Internal documents reveal extent of Paxil drug study chicanery

When I was a reporter for The Boston Globe in the 90s, an employee of Brown University's department of psychiatry handed me a raft of internal university documents. A number of these documents pertained to an ongoing clinical trial that compared the antidepressant Paxil to a placebo and older antidepresssant (imipramine) in the treatment of depression in adolescents. My book, Side Effects tells the story of how researchers at Brown University misrepresented data in this key clinical trial, known as study 329, so in the interest of transparency, I am posting these and other pertinent documents on my website.

The first document here is from the final GlaxoSmithKline report of study 329 (publicly available on its website), which shows that a 14-year-old boy who was in the Paxil arm of the study lost control, punching pictures, breaking glass and injuring himself in November 1994. He was hospitalized and seen as suicidal by the psychiatrists treating him. Yet this patient , known as #65 in the study, was not labeled as suicidal when the study was published in the July 2001 issue of the Journal of the American Academy of Child and Adolescent Psychiatry. (The following year, Paxil became the most widely prescribed antidepressant in the U.S.).

The second document here is a memo to Brown University's Institutional Review Board from Martin Keller, chief of psychiatry at Brown and principal investigator of study 329. In the memo, Keller reports that a teenager in the Paxil study was hospitalized in September 1995 due to becoming combative and suicidal. Yet in the memo, Keller says he has labeled this patient (#106) as noncompliant instead of suicidal as a result of taking Paxil. Likewise in the published 2001 study, this teenager is labeled noncomplaint and not included in the list of adolescents withdrawn from the study as the result of adverse side effects.

The third and perhaps most mysterious document is a memo from Keller to the IRBs at Brown and two of its affiliated hospitals, Butler and Bradley. In this January 30, 1995 memo, Keller reports that a teenage girl, patient #70 in study 329, ingested 82 Tylenol pills on January 19 and was hospitalized at St. Ann's Hospital. She was discontinued from the study at the end of January and coded as noncompliant according to another memo from Keller to Brown's IRB here. Yet according to the GlaxoSmithKline's final report, patient #70 in the same study was a 12-year-old boy enrolled in the trial on February, 22, 1995 and withdrawn on March 24 after suffering from chest pains. This patient had been randomized to the imipramine arm of study 329.

So the question remains: how did patient #70 go from being a teenage girl who overdosed on Tylenol to a 12-year-old boy with chest pains?

As I report in my book, Donna Howard, the former assistant administrator in Brown's department of psychiatry who gave me the internal Brown documents, said the data in study 329 was changed to satisfy the study's sponsor, SmithKline Beecham (which later merged with another company to become GlaxoSmithKline).

"Everybody knew we had to keep SmithKline happy and give them the results they wanted," Howard said. And she reiterated that comment last week when she was interviewed by Brown's own student newspaper, The Daily Herald.

In an editorial accompanying the news article, The Daily Herald editors raised the issue of why the university seemed to be judging faculty members by a different standard than it judges students who misrepresent or falsify data. Good question!