Did Drug Companies Try to Hide Suicidality?

Like the tobacco companies knowing nicotine was addictive yet hiding it from the public, did the antidepressant manufacturers know that their data implied increased suicide risk -- and try to conceal that from the FDA?  (or even from themselves? UPDATE 4/9/04 -- the Los Angeles Times obtained a document that suggests the FDA may indeed have tried to keep it from themselves: here's that story)

At least one company coded "possibly suicide-related events", as they are now called by the FDA, under the term "emotional lability", lumped there with other non-suicidal symptoms and behaviors.  Listen to Dr. Laughren of the FDA commenting on how they decided to request more specific data from manufacturers -- i.e. not just accepting the data as the company had filed it.  Notice how the response they got led to the ban on antidepressants for kids in Britain: 

(From the FDA 2004 hearing transcript: )

Right now let me talk a little bit about


 10   how the signal came onto our radar screen.  We had


 11   reviewed over the past three to four to five years


 12   pediatric supplements for 8 drugs, and we looked at


 13   the safety and efficacy data for these drugs.


 14             In the course of putting together a report


 15   for FDA, companies code their adverse event data,


 16   and they do this in their own ways.  We don't tell


 17   them how to code the data, they choose their own


 18   dictionaries and they set about coding the data


 19   before they send it in.


 20             This applied to any events suggestive of


 21   suicidality, as well as any other adverse events.


 22   We reviewed those supplements over this period of


 23   three to four years, and suicidality did not emerge


 24   as a matter of concern based on those reviews.


 25             However, the Paxil review did raise a




  1   question about data management, in that events


  2   suggestive of suicidality were coded under the


  3   general preferred term "emotional lability."


  4             This struck the reviewer as rather odd,


  5   and so in responding to GSK, we asked them to


  6   separate out the verbatim terms suggestive of


  7   suicidality under a term specific to suicidality.


  8             [Slide.]


  9             That request to GlaxoSmithKline resulted


 10   in additional work and ultimately resulted in a


 11   report on paroxetine and pediatric suicidality.


 12   That report went first to the MHRA -- that is FDA's


 13   counterpart in the UK -- and shortly thereafter to


 14   FDA in May of last year.


 15             That report indeed suggested an increased


 16   risk of suicidality associated with paroxetine use


 17   in particular in one of the three studies done in


 18   pediatric depression.


Wyeth wrote its own Warning letter to doctors, before the recent FDA Warning requirement, based on their own data, without further guidance by the FDA.  Yet, as pointed out in testimony in the 2/2004 hearing, this company had those data for years.  Did they just coincidentally get around to reviewing their own data just as the FDA began its review?  Here is a link to the table with those data.








Here are some other notes from the transcript review:

Conclusion by Dr. MOSHOLDER  (pg 223)

10             In conclusion, for the study of this issue


 11   of pediatric suicidal behavior associated with


 12   antidepressant treatment, the available


 13   pharmacoepidemiological data and postmarketing


 14   surveillance data is of limited utility, and


 15   randomized, controlled trial data should be


 16   superior to these sources.


Fundamental limitations in understanding these data:
"Because there are some exclusions in some trials of patients with some baseline suicidality, so the observed rates will not reflect what is going on in real life, and this might hamper our efforts in trying to investigate the risk because it will lead to underestimation" (pg 286, line 19)

Laughren reviews efficacy data in kids (pg 285, line 20)
Notes the FDA standard of two positive studies; fluoxetine has that, none of the rest do; but the latter is subject to interpretation, which he explains.  On the whole it's pretty suggestive these work for some kids, and that none is clearly better than the others

(291, line 1) He explains that these data are not "proof of the lack of benefit". 

Note the very interesting twist (289, starts line 5) where he explains that in the TCA days, the studies a company would have to submit where not set up to require "positive" studies -- and thus implies that the current set-up, so requiring, may warp the data in favor of efficacy.  Very damning of the process, if you read him close.