Confidence Game

Antidepressants may be placebos, but that doesn't mean they don't work.

Is it possible that everyone who's been listening to Prozac has been hearing things?

During the last decade or so, millions of Americans have taken Prozac and similar antidepressants, known as selective serotonin reuptake inhibitors (SSRIs), to help achieve happier, less anxious lives. But a new study suggests that people feel better after taking these drugs mainly because that's what they expect.

The finding will dismay many people who are convinced that an SSRI was the key to their self-improvement. Just as Viagra, originally approved for treatment of impotence, is now taken as an all-purpose sex enhancer, SSRI use has expanded beyond severe depression to include a wide range of dissatisfactions and problems in living.

In his book Creating Mental Illness, the sociologist Allan V. Horwitz observes that so-called antidepressants "work equally well for a broad range of disorders including panic, obsessive, and phobic conditions, as well as depressive and anxious states. They are also widely used for substance abuse and eating disorders and for general distress among both adults and children."

Not only that, but "these medications are promoted as ways to enhance the personalities of normal people by improving self-esteem, self-confidence, interpersonal relationships, and achievement." This sort of use was chronicled, somewhat ambivalently, by the psychiatrist Peter D. Kramer in his 1993 bestseller Listening to Prozac.

The broad range of applications for which SSRIs are recommended has contributed to the perception that they are too good to be true. A study in the current issue of the American Psychological Association's journal Prevention and Treatment lends substantial support to that view.

A team led by University of Connecticut psychologist Irving Kirsch analyzed clinical trial data for six widely prescribed antidepressants approved by the Food and Drug Administration between 1987 and 1999: Prozac, Paxil, Zoloft, Effexor, Serzone, and Celexa. They found that "80% of the response to medication was duplicated in placebo control groups."

In other words, subjects who received only the placebo improved almost as much as subjects who got the drug. The average difference in improvement was only two points on the Hamilton Depression Scale, which generates scores up to 50 or 62 points, depending on the version used.

The difference was so small, Kirsch and his colleagues report, that it could be due entirely to patients who surmised, based on side effects, that they were receiving the real drug and therefore had stronger expectations of improvement. If so, the much-ballyhooed SSRIs would be nothing more than placebos themselves. Hence the title of the study: "The Emperor's New Drugs."

"Our data suggest that the effects of antidepressant drugs are very small and of questionable clinical significance," Kirsch et al. conclude. Given the potential side effects of SSRIs, they say, "antidepressant medication might best be considered a last resort, restricted to patients who refuse or fail to respond to other treatments."

The study is accompanied by nine commentaries. None of them questions Kirsch et al.'s main finding, which is consistent with the results of earlier analyses. But some of the commentators argue that antidepressants may be more beneficial than the FDA data indicate.

"Are these widely touted, intensively marketed drugs as pathetically ineffective as these…data suggest?" asks Brown University psychiatrist Walter A. Brown. Perhaps not, he says, if SSRIs are especially helpful to certain kinds of people, such as the severely depressed, or if their effects last longer than the effects of placebos.

"Medication treatment is no less potent than other clinical alternatives," write three psychologists and a psychiatrist. "Psychological mechanisms may account for the bulk of its effects (on average), but it is at the least a very effective way of mobilizing those mechanisms."

Similarly, Daniel E. Moerman, a professor of behavioral sciences at the University of Michigan, says "a doctor with a drug (especially a hot, new one) can be a powerful 'meaning delivery system'…If physicians prescribed placebos alone, their effect would be significantly reduced because doctors have little confidence in them."

The same thing could happen, of course, if physicians and their patients take to heart the evidence indicating that the pharmacological effects of SSRIs are negligible. Calling attention to the placebo effect may be a good way to ruin it.

But perhaps the evidence won't matter. As Moerman observes, "Far too many people experience substantial benefits from these drugs (and/or from the context of their use) to simply let them go."