the American Society of Clinical Oncologists – otherwise known as “cancer doctors.” According to the New York Times, several people who had spoken out against a particular cancer drug received threats, even death threats. Apparently some patients felt they had the right to have this drug approved as a last resort even if many scientists felt it did not “work.” And if those scientists didn’t support “saving their lives”, these angry patients felt comfortable threatening them. This is an example of a system that is not working.
Or rather, the system is working just as we set it up to work. We have a $100 billion research enterprise feeding products into a $1 billion drug review system, funded to an ever growing degree by industry user fees. At the front end, we have some pretty strict rules for what constitutes “working.” And anything less than a high “non-randomness” score (i.e. 95% chance of not being random) won’t “work.” Try telling that to terminally ill patients who want to take a drug that has a score of 90 or 85.
Meanwhile, those user fees have been restricted for years from being applied to post market surveillance of newly approved drugs. If you can’t do post market surveillance, you don’t know how a drug is working in the real world until someone reviews random adverse event reports from doctors and starts comparing them to previously obtained clinical trial data that may or may not have tested for the reported effect. And then when deadly effects are discovered in some people, all hell breaks loose because nobody was monitoring whether the drug “worked.”
Avandia. Vioxx. Accutane. Thalidomide. All these drugs work for some people in some very important ways. For other people, they don’t just not work, they’re deadly. Until we have a system that works, it will be hard to tell whether the drugs are working ….or wasting both money …and lives.
Greg
2 comments:
self-interest note: I take Avandia.
For you to put these 4 words together in this context:
"Avandia. Vioxx. Accutane. Thalidomide."
... is simply irresponsible. Just like the meta-study of Avandia.
While I certainly agree with you that the current system doesn't work, I seriously, seriously doubt that anything the FDA will EVER do will make it work better for anyone, except for the drug companies and the politicians.
As long as there are ill people who have money and blood, there'll be robbers vampires to liberate it from them. All while claiming to do good.
rr
Greg, you leave open the question of exactly how you'd prefer the FDA to decide whether a drug "works" or not. My sense is that the agency has to set the bar somewhere, and the 95% confidence level is pretty well established as an arbitrary but accepted standard. CEOs like Richard Miller complain about the "arbitrariness" of that standard without actually outlining what they'd rather see -- except, of course, a standard that would let a drug like Xcytrin receive approval.
So what are your standards? What should the FDA be doing differently? Or are the frustrated would-be Provenge patients just the victims of the hype-and-retreat cycle that recurs so frequently in biotechnology?
Post a Comment