Editorial Note: This post has been put together by Peter Goetzsche and David Healy
Dear Dr Goetzsche
…
“At GSK we firmly believe that making more information available, including clinical study reports and anonymised patient-level data, will enable researchers to study the science behind today’s medicines more closely, to learn more about them and how they can best be used.
For GSK the transparency and access journey [i] started with online clinical trial studies in 2004 and has so far led us to the endorsement of the AllTrials Campaign last year, as well as access to anonymised patient data launched in May 2013. Ultimately this has the potential to improve patient care and drive open innovation.
When patients volunteer to take part in research they have a legitimate expectation that the data will be used to enhance knowledge and improve patient care. Our actions towards transparency are an essential part of meeting those expectations. By being open and accessible we become an even stronger and more trusted partner for society, patients and health care professionals and at the same time improve patient care.
We have been open in describing the steps we have taken. Our initiatives has been acknowledged and welcomed by governments, NGO’s and research institutions.
The recently published report from the Science & Technology Committee in UK, states regarding patient level data: “We recognise the efforts of some members of the pharmaceutical industry, particularly GSK, to increase clinical trial transparency and hope that other companies will act in the same spirit in implementing industry-wide principles for responsible clinical trial data sharing”.[ii]
Our initiatives have been commended by campaign groups as setting a new standard for greater transparency. In response to our commitment to develop a system for access to patient level data, Ben Goldacre, (author of “Bad Pharma” and co-founder of AllTrials) said: “It is a great initiative; I call for it in the book, I hope it happens at GSK, consistently, and I hope it is copied”.[iii]
Similarly, in response to GSK signing up to AllTrials, where the Cochrane Institute is one of the partners, and committing to post CSRs, Tracy Brown from Sense about Science said: “GSK signing up to the campaign is very important, off course, because they are a large global player in clinical research so they have a lot of potentially useful information to share, but also because they are finding a way to put in place the infrastructure needed to do this. Which makes it realistic for others and sets a new standard”.[iv]In a BMJ article published March 2013, CEO Andrew Witty explained our commitment to transparency[v] and our global position is also reflected in the recent submission of our views as part of the consultation process with respect to EMA’s ‘Policy 0070 on publication and access to clinical-trial data’.
We regret that you find a personal meeting irrelevant and would like to repeat our invitation. In GSK we find dialogue very valuable and ultimately to the benefit of patients, and thus our dialogue should be continued in a face to face meeting either in London or in Copenhagen.
A GSK & AllTrials coalition?
Here’s that Branded tone again – mentioned in SHIT Happens. A plaintiff note about how the world just doesn’t understand we are the good guys. GlaxoSmithKline (GSK) seem to have bought their own propaganda about being champions of transparency.
As they mention, the House of Commons has rolled over and put its paws in the air so pleased are they – despite being told by GSK that the UK is not a fit place for Pharma to run trials in. No reference here to the recent fines in the USA for $3 Billion against GSK or the fraud action by New York State against GSK in 2004 hinted at above as the kickstart to GSK’s journey to transparency and access.
Worryingly, there almost appears to be a hook-up between AllTrials and GSK. AllTrials have been campaigning to have all trials registered. When GSK endorsed this campaign, Ben Goldacre described it as a “cartwheel moment”.
In fact the idea of registering all trials – and in this sense the start of the journey – is one that Glaxo Wellcome endorsed 15 years ago when Richard Sykes was CEO of the company. GSK then drew back from this commitment, after Glaxo Wellcome merged with SmithKlineBeecham.
Getting closer?
The links between GSK and AllTrials have appeared to get even closer recently with Iain Chalmers of AllTrials co-authoring an editorial with Patrick Vaillance of GSK. (The Attitude of Chicks to Trojans and Horses). This editorial pushes trial registration. But it also endorses GSK’s proposal of only releasing some of the material from those trials and then only releasing data in response to a request that contains an analytic plan and even in response to such a request only offering a form of release that actively blocks investigators from having a clear view of what might actually have gone on in the trial.
Transparency GSK-style, like Gangham-style, seems to have gone viral. It went viral went it was a press release with not a shred of policy detail in sight – Won’t get Fooled Again. A case study perhaps for those who think that to get something to go viral needs Jupiter to be in alignment with a cat’s eye in Southern Africa when the wind blows East.
Everyone wanted to hear this message.
The Chalmers-Vaillance editorial justified GSK-style in part by making a case that patient confidentiality was extremely important and that the risks of identification were great. This is a strategy Pharma have deployed from the start of the current Data Access Debate – using the patient group Eurordis to make the case for them (The Data Access Wars).
It flies in the face of what the overwhelming majority of people signing consent forms probably intend – which is to make their data available for scrutiny by independent experts. If those of us who have been participants in trials thought some remote risk of a breach of privacy were being used to prevent a disclosure of details that would save someone else’s life but threaten GSK’s profits, most of us would likely be horrified.
It also flies in the face of the original use of consent forms in drug trials – which was solely to establish that you knew you were taking a novel, not-yet-marketed compound. See When Does Yes Mean No.
The moral high ground?
GSK are claiming the moral high ground of concern for patient welfare. Have they in the process managed to co-opt groups from the House of Commons to AllTrials who might otherwise hold them to account?
Study 329 is GSK’s most famous clinical trial. In this a large number of children became suicidal on paroxetine. Antidepressants like paroxetine – thanks to Study 329 but not thanks to GSK – now come with a Black Box Warning that they may cause suicide.
Study 329 was worth roughly $1 Billion to GSK. It enabled the company to get six months patent extension on Paxil-Seroxat.
The children who became suicidal on Paxil in Study 329 are more likely than others to become suicidal again if exposed to another SSRI.
Against this background
- What should a company genuinely concerned about the welfare of patients who have done so much for the welfare of the company (participating in this trial without payment) do at this point?
- Should GSK’s concern about patient welfare extend to informing the subjects in Study 329 of the link between their treatment and their problems and future risks they may run?
- Andrew – have you contemplated the possibility that if you don’t inform those affected you are in fact still to this day making suicide more likely for some of those affected?
- Has GSK – or any pharmaceutical company – ever gone back to any of the subjects enrolled in any of its trials to explain to them what might have happened to them and the role their drug might have played in what happened? Please send known examples.
- Would explaining to people how their injury arose not be a more telling indicator of concern for patients than an effort to keep identifiers out of the public domain?
- What would the patients who have participated in clinical trials think was the more important indicator of genuine company concern?
- What do Flaminia Macchia and Eurordis think would be the most appropriate indicator of genuine company concern?
Confidentiality
Following a legal action taken by New York State in 2004, GSK agreed to post Clinical Study Reports (CSRs) for their pediatric Paxil trials and other trials on the company website along with further details of their Avandia and other trials on the company website.
Any good investigative journalist could likely identify each of the children who became suicidal on paroxetine from the CSRs available on GSK’s website. Pretty well the only additional details that the patient level data that the company refuses to release contains are the patients’ initials. These could easily be redacted. This raises the question as to why GSK are so resistant to making the original case report forms (CRFs) available?
The only obvious other details that the CRFs contain not found in the CSRs are adverse effects. Its clear from looking at the CSRs, there are a very large number of mismatches (several hundred) between what the CSRs show and the CRFs are likely to contain.
Testing GSK
There is a team attempting to rewrite Study 329 according to the RIAT process. See Reading the RIAT Act. The difficulties they are having in getting data out of GSK have been covered by Ed Silverman of Pharmalot. [No longer at http://www.pharmalive.com/is-glaxo-keeping-its-commitment-to-releasing-paxil-trial-data].
1boringoldman has also covered this ground and the apparent growing links between GSK and AllTrials, and how GSK are using their model of transparency to block access to the data.
Because there is a mismatch between the CSRs and CRFs, the Study 329 RIATers are at a point where they require the raw data – the CRFs. GSK refuses to hand over the CRFs.
What would the now 35-year-old or so children who participated in Study 329 think should happen at this point?
Coalition politics
If you are a small player in the scheme of things, you may know just how the world should work but will never get a chance to implement any of your ideas. If you go into coalition with a bigger power you may get to implement policies you have always supported but have no chance to see brought into being. Or perhaps you can temper policies your partners might want to put in place that you disagree with.
What do AllTrials think should happen at this point?
David, is the team seeking to rewrite Study 329 the same as that which published the original study “Clinical trials and drug promotion: Selective reporting of study 329”?
I am not familiar with all the ugly details of this story, but as a scientist my skepticism is raised when I see an article on an extremely important public health issue published in a journal that I’ve never heard of, which is not indexed in Pubmed, which has an impact factor of 0.15, and which is apparently not peer reviewed. The first explanation that comes to mind is that the article was not able to pass review in a peer-reviewed publication, which does not in itself render in unreliable, but certainly raises some flags.
Just as the industry publishes papers that present only the side of the data that it wants people to see, the literature is full of papers by critics that show only the data that supports their position, which is taken out of context, or distorted (Ben Goldacre is a particularly notable offender). On pharmalot, I critiqued a recent BMJ paper that criticized industry for non-publication of 30% of large trials completed prior to 2009. The authors stated that the size of the trials eliminated the possibility that the trials were “low interest”, and that the only possible explanation for their not being published was to hide negative data. But inspection of the first 20 unpublished trials in their list (which took me only 20 minutes) showed that most of these trials were of drugs that were not approved, and which were never marketed for any indication. My personal feeling is that BMJ would never have permitted authors to make such broad statements with such a shallow level of analysis if the conclusions had not been in support of the journal’s own position.
Your thoughts?
John,
As a co-author on the publication you mention that was published in the International Journal of Risk & Safety in Medicine (IJRSM), I can assure you this critique of the Keller et al report of study 329 was peer reviewed. The article was originally comissioned by BMJ, but when submitted for publication, the BMJ lawyers decided that it was a risk for libel in the UK. We therefore sought publication outside of the UK.
It is very difficult to publish articles critical of industry-sponsored studies found to be guilty of scientific misconduct because many medical journals are heavily dependent on industry advertising and re-print orders, many for-profit journals exist to serve industry and finally because lawyers defending industry threaten libel actions. I think BMJ and PLos Medicine are exceptions but since they are published in England, they still face threats of libel actions. GSK has successfully blocked another publication by Jureidini and myself in the Lancet.
The team rewriting study 329 is not the same as the co-authors on the Selective Reporting paper published in IJRSM, except for the lead author, Jon Jureidini.
Leemon McHenry
Thank you. I appreciate your thoughtful and detailed response.
JT
David, I apologize, my statement above that the journal I refer to in the bottom paragraph above is indeed indexed in Pubmed. This was an error on my part.
But to follow-up on the two-way transparency issue, I thought it interesting that Dr. Juredini in the letter to the editor found here (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1939762/#!po=100.000) references his own, non-peer reviewed blog as a data source for his assertion that 19/20 antidepressant trials are unpublished. If this is true, why does he not publish the result in a journal that would subject his findings to critical review? Certainly the BMJ and other medical journals have published articles critical of the industry in the past.
I also found it noteworthy that when one goes to the site referenced by Dr. Jureidini, only members are allowed to post comments. And the criteria for membership include “Sharing our aims”, which a skeptic might interpret as “sharing our beliefs”.
So I guess I’m trying to drive home a point here, which is that transparency is a two way street. I appreciate your own history of openness in this regard.
I think that irrespective of other motivations and having things to hide, this sort of behavior would make many reluctant to share primary data with any and all comers.
John
Thank you your comments. Publishing that table on the healthyskepticism website was a way of making it available to readers like you to help you to check things for yourself.
I lack your confidence in the formal peer review process; after all the Keller study was peer reviewed.
I do however value genuine peer review, and would be grateful for you pointing out any fault you can spot in our paper in International Journal of Risk & Safety in Medicine, or any other papers that I have published.
Jon
Jon, I’m surprised and complimented that you took the time to respond. Thank you.
Yes, I am guilty of taking some shortcuts in forming my opinions, and have not yet taken the time to dig into all of the details of this story. I should look into it in greater detail, and arguably should have done so before commenting on it.
As a scientist in the industry I’ve spent my career doing what I regard as good and worthwhile things. I’ve worked on projects in cancer, cytomegalovirus retinitis (in the pre-protease inhibitor era), multi-drug resistant bacterial infections, disease-modifying therapies for Alzheimer’s disease, and AIDS. I’m currently seeking funding for some ideas for treating dengue.
I recognize that the commercial side of the industry has done some truly awful things. But in my experience, many critics of the industry engage in the same data cherry picking, data distortion, and conveniently shallow data analysis that the industry has in many cases been guilty of. For this reason I’ve broadly adopted the same skepticism toward papers by industry critics that I’m sure you have adopted for all papers coming out of industry. I place little weight on the conclusions drawn unless I re-analyze the data for myself, and I don’t have time to do that for each and every paper. I’ve reached a range of different conclusions on the occasions that I have taken the time to do this.
I don’t mean to convey any disrespect when I say this.
It’s good to interact with you and with David here and discuss issues honestly and respectfully. Maybe what we need is more of this. David and I are so far apart on the issues that I have trouble sometimes grasping what his basic assumptions are. But maybe with time we will both learn some new things.
How about if you open up your website to comments by non-members? Who knows, I might stop by….
This comment is posted on behalf of Graham Dukes, and addresses the issue of the responsibility of a company to people injured in a trial.
First I must reject the suggestion that the International Journal of Risk and Safety in Medicine is not peer reviewed or does not feature in authoritative bibliographies. I was founding editor of this Journal and both I and my successor have consistently used expert peer review; the Journal also features in most bibliographic systems.
Second on the question of a company’s responsibility for subjects injured in its trials, I have published on this on various occasions and summarized my conclusions in several books:
Dukes & Swartz (1988): Responsibility for Drug-Induced Injury. (Elsevier). In Chapter 16 ((Clinical Investigation of Drugs) there is long sub-section entitled “The Responsibilities of the Sponsor”.
These are said to include “a duty to ensure publication at least of the principal results, irrespective of whether these are favourable to the product or not”. There is also an account of a case (anonymized on the advice of the publisher’s lawyers!) in which a firm chose to dismiss a fatality involving an elderly patient during a trial as being “co-incidental”, and did not report it to the regulatory authorities; the effect was later found to be a characteristic of the drug and led to its withdrawal from the market.
This text does not go into the issue of compensation for injury during trials.
The Law and Ethics of the Pharmaceutical Industry. (Elsevier, 2006). Here there is a long section on “The Responsibilities of the sponsor” in clinical trials, pp. 298-301.
On the injury issue it includes the following text:
“It has long been considered that there is a moral obligation to compensate any participant who suffers injury (Childress 1976). There is also a broader belief that a sponsor should provide insurance to cover loss or injury on the part of the investigator, the institution and the participants, where such loss or injury is a consequence of their involvement in the trial.
The Association of the British Pharmaceutical Industry has long recommended compensation in such cases (ABPI 1994) and within the European Union national legislation now makes such insurance compulsory for patients or volunteers.
Sponsors and their insurers have on occasion seen fit to pay claims relating to a non-drug related injury, e.g. to a healthy volunteer patient who was injured when his bed collapsed, and another who incurred a nosocomial infection while in hospital for the study.” (p.301)
Transparency means getting everything about a piece of research out in public view for independent review and analysis by others–everything . . . the good, the bad, and the ugly. There will usually for room for disagreements. But putting out something but not everything blocks unfettered discourse and the opportunity to test for the truth of things in dispute. The first canon of empirical science is the “intersubjectivity principle,” which, simply put, means that two observers using the same means can look at and agree on what is being observed. Another canon of empirical science, the “replicability principle,” is dependent on the first. Commercial interests and politics have created world where it is possible for the FDA to regulate the practice of quasi-science in which the fundamental canons of empirical science are ignored. By “quasi” I mean “not.” In the US, this state of affairs is legislatively-induced madness. Were it not so, we would not be having this exchange of views.