Occasionally in medicine, analyses of data have consequences so weighty and far-reaching that the outcome lasts for years, involves several countries and regulatory agencies, and has massive implications for patients, doctors, and the pharmaceutical industry.
The just-published reanalysis in the British Medical Journal of data concerning Paxil (Seroxat in the UK), one of the most widely prescribed antidepressants and anti-anxiety medications in North America and Europe, represents one such example about the power of data, with the fallout destined to last for years.
In this case, the data come with a history involving not just ghost writers, withheld data, and misleading statements, but also industry fines against the drug maker, GlaxoSmithKline, as well as last-minute black-box warnings about the risk of suicide to those patients already prescribed its medication.
When these revelations first drew attention in October 2004, after a series of investigative reports aired on the BBC news program Panorama called "Taken on Trust," senior pharmaceutical regulators in the UK stated on camera that they were “disgusted” and “horrified” by the deliberate withholding of information. The president of the Royal College of Psychiatrists said this “has serious implications for the whole of psychiatry; it has serious implications for the whole of medicine.” Britons prescribed the drug for social anxiety disorder—whose official symptoms in DSM-5 continue to include “fear of going to parties," "failing to speak in social situations,” and even "overpreparing the text of a speech"—were told that we may come to see the masking of Paxil's discontinuation symptoms as “one of the biggest medical scandals ever.”
The study at the center of this, "Study 329," has long been criticized; it remains unretracted by both its official authors and contributors. (I have reached out to GSK but have not yet received comment.)
The drug maker did receive a $3 billion fine by the New York State Attorney General; it had also opened its internal archives for scrutiny. Meanwhile, Brown University, where the lead author Martin Keller was based, has opted for silence; former Brown psychiatry chair Keller has been the focus of investigation (see more from Boston Globe reporter Alison Bass) concerning his under-reporting of financial ties to the pharmaceutical industry, including to the maker of Paxil.
The just-published reanalysis of Study 329 will reignite the controversy tied to this study, for it concludes in detail that Paxil is neither safe nor effective in adolescents with depression. Having scrupulously adopted the same criteria and protocols that the first set of researchers used (whose findings were approved by the FDA), the latest study reaches a conclusion that is 180 degrees different from the original one while documenting precisely what some have feared, known about, and written about for years: that Paxil should not have received FDA approval in the first place. The study, published in 2001, claimed that Paxil was “generally well tolerated and effective,” when the full picture established by the inclusion of all data indicates that it was neither.
The reanalysis indicates more than a doubling of the suicidality rate and a tripling of the event rate than was first reported in 2001. It also shows a withdrawal or discontinuation syndrome serious enough to trigger suicidality. Had the FDA received the full data generated by Study 329, including the amount of suicide-ideation tied to paroxetine, it would have had no basis to approve the drug for anxiety or depression. Instead, Paxil was prescribed to millions of people worldwide—at peak, roughly 5,000 Americans began a new course of treatment on it daily. Study 329 was subsequently invoked to support the creation of depression in adolescents and children, and, later still, of pediatric bipolar disorder.
The reanalysis of such important data stems from a significant and welcome initiative that Peter Doshi, associate editor of the British Medical Journal, has dubbed RIAT, short for Restoring Invisible and Abandoned Trials. It stems from recognition that drug safety requires the release of all relevant data. It begins from the premise that transparency about data is a key way to prevent harm to patients.
The international team of researchers that carried out the reanalysis involved 12 North American academic psychiatry centers. Its assessment raises a host of urgent questions, not just about Study 329 and why some participants were given twice the official dosage of a comparison drug (imipramine) while others were not warned about the risks of suicide associated with paroxetine. The controversy tied to Study 329 also exemplifies the need to correct misleading data and modify the influence of that data on new and revised definitions of illness, including childhood depression and pediatric bipolar disorder.
The alternative is to accept the practices that drug makers have adopted for years, though with far greater intensity since the 1990s—for drugs such as Paxil to be prescribed to millions of people worldwide on the basis of faulty information.
More information on Study 329 and its reanalyzed data appears here: www.study329.org