Q: Recently there’s been a spate of studies and negative media stories raising questions about the effectiveness of antidepressants. As a practicing therapist, I’m not sure what clinical conclusions to draw from all this confusing information. Can you help?
A: In February 2010, the cover of Newsweek magazine declared: “Antidepressants Don’t Work.” Earlier that year, “Antidepressant Drug Effects and Depression Severity,” published by Jay Fournier and his colleagues in the Journal of the American Medical Association (JAMA), arrived at a similar conclusion: antidepressants’ effects are minimal to nonexistent in treating mild to moderate depression when compared to placebos. They corroborated the findings of Irving Kirsch and his colleagues published in “The Emperor’s New Drugs”—-the first metanalysis of data from pharmaceutical companies’ registration trials.
Registration trials are the studies pharmaceutical companies conduct to demonstrate the effectiveness of their drugs to garner FDA approval. In all cases, they must be double-blind, placebo controlled, randomized trials, with adequate sample sizes. For FDA approval, the medication must be shown to be superior to placebos to an extent that’s statistically significant.
The Kirsch metanalysis included outcomes from 38 studies, and all of the antidepressants surveyed had been approved by the FDA between 1987 and 1999. However, the conclusion from this metanalysis was that the registration studies universally showed the treatment outcomes from antidepressants to be only slightly better than those from placebos. In fact, comparing the outcomes, the active medications, on average, showed only a two-point superiority compared with placebos, as measured by standard depression rating scales. A two-point advantage, Kirsch and his colleagues argue, suggests that antidepressants aren’t effective in treating major depression. Jay Fournier and his colleagues’ conclusions were essentially the same, except that their article found that with severely depressed patients, outcomes from prescription antidepressants were significantly more effective than placebos.
Such conclusions are certainly noteworthy and a reason for concern. But before clinicians completely reject the use of antidepressants, it’s important to look more carefully at the medical treatment of depression and consider a number of issues not addressed by the Kirsch article. The first, and possibly most important, issue is that all the studies reviewed in the metanalysis involved the use of only one antidepressant compared to a placebo. In addition, all FDA antidepressant registration trials recruit subjects with a diagnosis of major unipolar depression, but they exclude subjects who have comorbidities, such as substance abuse or Axis II disorders, making the study samples unrepresentative of the types of clients most therapists see.
The largest study to date evaluating the medical treatment of depression in the kind of patients most often seen in clinical practices was the National Institute of Mental Health STAR-D study (Sequential Treatment Alternatives to Relieve Depression) in 2006. In this project, researchers recruited 2,876 patients, all of whom suffered from major, unipolar depression. Among the subjects, 80 percent suffered from chronic or recurrent major depression, 25 percent had been depressed continuously for two or more years, and 65 percent of subjects had complex psychiatric and/or medical comorbidities.
The STAR-D study began by treating all subjects with the antidepressant citalopram (Celexa). Aggressive dosing was used if necessary to bring subjects to full recovery. By seven weeks, 30 percent of those treated had reached full recovery. All subjects that hadn’t reached complete remission then moved into phase two.
This phase consisted of two treatment options: combining Celexa with another medication (Wellbutrin or BuSpar) or switching from Celexa to a different antidepressant (Effexor, Wellbutrin, or Zoloft). By the end of phase two, a total of 55 percent of the subjects had recovered.
Phase three offered the remaining subjects (those not reaching full recovery in phases one or two) other augmenters or switching possibilities. The final outcome was that 67 percent of subjects treated met criteria for full recovery. The one-year follow-up, however, showed that only 45 percent of the original sample remained in remission.
The STAR-D study didn’t include a placebo control and wasn’t an open study, meaning that it wasn’t blinded. However, it reflects both the type of patients commonly seen by most therapists (those with severe depression and complex comorbidity) and a sequential approach to medical treatment. Such a strategy clearly is indicated if the first treatment either fails or only provides partial relief. It’s again important to emphasize that in all registration trials, subjects are treated with only one drug.
Another clinically relevant feature to bear in mind when considering the results of the Kirsch metanalysis is that 91 percent of the studies reviewed lasted six weeks or less. In the treatment of depression, some patients respond to antidepressants during the first four to six weeks, but if augmentation or switching is done, this always requires more than six weeks. This is especially true if the goal is full recovery.
While treatment outcome results in both antidepressant medication and psychotherapy studies fall significantly short of full recovery for all patients treated, the truth is severe depression is difficult to treat. For depression, the following treatments have been empirically validated: cognitive therapy, interpersonal therapy for depression, behavior therapy, antidepressant medications, high intensity light therapy, exercise, and electroconvulsive therapy (ECT or shock therapy). Even with integrated approaches, many people experience partial recovery, but continue to have some lingering depressive symptoms.
The articles by Kirsch and Fournier, and the Newsweek story, present accurate data, but fail to mention sequential treatments. I believe this limits the generalizability of the results being reported. When complex research results are oversimplified in the media, readers may misinterpret the findings, and some people who might otherwise be helped may not seek treatment. In the case of depression this is especially unfortunate, given that the World Health Organization predicts that by the year 2020, depression will be the second leading cause of reduced life expectancy due to suicide and to a host of medical illnesses related to it, most notably heart disease.
Increasingly, mental health professionals are under scrutiny to choose treatments that have empirical support, but it’s also important for the profession not to be drawn in by sensationalized or distorted reporting of research findings. Without the appropriate critical discernment and appreciation for the nuances of clinical applications of research, empirical findings can often cause harm as well as helpful therapeutic guidance.
John Preston
John Preston, Psy.D., ABPP, is a professor with Alliant International University (California School of Professional Psychology), Sacramento. He has also taught at UC Davis, School of Medicine. He is the author/co-author of 21 books on various topics including depression, psychotherapy, psychopharmacology, borderline disorders, bipolar disorder, and post-traumatic stress disorder. He is the author of “Drugs in Psychiatry” chapter in the Encyclopedia Americana as well as Integrative Treatment for Borderline Personality Disorder: Effective, Symptom-Focused Techniques, Simplified for Private Practice. Dr. Preston is the recipient of the Mental Health Association’s President’s Award for contributions to the mental health community and the California Psychological Association’s “Distinguished Contributions to Psychology” award. He has lectured in the United States, Canada, Europe, Africa, and Russia.
