Ghostly Data

There are four strands to this section.

First Strand
This follows the question of ghost writing of scientific articles. Ghost writing has probably always happened to some extent. In the course of the 1980s with the vast increase in the number of company sponsored satellite symposia and the emergence of a journal supplement business, ghost writing almost certainly became commoner.

As pharmaceutical corporations reorganised themselves in the 1970s and 1980s, they outsourced their medical writing into medical writing agencies. This has the advantage for companies of being able to get agencies to compete against each other for business, which may come in the form of a contract for 10 or 20 or 30 articles. Agencies can potentially compete in areas of cost, but also in terms of access to the best journals or in terms of attracting a greater proportion of academic stars as ‘authors’.

Ghost writing has now evolved beyond journal supplements and there are grounds to think that a significant proportion of the articles on therapeutics in main stream journals such as JAMA, The New England Journal of Medicine, the BMJ and the Lancet may be written with help from medical writing agencies.

In the course of Motus v Pfizer, the lawyers for the plaintiffs, Baum, Hedlund, Aristei, Guilford and Schiavo turned up a document produced by Current Medical Directions (CMD), a medical writing agency based in New York, who appeared to have been working for Pfizer toward the end of 1998.

This document lists a portfolio of articles on Zoloft (Sertraline), some of which had already been published and some were still being written. This document offered us the opportunity to analyse what proportion the therapeutic literature might be written in a manner that could be described as non traditional.

We have attempted to do this by seeking out all articles on sertraline and comparing the previous number of articles published by all authors in both the CMD and the non CMD series as well as the journal impact factors in which CMD and non CMD articles had been published and finally looking at the citation rates of these articles after publication.

These efforts led to a paper -‘Authorships and Impacts…’ was first sent to the Journal of the American Medical Association because of their prior interest in this area. This paper received a series of mixed reviews.

The paper was modified in the light of those reviews and sent to the British Journal of Psychiatry, where it was reviewed by at least 7 people before publication.

The background figures are presented – a list of the impact factors for CMD journals and authors, citation rates for CMD articles, as well as a list of non CMD articles with literature profiles.

Second Strand
This second strand in the web site contains a section on missing data.

In September of 1991 Kahn et al in the American Journal of Psychiatry published an article giving suicides and suicidal acts in clinical trials of new antipsychotics submitted to the FDA for licensing purposes. This article lists suicides and suicidal acts on both new and older antipsychotics as well as placebo. See Table 1. Aside from a high rate of suicides and suicidal acts on new antipsychotics compared to placebo, the interest in this paper lies in the set of missing figures for suicidal acts for Lilly’s antipsychotic Zyprexa and Astra Zeneca’s Seroquel.

Writing to Lilly and Astra Zeneca brought different responses. Astra Zeneca supplied the missing data for suicidal acts on Seroquel. Lilly did not. Scrutinising the material that comes from an FDA freedom of information request does not give any pointers as to the numbers of suicidal acts, but it does to show that Zyprexa has the highest figure for number of suicides in a clinical trial programme of any current psychotropic drug. It appears that the placebo suicide listed in Khan for the Zyprexa trials may not have occurred during treatment with placebo but may in fact have occurred during the wash-out phase of the trial.

This lack of data led to the following correspondence with the Minister for Health in the United Kingdom. Ministers for Health worldwide have no powers to force a pharmaceutical company to make data public. However they do have some responsibility for ensuring that patients receiving treatment are able to give informed consent.

This issue remains unresolved nearly two years later, despite written confirmation from the Department of Health that it needs resolution.

This point links up to the issue of SSRIs and suicidal behaviour, particularly in children. In the mid-1990s, Pfizer conducted studies in children with obsessive compulsive and depressive disorders. The articles resulting from these studies are included on the CMD list of articles. Of the six articles listed here, only one makes a clear reference to suicidal behaviour in one child.

In fact as an Expert Report prepared for Pfizer on this data reveals there had been at least six suicidal acts in children taking sertraline.

Studies with Glaxo SmithKline’s Paxil in depressed children reveal a comparably high rate of suicidal acts on it compared with competitor antidepressants and placebo (Keller M et al J American Academy Child & Adolescent Psychiatry 2001, 40, 762-772). In this case the company defence appears to be that the suicidal behaviour was not attributed by the investigator to the drug.

However when investigators do attribute causality to the drug in the case of an adverse event, Glaxo SmithKline and other companies say that neither these investigators nor even their own company personnel can attribute causality in this way. That causality can only be determined by Randomised Controlled Trials. Appling this logic and sticking strictly to the figures would appear to indicate that whether or not the investigators believed that Zoloft and Paxil can make children suicidal, the figures indicate that they can and possibly do so at a higher rate than occurs in adults.

Strand Three
In addition to the possible ghost authoring of articles (strand 1) and the omission of important safety data (strand 2) there is a further way in which companies shape what gets published or what doesn’t get published.

In April 2000 Joseph Glenmullen’s Prozac Backlash was published. Shortly after publication a variety of media outlets received unsolicited reviews of this book. Attached are a set of reviews sent to Newsday in New York with a covering letter from Chamberlain Communications Group and a further set sent to the Boston Globe from Rasky Baerlein. These reviews from senior figures in US psychopharmacology echo themes that were first developed within Lilly in the early 1990s – that Prozac is one of the most researched drugs ever, that depression leads to suicide, and that the people likely to suffer following Prozac Backlash are the depressed people who scared off treatment go on to kill themselves.

A systematic campaign like this will deter most people from entering any debate on the possible hazards of antidepressants.

Strand Four
This section gives concrete instances, where it can be certain that some ghost-writing happened.

A. In 1999 I was approached by the pharmaceutical company, Pierre Fabre, to participate in a satellite symposium for their antidepressant Milnacipran at a European College of Neuropsychopharmacology meeting in London that September. This led to a series of e-mails. One of these came with a ghost written Healy article called “Bridging the Gap”. This was subsequently published with the apparent author as Siegfried Kasper. It appears in a journal supplement. The manuscript I wrote for the symposium ‘Antidepressant pharmacotherapy at the crossroads’, appeared in the same journal supplement. See Kasper IJP and Healy IJP.

B. In the course of 2000 while I was waiting to move to a position in the University of Toronto, I had overtures from Sid Kennedy about involvement in a meeting that had been organised by Wyeth Canada by CMED, a medical communications group based in Toronto. This was originally scheduled fro Cancun in Mexico but was later switched to Laguna Beach in California.

For a variety of reasons I was reluctant to get involved.

While visiting Toronto as participant in symposium celebrating the 75th anniversary of the University Department, in between interviewing colleagues for positions on the mood and anxiety disorders programme in the University and other business, Sid Kennedy organised for me to meet Jack Fava from CMED, who attempted to overcome my reluctance at being involved. Fava was happy for someone to deputise for me - Richard Tranter – and was offering to write our article for us..

After returning home and after finding out about the breach of my contract with the university of Toronto, I approached Richard as to his willingness to get involved, explaining that a manuscript would be written for us and that we were going to alter this as part of an experiment.

This leads into a sequence of e-mails from Jack Fava and Pauling Lavigne, which are filed under CMED-emails. The process also led to a series of articles from CMED-original to CMED-revised and CMED-final.

The meeting in Laguna Beach centred on an article written by Michael Thase and two Wyeth authors, which appeared in the British Journal of Psychiatry in 2001. the central claim behind this paper has also featured in advertisements that Wyeth had been running for Efexor, which was that unlike other antidepressants, which might get you better, Efexor got you well. This British Journal of Psychiatry paper featured prominently in an editorial that subsequently appeared in The Lancet under the title “How Tainted has Medicine become?”

Our CMED-revised changed CMED-original in two ways. The two changes are near the end of the paper. First it introduced some data from clinical trials on Remeron that were inconvenient for Wyeth’s claims about Efexor. Remeron (Mirtazapine) was one of the compounds Wyeth was most concerned about as it had in some respects a comparable mechanism of action to Efexor.

The second change involved references to the fact that if an antidepressant did not suit a patient that not only would it neither get you well not even better but could in fact make you worse to the point of making you suicidal.

Following CMED-revised the e-mail trace makes it clear that there was an objection from CMED. The objection was to our introduction of the material on Remeron.

The next I got to see of this article was after CMED-final had been submitted to the Journal of Psychiatry and Neuroscience. CMED-final has removed the reference to suicidality and inserted a reference to treating anybody who fails to respond with venlafaxine – see letter June 12th.

This led to a series of e-mails which can be found in the CMED-email series and letters on July 27th, July 30th, August 13th and August 30th (Letters).

There was also one phone call, the only phone call I received from Sid Kennedy following my breach of contract with the University of Toronto. In this call, Dr Kennedy proposed that I remove my name from the article given the difficulties I was having and instead write a commentary, which would reflect some of my concerns. For a variety of reasons, this proposal had its attractions. This commentary finally emerged. – see file JpsychNeuro.

In the meantime, the previous article was reviewed and revised on the basis of these reviews. It was subsequently published in 2002 and is listed there as having been apparently first submitted on September 10th 2001, and re-submitted following revision on May 15th and accepted on May 24th 2002. This article appears as CMED 2002.

Meanwhile my commentary was reviewed and the review suggested that this piece, which picked up on the issues of possible suicidality, did not handle this issue in sufficient depth. This led to correspondence with the editor of the Journal of Psychiatry and Neuroscience, Russell Joffe, in which a proposal for appoint counterpoint feature was raised and accepted. See JPN-emails.

This led to a paper on SSRIs and the risk of suicidality – JPCND Healy. The issues stalled for close to a year. Finally a counterpoint feature written by Yvon Lapierre appeared (JPN-YLP). Then both point (JPCNDHealy2) and counterpoint were sent for review by the Journal. Correspondence from the Journal had indicated that point and counterpoint would appear within weeks but this was followed up then by a letter fro April 9th from Dr Joffe indicating problems. I responded on April 16th.

Also included in this section is the Lancet Editorial on Tainted Medicine and Point-Counterpoint feature on the question of “Is Academic Psychiatry for Sale?” involving Michael Thase.

This series of articles meets the BMJ-Lancet article in a Lines of Evidence article published in Psychotherapy and Psychosomatics.