A Review of Side Effects, by Alison Bass

I finally got a chance over this past weekend to sit down and read Side Effects, the new book by former Boston Globe reporter Alison Bass. I read the book cover to cover, which surprised me, because I was expecting that this would be an information-heavy repetition of the same territory covered by other recent books, such as Claude Barber’s Comfortably Numb or Marcia Angell’s The Truth about Drug Companies.

But Side Effects is different and more riveting, because it tells the story of several individuals, peering into details of their professional and personal lives as the events unfold. It reads like a John Grisham thriller, but it teaches you everything you need to know about how some drug companies have used their marketing and legal muscle to lie about science.

The centerpiece of the book is study 329, a placebo-controlled clinical trial of Paxil (paroxetine) for the treatment of children and adolescents with depression. This study, first-authored by Martin Keller, then and now the chairman of Brown’s Department of Psychiatry, was published in the leading child psychiatry journal in 2001. Here is the abstract from the Journal of the American Academy of Child and Adolescent Psychiatry. As you can see, the conclusion reads: “Paroxetine is generally well tolerated and effective for major depression in adolescents.” As it turns out, it was neither tolerated nor effective.

Bass describes the experience of the lead investigating attorney, Rose Firestein, as she first read study 329:

Downloading the study onto her computer, Firestein read the full text of the report and read it again. And again. She grew increasingly baffled. It appeared as if the positive conclusion in the abstract simply did not correspond to the actual data.

One of the fun things about this book is that you can participate in solving the mystery as you read. Thus, I, too, downloaded the study and tried to see the problems Firestein detected—and believe me, it wasn’t easy, because so many different outcome variables were analyzed that it’s like cutting through a jungle with a machete to figure out where the relevant result are.

This, of course, was part of the problem. The study went on a fishing expedition after the data came in, looking for a measure that would make Paxil look good, because the two primary outcomes were negative. This has been compared to changing the score of a football game by moving the goalpost after the last quarter.

But such manipulations are already so well known that they are no great surprise. What is surprising is how deliberately the drug company went about deceiving us. In a confidential memo from GSK’s Medical Affairs team that was revealed anonymously, we learn that the company knew exactly what it was doing. To quote again from Side Effects:

The memo acknowledged that study 329 had “failed to demonstrate a statistically significant difference from placebo on the primary efficacy measures.” Even so, “a full manuscript will be progressed,” the memo’s authors wrote. “It would be commercially unacceptable,” they added, “to include a statement that efficacy had not been demonstrated, as this would undermine the profile of paroxetine.”

And it goes on. I don’t want to steal any more thunder from this magnificent book, so I encourage you to get your own copy. Just don’t expect to get much sleep for a couple of days after you start it, because you won’t be able to put it down.