Back in 2001, an infamous study was published in the Journal of the American Academy of Child and Adolescent Psychiatry that declared Glaxo's Paxil antidepressant - called Seroxat in the UK - was "generally well tolerated and effective for major depression in adolescents." Known as study 329, the findings were used to widely promote the drug, which became a huge seller.
Of course, the study was later held in disrepute after it was learned the results didn't tell the whole story. In fact, 329 was one of three studies cited by former New York Attorney General Eliot Spitzer, who filed a suit charging Glaxo with "repeated and persistent fraud,” alleging the drugmaker had promoted positive findings, but hadn’t publicized unfavorable data (back story).
As it turns out, study 329, which already had a sordid history that included ghostwriting charges (here's some background), were worse than imagined. A new study in the International Journal of Risk & Safety of Medicine discloses that, after sifting through some 10,000 documents that surfaced during Paxil litigation, highly selective reporting was used to skew the results favorably.
The latest study prompted a complaint concerning possible scientific misconduct to be leveled against study 329's lead author, Martin Keller, a psychiatry professor at Brown University, by David Egilman, a clinical associate professor in community health at Brown. We are awaiting a reply from Keller and Brown provost David Kertzer, who was sent the complaint by Egilman. UPDATE: A Brown spokesman declines to comment.
Back to the new study. A key finding about 329: "There was no significant difference between the paroxetine and placebo groups on any of the eight pre-specified outcome measures. However, by the time the data were analysed, many other new measures had been added to the list of secondary outcomes."
For instance, along the way, "four of the eight negative outcome measures specified in the protocol were replaced with positive ones." In fact, the study finds that in 329 "...disclosures of overdose and mania were edited out, and serious adverse events on (Paxil) were attributed to other causes." And the original authors of 329 didn't conduct any analysis to determine whether side effects occurred more commonly on Paxil compared with a placebo.
By their calculations, though, the researchers write that serious adverse events in 329 were significantly more likely to occur in patients taking Paxil (12 percent) vs. placebo (2 percent). Adverse events requiring hospitalization were 6.5 percent on Paxil vs. none on placebo. Severe nervous system side effects - 18 percent on Paxil vs. 4.6 percent on placebo.
"Our detailed case study of proprietary documents from GSK regarding this study adds to the evidence that flaws in industry-funded research can be severe and difficult to detect," the researchers write. "The documents reveal that the published conclusions of study 329 and information provided by Glaxo to health professionals understated adverse effect rates and emphasised post-hoc measures that were not consistent with the unpublished, protocol-defined primary and secondary outcomes."
[We should note that the latest study points out that the lead author was contacted by plaintiffs' attorneys to conduct a review of Paxil data as part of their litigation against Glaxo. This is noted in the study, by the way].