"Study 329," one of the most controversial studies in psychiatry, has come under fire again, this time by a motley crew of seven academic researchers and retired psychiatrists publishing a paper in one of the most respected medical journals.
The original 2001 study, carried out by Brown University researchers and funded by pharmaceutical giant GlaxoSmithKline, claimed that the company's antidepressant Paxil was safe and effective for use in kids with depression aged 12 to 18.
But the new paper, a massive effort reanalyzing tens of thousands of pages of reports, finds that neither of those claims is true.
"Our basic message is, this study shouldn't have been published, and once it was published it shouldn't have stayed there," Mickey Nardo, a retired psychiatrist from Georgia and one of the new study's authors, told BuzzFeed News. "This is a clinical trial that felt like it had been jury-rigged."
The new paper, published Wednesday evening in the British journal BMJ, was two years in the making, after GSK gave Nardo and his colleagues 77,000 documents from the original study 329.
After reanalyzing one-third of the documents, the researchers found that suicide rates were much higher in kids who took Paxil (also known as paroxetine) than those who received a placebo, and that the antidepressant was barely better than placebo at alleviating symptoms of depression -- a small enough difference that it was likely just chance.
Independent psychiatry researchers say the new analysis highlights a problem that plagues many types of studies.
"If you just go rummaging around in the data, making comparisons between all of these various things, over time you'll turn over things that are statistically different," Mark Olfson, a professor of psychiatry at Columbia University, told BuzzFeed News.
The link between Paxil and suicide is well-known. In 2004, a black-box label was added to Paxil and other drugs in the same class warning that they increased suicide in kids. GSK has also been the subject of several major lawsuits around suicides and attempted suicides, and the company has paid out an estimated $390 million in penalties from these suits to date.
In 2012, GSK paid $3 billion in fines over false marketing of products, including the illegal marketing of Paxil to children.
The new study is not the first time a group has pointed out discrepancies in the data. A report commissioned by Senator Chuck Grassley found that not only did the study manipulate statistics to make the suicide risk look smaller, but that GSK hid that risk when they applied to the FDA for the rights to sell the drug to kids.
Some researchers have concerns about bias among the authors of the new study, whose members include David Healy, a psychiatrist who often speaks out against the pharmaceutical industry, and Jon Jureidini, a psychiatrist who has long agitated to have Study 329 retracted by the journal that published it. Mickey Nardo runs a blog called 1 Boring Old Man, where he has criticized the study a number of times.
"The group is a self-appointed watchdog," Jeffrey Lieberman, chair of psychiatry at the Columbia University College of Physicians and Surgeons, told BuzzFeed News. "One wonders what the motivation is, and how objective they're going to be."
But the watchdogs were ready for that criticism. "All of us have conflicts of interest — namely, that we don't like this study," Nardo told BuzzFeed News. "The paper is so long because we were very careful to stick to the rules and show our work."
This paper is part of the Restoring Invisible and Abandoned Trials Initiative. Several journals, including the BMJ, welcome papers by groups that reanalyze clinical trial data from other researchers.
But some aren't sure this double-checking is widely applicable.
"It's so labor intensive, I'm not sure how we can generalize it. It's an amazing work, but who can fund this?" Khalid Afzal, a professor of child and adolescent psychiatry at the University of Chicago, told BuzzFeed News.
This paper, at the least, is an example of what can be done with open source data — not just double checking, but finding new insights in the same data, too.
"No single study will ever be the final word on the hypothesis that's being tested," said Bryan King, the director of child and adolescent psychiatry at the University of Washington. "The real value is, people who have new ideas can go to the repository and, potentially, have questions answered without having to do a new study."