Roughly one in eight Americans older than 12 is on antidepressants, and a quarter of the people who take these drugs have done so for more than 10 years. That’s according to some of the most recent mental health data collected by the Centers for Disease Control and Prevention (CDC). Considering that tens of millions of Americans are now taking selective serotonin reuptake inhibitors (SSRIs) and other antidepressants, one would reasonably expect that their efficacy has been proved beyond all doubt. But experts who study antidepressants are still debating whether they’re superior to placebos.
“There is data on both sides,” says Michelle Newman, a professor of psychology and psychiatry at the Pennsylvania State University and director of PSU’s Laboratory for Anxiety and Depression Research.
A research review published earlier this year in the Lancet concluded that the 21 antidepressants it analyzed significantly outperformed a placebo. While many media outlets declared this a victory for antidepressants, some expert reactions were more tepid. “It’s tricky, because these drugs do work, but you have to define what you mean by ‘work,’” says James McCormack, a professor of pharmaceutical sciences at the University of British Columbia.
“So, if one out of 10 people who take something get better, would you say it works?”
McCormack co-authored a BMJ editorial that critiqued the Lancet data. While he’s complimentary of the researchers who conducted the review, his editorial says of the data: “These findings do not support the widespread calls in the popular press for more people to take antidepressants.” Pulling in additional research on placebos, McCormack and his BMJ co-author argue that the Lancet data really showed that for every 10 patients who start taking antidepressants, five will report feeling better. But four of them would have felt better even if they’d taken a placebo. “So, if one out of 10 people who take something get better, would you say it works?” he asks.
Further complicating the evidence is the fact that the results of many antidepressant studies never see the light of day. “Drug companies have no cap on the number of controlled trials they can conduct,” Newman says. While the Food and Drug Administration (FDA), which oversees the prescription drug approval process, requires that pharmaceutical companies submit data from all their trials, they don’t require that drug companies publish all their findings.
To approve a drug, the FDA stipulates that a “developer” (basically, a drug company) “must have adequate data from two large, controlled clinical trials,” according to publicly available agency information. As Newman points out, there is no limit to the number of studies a drug maker can conduct to get the two needed for approval.
“The reality is that there are way more trials that have found no difference than that have found superiority,” Newman says of trials pitting prescription antidepressants against placebos. “Those null trials never get published or released and are not included in most meta-analyses — including the most recent one published in Lancet.”
An exception, she says, are some research reviews conducted by Irving Kirsch, the associate director of the Program in Placebo Studies at Harvard Medical School.
Back in 1998, Kirsch and one of his graduate students published a research meta-analysis that found the majority of an antidepressant drug’s benefit — at least 75 percent of it — was attributable to placebo effect. “Needless to say, our meta-analysis proved to be very controversial,” Kirsch has since written.
“If you have a common cold and I treat you with a drug, there’s a 100 percent chance you will feel better within two weeks — but that’s not because of the drug.”
After a lot of pushback from critics and clinicians, Kirsch repeated his analysis — only this time, he also included, thanks to a Freedom of Information Act request, all the unpublished data the drug manufacturers had submitted to the FDA. More than half of these trials found placebos either equaled or outperformed antidepressants. When pooled together, 82 percent of an antidepressant’s benefit was attributable to placebo effect, Kirsch’s review concluded.
It’s important to note that while many people consider a placebo to be a fake treatment — or the equivalent of no treatment at all — that’s not accurate. Decades of research have shown that placebos can have a real and meaningful effect on patient outcomes — even when used to treat conditions that are not psychological in nature.
“Large placebo effects have been seen in patients with hypertension, Parkinson’s, and even multiple sclerosis, and these are all illnesses where no one questions whether they’re biological,” says Gerard Sanacora, MD, a professor of psychiatry at Yale University and director of the Yale Depression Research Program. Sanacora mentions imaging research that suggests placebos can cause measurable, opioid-like changes in a patient’s brain and cerebrospinal fluid.
If antidepressant drugs provide a benefit — even a small one — over placebos, that’s meaningful. “The unintended message from some of these placebo studies is that [antidepressant] medicines don’t work, but that’s really not true — they work really well,” Sanacora says. And even if a portion of a drug’s benefit does stem from placebo, it’s still true that you won’t get any of this benefit if you don’t take the drug, he says.
Another important point: There are very few clinical trials that compare antidepressants to no treatment at all. Depression can be a lethal disease, so it would be unethical to deny depressed people treatment for the purposes of drug-development research, says Charles Nemeroff, MD, a professor of psychiatry and behavioral sciences at the University of Miami and chief of psychiatry at the University of Miami Hospital. “So, even in the placebo group, patients are usually going from nothing and feeling awful to going to a clinic and being seen weekly by nurses and other people involved with the study,” he explains. This figurative “laying on of hands” is almost certain to provide some benefit, Nemeroff says.
It’s also well established that depression is a cyclical condition — its severity tends to ebb and flow. McCormack says these natural ups and downs mean that you need to know what would happen to those in a no-treatment group to properly tease out both the placebo effect and the treatment effect. “If you have a common cold and I treat you with a drug, there’s a 100 percent chance you will feel better within two weeks — but that’s not because of the drug,” he says.
There are additional problems with the data on antidepressants’ benefits.
In a recent review article, Harvard’s Kirsch points out that in trials pitting antidepressants against placebos, participants are informed of the antidepressants’ side effects. These side effects are common, and their emergence could tip off participants to the fact that they received the “real” drug and not the placebo. This could both increase the antidepressant’s apparent effectiveness while decreasing the placebo’s, which may help explain why antidepressants usually outperform placebos in big review papers.
There’s some research that supports this argument. One 2009 review found that antidepressant trials that don’t include a placebo group find much bigger drug effects than studies that do include a placebo group. These findings could be seen as support for the argument that if study participants truly didn’t know whether they were on a placebo or an antidepressant — that is, if side effects didn’t tip them off — the two treatments might be equally effective.
Put all this together, and it’s unclear how doctors should apply the existing research when treating people with depression. On the one hand, antidepressants clearly work for many people — even if some or all of their benefit is attributable to placebo effect or the cyclical nature of the condition. On the other hand, antidepressants come with side effects — including sexual dysfunction, sleepiness or insomnia, and weight gain.
Some doctors — especially those outside the United States — skirt this issue by prescribing placebos to their patients. Based on the existing evidence, you could make the argument that this provides most of an antidepressant drug’s benefits without the risk of side effects. But many doctors consider this unethical, and American clinicians who try it could be sued.
Sanacora says that, in most cases, it’s prudent to begin treatment with non-drug psychotherapy and add antidepressants only if a patient isn’t responding. McCormack adds that giving patients extremely low doses of antidepressants — a quarter of a normal dose — is another way for clinicians to use these medicines (and any associated placebo benefits) while reducing the chances of exposing their patients to unnecessary side effects.
But Newman argues that even combination therapy has drawbacks. “Doing both can undermine the efficacy of the psychotherapy,” she says. Patients who take pills tend to relapse as soon as they stop and tend not to put the same effort into psychotherapy that they would have without meds, she says. “My viewpoint is that antidepressant efficacy is overrated and antidepressants are overprescribed,” Newman adds.
The future will no doubt bring new drugs, along with new analyses of their effectiveness. In the meantime, the debate over whether antidepressants outperform placebos rages on.