The new report isn’t based on anything new, though. It’s based on a study published a year ago – a seriously flawed study that I dissected at length on this blog in November 2008. The FDA took a year to review this study and then said yep, we agree. I should mention that the 2008 study has one condition before recommending Crestor: that the patients should have elevated levels of C-reactive protein (CRP). I described in earlier blog post the problems with this recommendation – including the fact that they study’s leader has a patent on the CRP diagnostic test.
Does it matter that, as I explained a year ago, the study was funded by AstraZeneca? How about the fact that this huge study found only a tiny benefit? That’s right: the benefit is so small that you’d have to treat 95 patients for 2 years in order to prevent one heart attack. As I wrote last year, and as others have pointed out, when the number needed (NN) to treat is over 50, the result is probably just statistical noise. Or to put it another way, if you take Crestor for 2 years, you have about a 1% chance of getting any benefit. Compare that to, say, taking vitamin C to treat scurvy, where the NN is about 1. With any decent drug, if you take it, you should have a very high likelihood of benefit.
We know who’s going to benefit here: AstraZeneca. Does it matter to them that there is a significantly higher risk of diabetes in people who take Crestor? There was also a higher number of deaths in the placebo group caused by gastrointestinal disorders, but the FDA dismissed this number because it didn’t reach statistical significance. (Kudos to the WSJ and Reuters reporters for highlighting these problems.) And no one besides me (see my Nov 2008 blog post) seems to have noticed some of the other problems with the study, such as the fact that the placebo group contained an excess in the number of patients with conditions that raise the risk of heart disease.
Does it matter that, as the FDA reported in 2005, “Rhabdomyolysis (serious muscle damage) has been reported in patients taking Crestor as well as other statin drugs.” Or that, as the same report said, “Kidney failure of various types has also been reported in patients treated with Crestor as well as other statins”?
So who the heck are these FDA scientists? After considerable searching, I found the lengthy FDA report here. It’s a “briefing document” for a FDA advisory committee that will meet on Dec. 15. It basically reviews the study in great detail and simply repeats the study results. What became clear to me once I downloaded the document, though, was that it was written by AstraZeneca! [Note: thanks to the first commenter on this blog, I also found the separate FDA-sponsored document - see comments.]
The reports from the AP, Reuters, and the WSJ I cited above mention an anonymous FDA reviewer, but apparently this “reviewer” did not want to be named on the record. It's too bad none of the reporters revealed the name of this anonymous FDA scientist, but the roster of the FDA panel reviewing Crestor is here.
It’s a shame that the FDA’s own scientists can’t take a more critical look at the evidence. Look for AstraZeneca stock to go up next week. Let me close with a video from Stephen Colbert, who had a great skeptical spoof of this last year on his show. Skip forward to about the 2-minute mark in this video to see the Crestor segment.
| The Colbert Report | Mon - Thurs 11:30pm / 10:30c | |||
| Cheating Death - Women's Health | ||||
| ||||
As Stephen says in the video, "this is a great breakthrough in the battle to find things to prescribe to people who don't need them".
Nice.
There are two briefing documents, one by the sponsor and one by the FDA. You must be looking at the one by the sponsor. Go back to the FDA webpage and you will see a link to the FDA review. The name of the FDA reviewer is disclosed.
ReplyDeleteMea culpa - and thanks for catching this. You're right, the FDA reviewer, Mary Dunne Roberts, wrote a separate document, somehow I missed it the first time. (I don't know why none of the reporters mention her name.) The end of the report contains the names of two statistical reviewers, David Hoberman and J. Todd Sahlroot.
ReplyDeleteThe reviewers did not notice, unfortunately, the problem I pointed out last year that the placebo group was not balanced for all risk factors; she just accepted the claim in the study that there were no significant differences between the groups.
They do point out, though, that "it is not clear whether or not rosuvastatin [Crestor] confers a total mortality advantage compared to placebo." That is, the study looked at particular treatment endpoints (effects), but it did not look at death, where there is not a clear benefit to Crestor.
We'll see what the Advisory Panel decides on Dec 15.
You point out that the study was funded by the manufacturer. Who instead should fund such studies? How much would it cost for all the necessary research to be done by third parties?
ReplyDeleteJosh makes a fair point. Who will fund all the development studies. Are you arguing, ex post facto, that every industry study ever conducted on every product cannot be trusted and that the efficacy of the products are not proven? How about a scientific study using clinical endpoints to make your care instead of just making assertions based on methodological quibbles. And for that matter, are government sponsored studies any better?
ReplyDeleteYou guys seem to miss my point. The mere fact that the study is funded by the manufacturer does not invalidate it - certainly not. But it does mean that there might be a bias on the part of the investigators. In fact, a study not long ago showed that a large proportion of manufacturer-funded studies produced positive results for the drugs being studied. This means we need to scrutinize these studies carefully - and be skeptical.
ReplyDeleteThat's what I did with the Crestor study - I looked at the details. And in my view, the benefits are so small that they might be clinically insignificant, and perhaps even harmful overall when you consider the side effects of putting millions more people on this drug.
In the 1970's, 70% of all university research was funded by the government. The research was helpful to me, as a physician, in caring for my patients.
ReplyDeleteBy the mid-1990's, 70% of university research was funded by industry. This led to my being less effective as a physician.
Nutrasweet research is an example of why we, the tax-payers, should pay for research. Recent news reports noted that the majority of research showed Nutrasweet to be safe. However, when one looks at the funding for those studies, almost all of the studies saying Nutrasweet is safe were funded by industry. These are the same people that told us Thalidomide led to safer pregnancies.
One year, treatment protocols indicate I should use a certain drug, Crestor or Zyprexa. A few years later, I learn that I have caused a lot of damage to my patients using these "indicated" agents.
Either we pay for the research, or we pay for the effect of the mistakes.