an addendum…

Posted on Tuesday 29 October 2013

Yesterday, the American Academy of Child and Adolescent Psychiatry wrapped up its 60th Annual Meeting in Orlando Florida. As you might recall, they published Efficacy of Paroxetine in the Treatment of Adolescent Major Depression: A Randomized, Controlled Trial in their journal in 2001, concluding that "Paroxetine is generally well tolerated and effective for major depression in adolescents." I understand that at their meeting last week, Paxil Study 329 was not open for discussion, just like last year when I filed a formal request for retraction. Appeals to the Authors, their sponsoring Universities, the editor of the Journal of the American Academy of Child and Adolescent Psychiatry, the Academy’s leadership and Ethics Committee, the NIH Office of Research Integrity, and GlaxoSmithKline have all been met in this same way. And yet Paxil Study 329 has been central in any number of successful  suits against GSK, including last year’s record $3 B settlement and has by now a rich literature of its own.

Ed Silverman of Pharmalot reports below on the progress of the team working on writing up Paxil Study 329 for publication as it should’ve been written up 12 years ago, part of the RIAT [Restoring Invisible and Abandoned Trials] Initiative. I’m a member of that RIAT team, and as Ed reports, we’ve hit an impasse trying to get the patient level data on Adverse Events that have been omitted from the publicly available study data:
by Ed Silverman

For the past year, GlaxoSmithKline has vowed to usher in a new era of transparency by creating a system to disclose detailed clinical trial data, a widely publicized move that has been hailed by many critics of the pharmaceutical industry who have accused drugmakers of deliberately concealing vital information that should be accessible to others in order to confirm safety and effectiveness. Now, though, a group of researchers is putting the drugmaker to the test by requesting detailed data for an infamous study of its Paxil antidepressant, but are squabbling with the drugmaker over information being sought. In the process, the dispute is raising questions about whether Glaxo complied with a 2004 consent order with the New York State Attorney General to publicly disclose the Paxil trial data.

At issue is data for Study 329. Glaxo participated in preparing, publishing and distributing what US authorities called a "misleading medical journal article" because the results reported that a Paxil clinical trial demonstrated efficacy in treating depression in patients under age 18, when the study actually failed to make the case. The trial missed its endpoints and also figured in a ghostwriting controversy [here is the study]. For more than a decade, the study has haunted Glaxo after it became known that suicide risks in what was one of the best-selling antidepressants had been minimized. The episode, which led Glaxo to the 2004 consent order, factored into the $3 billion settlement with the US Department of Justice last year for illegal promotion and pricing activities [more here]. For its part, Glaxo maintains it has fully complied with the order (more on that later).

However, a loose-knit group of researchers led by Jon Jureidini, a child psychiatrist at the Women’s and Children’s Hospital in Adelaide, Australia, and a professor in psychiatry and pediatrics at the University of Adelaide, is haggling with Glaxo over certain data referred to in the consent order and they maintain the drugmaker has balked at their request and raised questions about its commitment to releasing data. The tussle underscores the ongoing tension over the extent to which certain data – notably, anonymous patient-level data – can or should be released. Last week, a group that includes academics and consultants, some of whom are affiliated with the Multi-Regional Clinical Trials Center at Harvard University, suggested an independent panel should be created [see this]. And a recent essay in The Lancet, Glaxo R&D chief Patrick Vallance and Iain Chalmers, one of the founders of the Cochrane Collaboration, argue that open access to this detailed data poses many risks…
Silverman goes on to flesh out the issues involved in the current conflicts in that way he does better than any one around. I’ll leave it for you to read. My comments today aren’t about those specifics and Ed’s better at it.

Looking back, my first encounter with Paxil Study 329 was in December 2010 when Paul Thacker, then at POGO, exposed the ghost-writing of Sally Laden, the actual author of the article [see roaches…]. And over the last three years, the article has been the gift that keeps on giving, teaching us about the methods used to distort industry sponsored clinical trial reports in our literature. There’s no question about the article. The overall conclusion isn’t correct ["Paroxetine is generally well tolerated and effective for major depression in adolescents"]. More important, the study itself doesn’t support that conclusion – obvious to the sponsors in this 1998 internal GSK Memo as soon as the blind was broken:
There’s no shame in being wrong. The scholarly literature is filled with wrong conclusions. The shame is that GSK, the Authors, and the Journal’s Editor were all made aware that the study didn’t support the conclusions and published it anyway. And thanks to a library of subpoenaed documents, there’s too much known about the process and events that went into this publication to even support the usual excuse of plausible deniability. So there are two reasons behind the move to republish this study as it should’ve been reported twelve years ago:
  • The first is obvious – to correct the academic literature. This article shouldn’t be there. If they won’t retract it or correct it, then the only choice is to add a corrected version of the report on the clinical trial.
  • Second, to make the statement to future sponsors, authors, academic institutions, and journal editors that if they engage in this kind of scientific misbehavior, whatever their motive, they should be prepared for a public statement of what they did as a potential consequence.
The RIAT Initiative is a forgiving proposal. Once an article is identified as defective or missing in action, those responsible are offered the chance to address the problem themselves. A RIAT team is only assembled if they refuse, and has to jump through the same hoops as any submission to an academic journal with the added burden of gathering and working with someone else’s data.

The later part of the Pharmalot post is about the RIAT team’s request for the full patient-level data from Paxil Study 329, particularly from the Adverse Effects – reporting that is omitted in the online data [Appendix H]. Ed tells that story accurately and I’ll defer to his version. But there’s one piece to add to his report:

Several things happened in 2004. Eliot Spitzer’s settlement in New York [from Pharmalot]:
For instance, page 7 of the order says that, beginning in February 2005, Glaxo must provide public access to the clinical study reports of Paxil as a treatment for major depressive disorder in the pediatric population for a 10-year period [here is the consent order]. An attached document defines clinical study report as including all data. And data is defined as all the results and outcome measurements obtained from a clinical study [here is the document].
But there was also a request from the FDA that GSK reevaluate their data looking at suicidality – the part that ultimately lead to the Black Box Warning added to the antidepressants. GSK reported the results of that reevaluation in 2006 along with publishing the two unpublished negative studies [see timeline above – studies 701 and 377]:
by Apter A, Lipschitz A, Fong R, Carpenter DJ, Krulewicz S, Davies JT, Wilkinson C, Perera P, and Metz A.
Journal of Child and Adolescent Psychopharmacology. 2006 16[1-2]:77-90.

CONCLUSIONS: Adolescents treated with paroxetine showed an increased risk of suicide-related events. Suicidality rating scales did not show this risk difference. The presence of uncontrolled suicide risk factors, the relatively low incidence of these events, and their predominance in adolescents with MDD make it difficult to identify a single cause for suicidality in these pediatric patients.
It’s obvious from that report that the Adverse Effects information in the electronic database used for the JAACAP article was incorrect. And it is only the electronic database print-outs that are posted on GSK’s Website. The patient level information in Appendix H is totally missing rather than redacted. So there is a concrete reason for wanting to look at the patient level data that was clearly available to the authors of the 2006 GSK article. And any reading of the 2004 consent order can’t miss the words all data. The point of the RIAT Initiative is to republish the study as it should’ve been published originally. The point of the consent order GSK signed in 2004 is that they agreed to post "all" of the data. So we are simply asking for information that is known to be pertinent [from the 2006 GSK-written publication] and should be available under the 2004 court order. Why it has been reframed into a debate about GSK’s newly proposed program for data transparency is open for speculation and interpretation, but well off the actual point…
    October 30, 2013 | 2:43 PM

    You’re a white knight in the battle for patient safety and accountability for drug makers! Thank you.

Sorry, the comment form is closed at this time.