Correspondence With GSK

2013-06-13 – Publication of RIAT Declaration

Riat Cover image

 Editorial |Full article | Print edition

2013-04-26 — Dr. Jureidini writes to GSK CEO Sir Andrew Witty. The letter states:  I write to you as the CEO of GlaxoSmithKline in regard to an on-going complaint about a fraudulent journal article under the lead authorship of Martin Keller… In light of a recent $3 billion settlement in which your corporation pleaded guilty to misbranding of paroxetine (Paxil), we request that you write to Dr. Andrés Martin, the editor of Journal of the American Academy of Child & Adolescent Psychiatry to request retraction of the Keller et al. article.

2013-05-03 — Dr. John Kraus responds to Dr. Jureidini on behalf of Andrew Witty.  He declines to request retraction, and refutes the suggestion that the reported findings were fraudulent.

2013-06-14 — The RIAT team sends an email to GSK, Sir Andrew Witty (CEO) and Patrick Vallance (President  of Pharmaceutical R&D), notifying them of the RIAT article publication and requesting that if they plan to restore any old GSK trials, they respond as soon as possible.

2013-06-30 — RIAT to GSK


2013-09-04 — Jureidini to Witty: Dr Jon Jureidini (JJ), requests access to Study 329 data, reminding Sir Andrew Witty of the GSK commitment under the 2004 consent order of the New York State Attorney General’s office to make Study 329 data public.

2013-09-06 — Shannon to Jureidini: Reply by James Shannon (JS), GSK Chief Medical Officer, on behalf of Andrew Witty, enumerating the information available  re: Study 329 on the GSK website, and explaining that Dr Jureidini could submit a research proposal and a GSK “Independent Review Panel” would consider making anonymized raw datasets available to his research team.

2013-09-30 — Jureidini to Witty: JJ confirms that the on-line request has been submitted and requests that the GSK Data Access System make the CasevReport Forms (CRFs), which are the individual detailed patient records, available to the team  with personal identifiers removed. He also objects to having the data only available in the form of scanned images, which not searchable or otherwise amenable to compilation or other manipulation.     

2013-10-12 — Shannon to Jureidini:  JS responds advising Dr Jureidini that GSK does not make all CRFs available to researchers. GSK will provide Case Report Forms, which include CRF information only for patients experiencing serious adverse events. Of course, if GSK decides that an adverse event in its Paxil group was not attributable to the drug, no CRF would be part of the CSR, rendering the CSR useless for the researchers’ purposes. GSK notes that its CRF data is available for audit by the FDA. One wonders if the FDA auditors are forced to look at scanned images, taking months to sort through data which could analyzed.

2013-10-27 — Data Sharing Agreement  GSK standard research agreement

2013-10-28 — Data Application JJ applies for access to the 329 data on behalf of the RIAT team

2013-10-28 — GSK Confirmation of Application Receipt    GSK Confirmation of Application Receipt:  GSK acknowledges the receipt of the 329 team application,          saying that no changes can be made after this date but more information may be requested.

2013-10-29 — Jureidini to Shannon: JJ requests access to the CRF data noting that the only personal identifiers are initials which can be redacted in minutes, that     the 329 team has no interest in identifying patients, and that without access to the CRFs the obvious errors coding the adverse  events in the original study cannot be corrected by the research team.  He also enquires regarding follow-up of patients who had enrolled in 329. JS asks if “those who became suicidal or violent on Paxil” were “subsequently advised of the role of the drug in their dangerous and distressing feelings”.

2013-11-05 — Shannon to J Jureidini:  JS responds that the request has been sent to the independent review panel, and reiterates that it is not GSK policy to make CRFs available.  However, he suggests a phone call to discuss the issue.

2013-11-08 — Shannon to Jureidini:  JJ thanks JS for his assistance and states that he prefers written communication to a phone conversation.  He asks about the best way to transfer all CRFs, de-identified, with personal narratives intact.

2013-11-19 — Shannon to Jureidini:   JS proposes providing anonymized CRFs through the same method as the CSR.

2013-12-10 — Jureidini to Shannon:   JJ confirms that he does want access to the data, and suggests that the agreement should be interpreted to allow corrections of the original data to be recorded in the spreadsheet and deposited with BMJ. He reminds Dr Shannon that he has not answered the question regarding follow-up of 329 participants.

2013-12-12 — Shannon to Jureidini:   JS writes that he was awaiting confirmation before asking his staff to proceed with drafting a data sharing agreement. He states that follow-up of subjects is the responsibility of the Investigator and treating physician and has nothing to do with GSK. He asks about the discrepancies in the adverse event tables references in Dr Jureidini’s Oct 29 letter. He again suggests a phone call.

2014-01-06 — Jureidini to Shannon:  JJ states that since JS has not contradicted his assumptions about constraints on access to data and publication that this signals  agreement. He notes that he prefers written communication so that there is a written record. He asks about the basis of the JS assumption that patient follow-up is entirely the responsibility of the Investigator and the treating physician, and inquires whether GSK ever provides guidance in this regard.

2014-01-20 — Shannon to Jureidini:  JS assures JJ that the data sharing agreement is standard wording intended only to protect patient confidentiality. He also cites a reference for his contention that it is the Investigator’s primary responsibility to follow-up study participants. He once again offers to talk on the phone or face-to-face, “to ensure that (the Study 329 RIAT Team) understands and is able to effectively navigate the datasets”. He disagrees with JJ’s contention that GSK should have recognized the clear indications that Paxil produces suicide-related adverse events, claiming that:  “A statistically significant difference in suicide-related adverse events… emerged only when adverse event data was interrogated further and data from individual trials were pooled together.”

2014-01-21 — Data Sharing Agreement (Proposal 669)   A copy of the data-sharing agreement is sent to JJ, signed by Russell Brooks, VP Legal Operations for GSK.

2014-02-11 — Jureidini to Shannon:  JJ continues to challenge JS’s statements regarding follow-up, noting that individual treating physicians would have no knowledge that Paxil can cause suicidality, and that the evidence of this side effect is apparent from a simple scan of patient narratives. He is not buying the contention that GSK did not know about the suicidality and repeats his belief that GSK has some follow-up responsibility to the Study participants in this regard.

2014-03-11 — Shannon to Jureidini:  JS acknowledges only JJ’s point that paroxetine increases suicidality without acknowledging responsibility for follow-up. He advises that the Study 329 RIAT Team has been granted access to the CRF data, and that 250 records will soon be available with  access to the rest to follow shortly after.

2014-04-21 — Jureidini to Shannon:  JJ continues to press for answers to questions he considers paramount:  Why JS would use statistical significance as the only indicator of a problem, why other methods besides RCT were not used, and why participants did not receive the benefit of a warning that any symptoms they may have experienced could have been caused by Paxil.

2014-05-29 — Shannon to Jureidini: JS sidesteps the questions raised by JJ, saying:  “I hope you agree that randomized controlled trials provide the most reliable evidence to determine the efficacy and safety of a treatment.”

2014-09-09 — Jureidini to Shannon: JJ sends JS a copy of the article submitted to BMJ for publication in early September.

2014-09-24 — Shannon to Jureidini: JS provides some suggestions for improving the RIAT paper:

  1. He repeats the GSK position that the original study was not false, fraudulent, misleading or ghostwritten and wants this GSK view represented in the paper.
  2. He wants the 329 new findings compared to other studies that have reanalyzed safety data, specifically:
    • Suicidality in pediatric patients treated with antidepressant drugs, Archives of General Psychiatry, 2006, Hammed, Laughren and Racoosin. This was a meta-analysis of trials submitted to the FDA re: SSRI impacts on paediatric patients.  Thus, the studies excluded all trials showing negative results or they would never have been submitted.  Even with this bias, the study still found a “modestly increased risk of suicidality”, which apparently had escaped the prior notice of FDA officials.
    • Evaluation of Suicidal Thoughts and Behaviors in Children and Adolescents Taking Paroxetine, Journal of Child and Adolescent Psychopharmacology, 2006. Alan Apter et al found that: “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.” In other words, the authors imply that it was the depression, not the paroxetine, that probably caused the increased suicidality. The standard mantra of pharma re: SSRI suicidality.
  3. He implies that it is a problem that the CRFs were reviewed without the reviewers being blinded to the study drug.
  4. He complains about the new investigators overriding the judgement of the original investigators in coding.
  5. He complains that the reviewers did not review 100% of the CSRs. (While this is true, it was GSK that made 100% review all but impossible by providing the data in the least user-friendly form possible.)
  6. Citing lack of comparability of adverse events between the new and old studies, he expresses concern that readers might be misled.

2014-10-14 — Jureidini to Shannon (re BMJ review):  JJ does not address the JS comments, noting only that there are “few points of agreement”. He asks, in the original study, whether the HAM-D depression item was designated as an outcome measure prior to breaking the blind.

2014-10-27   Kraus to Jureidini:  Response letter from John Kraus, GSK VP of medicines development (JK), on behalf of JS, who is out of the office. He recommends that variables of interest to the Study 329 RIAT group should be analyzed in accordance with the original study protocol. He suggests that pairwise comparisons are consistent with the protocol.

2014-10-31 — Jureidini to Kraus:  JJ explains that pairwise comparisons would be meaningless. He clarifies that he is not looking for assurance that HAM- D was designated prior to breaking the blind, he is looking for evidence that this was the case, noting “unless the analytical plan is produced, we can only assume that it does not exist.”

2014-11-08 — Email Kraus to Jureidini:  JK attaches an August internal request for an SAS analysis of HAM-D before the blind was broken in October, incorrectly implying that this proves that there was an analytical plan that included this as a planned measure. Presumably if there was an analytical plan with this included, GSK could have produced it. See attachment

2014-11-13 — Letter to Dr Kraus:  JJ writes to JK, once again requesting the analytical plan.

2014-11-20 — Email Kraus to Jureidini:    JK responds by email and restates his position that the file note he already sent ought to be good enough.