In a new review of data from a 2001 trial, researchers say there is no evidence that paroxetine (Paxil), an antidepressant that is approved for treating depression, obsessive-compulsive disorder and social anxiety in adults, is effective or safe for teens. The original study had stated that paroxetine is “generally well tolerated and effective for major depression in adolescents.”
In the re-analysis, published in the BMJ, an international group of scientists conducted a thorough review of Study 329, which was originally published in the Journal of the American Academy of Child and Adolescent Psychiatry. Since its publication, reports of suicidal thoughts and behavior among teens taking the drug raised questions about paroxetine’s effectiveness and safety, and in 2002 the U.S. Food and Drug Administration began investigating the drug’s potential dangers among adolescents. The following year, the agency advised doctors not to prescribe it to teens suffering from depression.
Doctors and researchers have called for the journal editors to retract or revise the paper; the editor-in-chief, after reviewing it, ultimately said there was "no basis [was] found for editorial action against the article."
In 2012, the drug’s maker, GlaxoSmithKline (GSK), paid $3 billion in fines for illegally marketing certain of its drugs, including paroxetine, for unapproved uses and for failing to report drug safety information to the FDA.
Only one antidepressant, fluoxetine (Prozac) is approved by the FDA for treating depression in teens. But other medications in the same class, known as selective serotonin reuptake inhibitors (SSRIs), are commonly prescribed by doctors for younger patients in so-called off label use. Their comfort in doing so comes from trials like Study 329.
Jon Jureidini, professor and research leader of critical and ethical mental health at the University of Adelaide, charges that the authors of the original 2001 paper “deliberately misrepresented the outcomes of the study” and changed the protocols of the study without following the proper procedures to do so.
Dr. Martin Keller of Brown University, who led Study 329, did not immediately respond to a request for comment on the current reanalysis.
All trials are designed with a strict protocol that defines, before the trial starts, what the researchers will measure and how they will measure it. These protocols are approved by review boards to ensure that they are scientifically sound and that they follow ethical guidelines for protecting the safety of the trial participants, who volunteer for the study. Any change in the objectives needs to be submitted to and approved by the board again.
However, that’s not what the authors appear to have done for Study 329. While they laid out two primary objectives—where the participants ended up on a depression rating scale, and the change in this measure from the start to the end of the study—these were not the outcomes they reported on.
Instead, they reported on a modified version of one of the outcomes and four additional outcomes that were listed as secondary, not primary, objectives. The American Academy of Child and Adolescent Psychiatry referred questions about the study to the Keller and his co-authors, who did not return request for comment.
Another concern Jureidini has with the original study is that adverse events were not reported in the proper way, with suicide attempts severe enough to require hospitalization labeled as “emotional liability."
GSK issued a statement in response to the study noting that “the findings from this team’s analysis appear to be in line with the longstanding view that there is an increased risk of suicidality in pediatric and adolescent patients given antidepressants like paroxetine. This is widely known and clear warnings have been in place on the product label for more than a decade. As such we don’t believe this reanalysis affects patient safety.”
Jureidini and his team were able to make these discoveries because they had access to the raw data generated from the trial, something that is not routine when clinical trials are published. Most scientists, and even the reviewers who review papers for publication, generally see the data packaged in the way that the authors present it. The current reanalysis is part of an effort by researchers to make such raw data more accessible and transparent to ensure interpretations are correct and reliable.
GSK posted much of the data on its website and added an additional 77,000 pages related to Study 329 after some negotiation. “It’s not the point of this paper to humiliate GSK or to accuse anyone of fraud,” Jureidini says. “It’s an attempt to honestly present the data. This is what the original paper should have looked like. If the original paper had reported things this way, there never would have been a problem.”
Decisions that doctors and parents make about whether to use medications to treat depression in children might have been different, too. In 2003, the FDA required black box warnings on SSRIs alerting people of the risk of suicidal behavior in children taking the medications. One study also published in the BMJ found a drop in the use of antidepressants among children between 2000 and 2010 and an increase in suicide attempts. Whether the trends could be attributed to untreated depression among teens isn’t clear, but the data highlight the delicate balance that mental health professionals and parents face in deciding how to treat depression in teens.
“It’s a very uncomfortable thing for us to be saying,” says Jureidini about his findings. “I can hear the responses to this study: that what you are doing is poisoning people’s minds in relation to antidepressants, and psychotropic drugs and psychiatry in general, that you are going to do terrible harm. We’re trying to bring out the truth.”
With the current re-analysis, more doctors may weigh the risks of potential increased suicidality more heavily before prescribing the drug to teens. At the very least, physicians who decide to prescribe paroxetine should monitor their patients more carefully for any potential signs of worsening depression or desire to harm themselves.
“It’s possible to make data available in a way that opens it up for the scientific community,” says Jureidini about addressing concerns over participant confidentiality. As this study shows, “it’s imperative that the scientific community should have access to it.”