Welcome to the Sausage Factory: Multiple Fraud in a Paxil Study

Dr. Jay Amsterdam, a professor of psychiatry at the University of Pennsylvania, recently lodged a very interesting complaint against five authors of a 2001 study that compared Paxil to another drug and placebo for treatment of bipolar disorder. The paper reports research paid for by SmithGlaxoKline, the makers of Paxil.  For a subgroup of patients, it says, Paxil worked better than the other drug and better than placebo. Paxil supposedly had fewer side effects than the comparison drug. Amsterdam accuses the five academic authors of plagiarism — meaning they put their names on a paper they didn’t write (like a student who buys a paper). He also says the paper grossly misrepresents the results (because the subgroup analysis was completely ad hoc and the side effects description utterly wrong). So if they did write it . . .

The paper has been cited hundreds of times.  Given the actual results — Paxil had worse side effects than the other drug, and the subgroup result means little — this is no small matter.

As Spy magazine has said, if you cheat your customers, don’t fire anyone. Email included with Amsterdam’s complaint suggests he was upset because he was not an author on the paper. Why? Well, the study was done at many sites and there could be only one author per site — according perhaps to SmithGlaxoKline. At Penn, the work (enrolling subjects) was first given to a junior faculty member named Laszlo Gyulai. However, Gyulai couldn’t enroll enough subjects. Amsterdam was asked to help and paid for doing so. He ended up enrolling more subjects (12) than Gyulai (7). Yet Gyulai was an author and he was not! This greatly bothered him. He considered it  “misappropriation” of his data, said Gyulai had engaged in “the theft and publication of a professor’s data”, and wanted Gyulai censured. Perhaps Gyulai had considered Amsterdam’s non-authorship okay because many professors who contributed subjects were not authors. Whatever the reason, it appears that authorship was determined by the firm that did the ghostwriting, Scientific Therapeutics Information, presumably following orders from SmithGlaxoKline.

I don’t know why Amsterdam waited ten years to complain. Since 2001, however, the ghostwriting problem has become much clearer. In 2001, Amsterdam complained to his department chair, Dr. Dwight Evans, about the situation. In 2010, Amsterdam learned that Evans had benefited from ghostwriting. That’s how common it was.

There’s also this:

POGO [Project on Government Oversight], in a letter to President Obama [related to Amsterdam's complaint], asked that he remove Amy Gutmann, president of the University of Pennsylvania, from her position as chairman of the Presidential Commission for the Study of Bioethical Issues, until the two cases involving Dr. Evans are fully investigated and resolved.

Chairman! Another indication how common and tolerated ghostwriting is. It is as if an obesity expert, appointed head of the most important obesity committee in the country, charged with recommending how to stop the obesity epidemic . . . is fat.

Perhaps British journalistic phone-hacking has been more common than misrepresentation of results by med school professors but the latter, I’m sure, has done more damage.

Attachments to the Amsterdam complaint. Pharmalot weighs in. Some of the accused defend themselves.

2 Responses to “Welcome to the Sausage Factory: Multiple Fraud in a Paxil Study”

  1. AZ Says:

    I found a case where a public health official outsourced the writing of a report. I checked the properties feature of the document. I googled the listed author and found the name attached to a scientific writing firm abroad. I emailed that person and the person confirmed doing the design for the report. I sat on the information, but I probably could have done something with it when the individual came before the Senate for confirmation to higher office.

  2. JeffR Says:

    I think your writing on this subject is valuable and I’d love to see it reaching a wide audience.