January 17, 2008

Are Drug Companies Hiding Negative Studies?

Let's skip right to the punching.

I.  The Not-Punching Part

A study in the Wednesday edition of the Journal of the American Socialist Party reports that 31% of antidepressant trials were not published, and almost all of the unpublished basically showed negative results.

This is bad, obviously, which is why we need a website for all raw data.   But let's be clear: this was a review of studies found in the FDA registry.  The FDA had this data, and used it to evaluate the meds. No one hid the data-- they gave the data to the FDA, all of it.  What didn't happen was publication.

So the real question is why didn't they get published.

Certainly, Pharma doesn't want negative studies published.   But these are Phase 2 and 3 clincial trials.  They're not done down at Lilly HQ-- these are done at universities.  Pharma didn't block their publication-- they were blocked by the academics who did them, and the journals themselves.

Hi.  Is this thing on?

You say: why would the academics themselves, committed as they are to science, block publication?  Because they'll lose their jobs, that's why.  Academics need grant money, and negative studies don't get grant money.  Not from Pharma, not from NIH, not from anywhere.   "Hi.  Nothing we've done is better than placebo.  Can I have tenure now?"

But more importantly, they don't publish the studies because they're not in the CIA.  When a study comes out negative, the academics don't break out the special redacting marker, they don't say, "better bury this."  They say, "Mama Mia!  It didn't work!" and start over.  Their bias is that the study should be positive; so if it comes out negative, the unconscious assumption is that something was wrong with the study.

You go up to a girl in a bar, with your new playa skillz, and she maces you in the face.  You don't assume you're a tool, you assume she's gay.  Oh, and you don't change your skillz.

Next are peer reviewers, who are unanimously dismissive of any study that doesn't separate from placebo.  They don't think its worthy of publication unless it showed a positive result: "this study does not meaningfully add to the existing literature..." 

Keep in mind these studies were done > 5 years ago, back when the culture wasn't "everything has equivalent efficacy."  Nowadays, that's the hot topic-- studies showing Pharma sucks, or branded meds are no better than generics.  Back "then" journals were all about finding the next big thing, the assumption of progress, etc. There's no room in journals for the null hypothesis.

Lead researcher and psychiatrist Erick Turner points out to The Wall Street Journal that doctors unaware of the unpublished studies can make inappropriate prescribing decisions for their patients.
My retina just detached.  Seriously?  That's the problem?

II.  The Punching Part

Says the study:

There can be many reasons why the results of a study are not published, and we do not know the reasons for nonpublication. Thus, we cannot determine whether the bias observed resulted from a failure to submit manuscripts on the part of authors and sponsors, decisions by journal editors and reviewers not to publish submitted manuscripts, or both.

That's it.  Two sentences.  Ok, let's grant them the asylum of ignorance.  Explain, then how those two sentences can be interpreted:

But Dr. Jeffrey M. Drazen, editor in chief of the New England Journal, explains to the New York Times why the study is so alarming for doctors and patients. "When you prescribe drugs, you want to make sure you're working with best data possible..." he says. Moreover, patients who agree to be guinea pigs "take some risk to be in the trial, "and then the drug company hides the data?" he asks. "That kind of thing gets us pretty passionate about this issue."

You have to have the deluded bravado of a DJ Khaled video to say the drug company is hiding data when, in fact, you are the one who is responsible.

The authors are themselves peer reviewers.  Did they ever review a negative article that they recommended publication?  

Lead author Erick Turner has "30 publications in peer reviewed journals."  How many of those publications had negative results?  One: B12 was not effective for seasonal affective disorder.  So did he submit negative studies and they were rejected, or did he simply discard them?  Turner was also a reviewer for the FDA-- why not simply release all that data?  Open acccess?  Don't give me this crap about clinicaltrials.gov.  Don't ask Pharma to put their data there. You already have the data-- just release it.  To his credit, he has already made this exact recommendation.  In 2004.  Making this article superfluous...

Dr. Drazen, above quoted editor-in-chief of the JASP, has to be insane.  HE'S THE ONE MOST RESPONSIBLE FOR THE REJECTIONS OF THE NEGATIVE STUDIES.  His predecessor was Marcia Angell-- arguably the single worst thing that happened to medicine, ever.  Together they form an impenetrable wall of meaningless social policy articles that no simple negative study could ever penetrate.    They reject articles showing Prozac is better than Zoloft; what chance does a Prozac=Zoloft article have?

But the cake goes to anchor author Dr. Rosenthal.  You know what his area of research is?  "Self- fulfilling prophecies" i.e. "the effects of experimenters' expectations on the results of their research."  This guy should know better-- I'm sure he knows better-- than to publish a study like this and not comment on the responsibility-- ok, the effect-- of the academics themselves.

But that would be asking the dog to bite the jaw that it was eating with, which is both impossible and painful.