Studies May Overstate The Benefits of Talk Therapy For Depression
Wednesday, September 30, 2015
The benefits of talk therapy for depression have been overstated in the scientific literature, according to a study in the journal PLOS ONE.
The finding comes several years after a similar study reached the same conclusion about antidepressant drugs.
Both talk therapy and antidepressant drugs "are efficacious," says Steven Hollon, a professor of psychology at Vanderbilt University and an author of the study, which was published Wednesday. "They're just not as efficacious as we think they are."
For each treatment, researchers found that the apparent effectiveness was inflated by publication bias. This sort of bias occurs when studies finding that a treatment works are more likely to be published than those with a negative finding.
"It's like flipping a bunch of coins and only keeping the ones that come up heads," Hollon says. The result is that anyone who reviews the published literature on a particular treatment will see a distorted picture.
The study of talk therapy involved a review of 55 National Institutes of Health grants awarded between 1972 and 2008. The grants paid for clinical trials of psychotherapy for depression.
But results from nearly a quarter of these trials were never published, says Erick Turner, a psychiatrist and researcher at Oregon Health and Science University and an author of both the 2008 study on depression drugs and the new one on talk therapy.
Turner and his colleagues were able to obtain the unpublished results from the researchers who did the trials. "And when you bring in the unpublished data it brings down the apparent efficacy of psychotherapy for depression" by about 25 percent, he says.
The new finding could help reverse an unfortunate side effect of the 2008 analysis of depression drugs, Turner says. After the study came out, it was used by critics of antidepressants to suggest that people with depression should avoid drug treatment.
"There was a wave of just simple antidepressant bashing that went on and has continued," Turner says. And the results were used to suggest that people with depression should choose talk therapy over drugs.
That made no sense to Turner. "Why should we be recommending this other treatment when it might be just as fraught with publication bias as the drug literature is?"
So Turner was happy to join Hollon and other researchers interested in finding out whether the psychotherapy research had the same bias as drug research.
The result of that effort should reduce unjustified criticisms of drug treatment, Hollon says. "This article if anything kind of evens up the playing field," he says.
Publication bias may have led psychiatrists and psychologists to be too optimistic about both talk therapy and drug treatment, Hollon says. But he says that's less likely to be a problem for patients.
"When people are depressed they are usually unduly pessimistic (about treatment)," he says. "The biggest problem we have in depression is getting people to avail themselves of treatment, whatever it is."
And the act of starting treatment produces a powerful placebo effect, which can make a big difference even if the treatment itself produces only a small benefit, Hollon says.
Of course, publication bias isn't limited to depression treatments. It's a widespread problem throughout the research world, says Kay Dickersin, a professor of epidemiology at Johns Hopkins Bloomberg School of Public Health.
"We're rewarded for getting papers out, for finding results that will bring attention to our universities and to ourselves," she says. "We aren't rewarded for being honest; we're rewarded for making a splash."
And that presents a challenge to both doctors and their patients, Dickersin says. "I think the question that's really arisen is, how much of what's out there should we really believe."
One solution, Dickersin says, is to have research grants include a requirement that all study results be made public, even if they're never published.
Copyright 2015 NPR. To see more, visit http://www.npr.org/.
To view PDF documents, Download Acrobat Reader.