The House of GSK

In a just published article in the BMJ, Peter Doshi notes how in recent months the English pharmaceutical company GlaxoSmithKline (GSK) have assiduously portrayed themselves as advocates of transparency and in support of access to clinical trial data.

Well in support of ‘Responsible Access’. Responsible essentially means that a researcher commits to the primacy of RCTs and statistical significance over an analysis of adverse events. It would not for example be responsible to claim that an SSRI causes suicide, a statin muscle damage or cognitive failure, or hypoglycemics cause hypoglycemia unless a trial has shown this to happen to a Statistically Significant extent – and they never do.

This scenario suggests a variation on the old joke that Cricket is the English idea of fair-play – eleven against one. Efforts to get at the raw data are just not Cricket.

Gunfight at the GSK corral

A failure to be converted to a Responsible way of looking at the data underpins the stand-off between GSK and the RIAT team attempting to restore Study 329 to what it should have been. Study 329 is GSK’s most famous clinical trial. RIAT stands for Restoring Invisible and Abandoned Trials (see Reading the RIAT Act).

The Doshi article along with the correspondence between GSK and the RIATers makes for eye-opening reading.

To recap, in 1998, SmithKline Beecham reviewed the final data from Study 329, a study begun in 1994, comparing Paxil, imipramine and placebo. They concluded that it demonstrated that paroxetine (Paxil-Seroxat) did not work for children. They decided they couldn’t show the data to FDA but they could go ahead and publish the “good bits” of the study.

This ultimately led to a publication in 2001 in the leading journal in the field of child psychiatry, the Journal of the American Association of Child and Adolescent Psychiatrists, with an authorship line to die for. The first author was Marty Keller of Brown University. There were over twenty others.

The article states clearly that Paxil is safe and effective in children. Most doctors eyeballing it, and its distinguished authors and the journal in which it was published, would be much more inclined to use Paxil afterwards. That so many doctors in New York State in fact went on to prescribe so much Paxil to children led New York State to take a fraud action against GSK in 2004. This study was later at the center of the US Department of Justice’s case against GSK that resulted in a $3 Billion fine.

None of the apparent authors, it transpired, were authors in the sense people in the street would be likely to understand authorship. The real author – Sally Laden – appeared nowhere on the authorship line. The story behind the publication of 329 is laid out on Healthy Skepticism along with the efforts by Leemon McHenry and Jon Jureidini to get the “authors”, or the cuckolded institutions who lent their prestige to this ghostwritten article, or JAACAP, the journal in which it was published, to retract.

Which all refuse to do. And GSK as of a few weeks ago state that “GSK does not believe the article is false, fraudulent or misleading” (see GSK to RIATers).

Enter stage left

This is astonishing but there is another equally astonishing story to tell.

The temptation is for non-academics reading this to glaze over at the sight of theologians arguing about how many angels can dance on the head of a pin. But in fact those of you who have nothing to do with healthcare – the Irresponsibles – as it turns out were then and always are better placed to know when there is a problem. It was the doctors, ethicists and theologians, the Responsible Adults, who missed it. And if this is true of 329, it is likely to be true of everything else as well.

Study 329 began to unravel because a journalist, Shelley Jofre, working for BBC’s Panorama, accidentally left to her own devices, begin to dig. She went to the American Psychiatric Association Meeting in Philadelphia in May 2002, clutching the Keller-Laden paper. She was interested to interview some of the “authors”. One of her questions was “What is emotional lability?” There was a surprising number of children in 329 who became emotionally labile – what’s this? she asked. No-one seemed to know.

At APA, she approached one of the 329 authors Neal Ryan, who gave a non-answer and quickly got in touch with GSK to let them know a journalist was asking questions.

A few weeks later in July and again in August, GSK send a dossier to FDA seeking approval for Paxil for children.

A few weeks after that, on October 7th, to coincide with World Mental Health day (October 10th), Newsweek hit the streets featuring a depressed teenager on the front cover and a strapline Teen Depression: 3 million kids suffer from it. What you can do. The message inside was that Prozac was already being marketing for children and Paxil and Zoloft were about to be. What you can do – is get children on Prozac, Paxil and Zoloft.

Celebrating world mental health day

Three days later, on October 10th, World Mental Health Day, and the 40th anniversary of the passage of the 1962 amendments to the Food and Drugs Act, FDA sent GSK a letter saying Paxil was approvable for kids.

Among other things, the letter notes that GSK had already told FDA that Study 329 showed that Paxil doesn’t work for depressed children. FDA were happy to go along with GSK’s suggestion that this should not be mentioned in the labeling of the drug. Given that the Study 329 publication majored on how effective Paxil was, it would have been inconvenient if the label said otherwise.

There are many notable things in this approvable letter. Perhaps the most interesting is FDA asking GSK to clarify just what emotional lability meant.

FDA have later spun this into a story that their reviewers detected there was an increase of emotional lability events in Study 329 that needed looking at. In fact this idea came on their radar after conversations between Jofre and third parties led to a visit to FDA on August 28th and a suggestion to FDA that they explore the issue of emotional lability.

FDA at the time had a few weeks to respond to GSK’s request to market Paxil for children. Neither Jofre nor the third parties knew at the time that GSK were seeking pediatric approval for Paxil.

The secrets of Seroxat – The perils of Paxil

On October 13th Jofre’s program, The Secrets of Seroxat aired on BBC. This led to one of the biggest responses the BBC had ever had to a Panorama program – 65,000 calls and over 1000 emails.

Panorama had never repeated a topic before. Jofre and Panorama have revisited paroxetine and GSK a further four times – each time demonstrating that when it comes to the blindingly obvious sometimes what you need is plain Curiosity rather than Responsibility.

In dealing with the RIAT team, GSK still insist there is nothing wrong with the Keller-Laden paper and show no interest to co-operate with setting the record right, even though the “fraudulent” version is still the only one on the public record. As such it is still being built into guidelines on the management of children who are depressed – a trial that showed paroxetine doesn’t work and leads to a statistically significant excess of suicidality (emotional lability).

When facing requests from others such as Peter Gøtzsche to be Responsible (see GSK’s Transparency and Access Journey), GSK have typically invited them to meet GSK experts in GSK House. This is more awkward with a RIAT team that is dispersed from Atlanta to Adelaide.

It’s a lot more awkward with a journalist like Jofre, who asks herself along to GSK House with a list of questions in need of an answer.

Despite a haircut between programs, Jofre and Andy Bell produced a second Panorama program, in May 2003, Emails from the Edge that brought the walls of the Temple down. A few days after the second Panorama program, Britain’s MHRA blew a fuse and Seroxat-Paxil was on its way to being banned for children. To be continued…

The moral of the story perhaps is GSK figure letting Responsible adults in to the Club is no problem, but they recognize that a motivated lay-person, particularly a woman spells trouble.