In the early 1990s, everyone was listening to Prozac right up until they weren't anymore. By the middle of that decade, the media had begun spotlighting disasters involving the life-changing wonder drug cases of suicide, even mass murder, apparently spurred by Prozac and similar medications. And by 2005, the Food and Drug Administration had mandated a black-box warning on all antidepressants, cautioning against an increase in suicide risk in certain patients.
Since then, a small but vocal minority of researchers have also questioned whether the mood-enhancing benefit of antidepressants amounts to anything more than a psychological artifact. They point to studies that suggest the drugs' seemingly powerful effects are the same as those of a sugar pill. Most recently, a headline-grabbing Journal of the American Medical Association (JAMA) paper published in January found that antidepressants worked no better than a placebo in patients with mild or moderate depression (but the study did conclude that medication helped the most persistent and severe cases). For some observers, however, the judgment was sealed. That month, Newsweek ran a cover story bemoaning "The Depressing News About Antidepressants."
But how can the same drug be at once poison, panacea and placebo? With about 7% of the American population estimated to suffer from depression at any given time, and some 67 million prescriptions written for the top three antidepressants alone in 2009, the answer is of wide public-health and economic interest.
The Placebo Studies
Less emphasized than the study's findings was that fact that the January paper in JAMA was a meta-analysis not an original clinical trial, but a review of the combined results of multiple previous studies. Although meta-analyses can be useful for summarizing large amounts of data, they can sometimes be misleading. For one thing, such papers are only as good as the studies on which they rely, and in cases like this, in which individual responses to a class of medication vary widely some people improve dramatically, while others get much worse the particular studies the authors decide to include can sway the results.
The JAMA meta-analysis wound up including just six studies out of the more than 2,000 the authors considered on only two drugs, the selective serotonin uptake inhibitor (SSRI) Paxil and an older tricyclic drug called imipramine, though there are dozens of antidepressants on the market. (Paxil happens to be the most controversial of all SSRIs; it is most strongly associated with negative side effects that many similar medications do not have, such as weight gain, withdrawal symptoms and possibly even birth defects.) The meta-analysis also did not include studies that had a "placebo washout period" those that start all patients on placebo before introducing the antidepressant, in order to identify and eliminate people who get better on their own. While drug companies argue that these studies help them gauge the real effects of the drug, critics say they give antidepressants an advantage that doesn't reflect clinical practice either way, eliminating these studies strengthens the placebo effect in a meta-analysis.
Further, when data are aggregated, patients' opposing responses to antidepressants can cancel each other out. If one patient's depression score drops 20 points while taking medication, and another's increases by the same amount, in aggregate there has been no effect. "You can have an overall slightly significant change in active treatment vs. placebo, but what you may see when you look at the figures on a patient-by-patient level is that there are dramatic differences," says Richard Tranter, a psychiatrist and consultant with the North West Wales NHS Trust in the U.K., who was not involved in the JAMA study.
Overall, the scientific evidence confirms that not all patients respond the same way to antidepressants, and not all antidepressants even those in the same class work the same way in any given individual. One drug can, in fact, have profoundly different effects in the same patient at different times in his or her life.
Yet even with its narrow selection of studies, the JAMA review still showed improvement in people with severe depression and those whose illness was mild but chronic. And the troublesome data from placebo-controlled antidepressant trials is not unique to this type of drug. Notably, about half of all clinical trials of opioid painkillers like OxyContin, which are powerful and known to work, show the drug to be no more effective than placebo, according to Dr. Gavril Pasternak, an opioid expert at Memorial Sloan-Kettering Cancer Center. Again, it's not because the medication is ineffective, but because of the huge variance in individual response and metabolism of the drug.