by David Feifel, MD, PhD

Dr. Feifel is Associate Professor In Residence, Department of Psychiatry; Director, Neuropsychiatry and Behavioral Medicine Program; Director, UCSD Adult ADHD Program, University of California, San Diego Medical Center, San Diego, California.

A recent article appearing in the January issue of the New England Journal of Medicine has created quite a controversy. The paper titled, “Selective Publication of Antidepressant Trials and Its Influence on Apparent Efficacy”[1] has cast a shadow on the efficacy of antidepressant medication in depression. The authors obtained and reviewed the results of all Phase II and Phase III randomized, placebo-controlled studies of drugs for the short-term treatment of depression registered with the US Food and Drug Administration (FDA). They then compared these registered studies with those in the literature and found that only 69 percent of the studies on file with the FDA were published, and the studies deemed to have clearly positive outcomes were much more likely to be published (37/38) compared to those resulting in a negative or equivocal outcome (14/36). Furthermore, many of the published studies that were deemed negative or equivocal were presented in a way that made the outcomes appear positive, according to the authors.

Based upon their findings, the authors concluded that there is a strong publication bias against publishing studies on antidepressants with findings that are not positive. In total, 94 percent of the reviewed studies that had a positive outcome were published; whereas, for the entire set of studies (published and unpublished) only 51 percent were actually clearly positive. The authors argued that this produces an inflated perception of the efficacy of antidepressants. When the authors calculated the effect size (a mathematical measure of the magnitude of efficacy) of the entire set of studies versus those that were published, they found that the selective publication bias inflated it by 32 percent and that the effect size for the entire set of studies was a very modest 0.31.

Prima facie, this high-profile paper presents some very disheartening and unwelcome news about the main tool psychiatrists use to treat depression. Only 51 percent of short-term, randomized, placebo-controlled studies in depression produce clear evidence of efficacy for the antidepressant, and the overall magnitude of the effect is modest. The media, most notably the New York Times, were quick to pick up on this paper and its obvious implications. However, we should be careful not to over interpret the finding of this paper. Aside from some methodological problems that seem to exaggerate the bias, we should be aware that a similar publication bias has been reported for many other types of medications in other fields of medicine.[2] More important to emphasize, however, is that the analyzed studies were all short-term studies of depression (typically 6–12 weeks). These are the most common type of randomized, controlled studies conducted because they are the quickest and least expensive means of obtaining FDA approval. To be eligible for these short-term studies, patients must typically meet the criteria for major depressive disorder and be in the throes of an active depressive episode. Drug manufacturers specifically make the entry requirements a high score on a depression-rating scale, such as the Hamilton Depression Rating scale, because it is easier to show improvement in patients who are highly symptomatic compared to patients who are only modestly symptomatic. However, major depression is a disease characterized by recurrent acute episodes of depression that are typically self limiting. Furthermore, depressive episodes are highly responsive to positive psychosocial interventions, such as those created by a placebo-controlled clinical trial with weekly visits to healthcare providers. It is no wonder, therefore, that there is a very high placebo response rate in short-term clinical studies of depression,[3] significantly higher than other disorders in medicine, even other disorders in psychiatry, that are not as vulnerable to psychosocial aspects. This robust placebo response rate in short-term depression studies has, for many years, been the bane of pharmaceutical manufacturers of antidepressant medications who spend millions of dollars trying to demonstrate their drugs’ efficacy only to be undermined by large placebo response rates. It also means that the results of short-term, randomized, clinical trials of depression are not good barometers of the efficacy of antidepressants in clinical practice (where the psychosocial intervention is typically not as strong as it is in a clinical study).

To appreciate the efficacy of antidepressant medication, we would do better to look at a different category of clinical studies. “Continuation” or “relapse prevention” studies use a very different approach to measuring antidepressant efficacy than short-term antidepressant response studies. In these studies, patients whose depressive episodes have improved in response to antidepressant treatment (typically in an open-label fashion) are randomized, in a double-blinded fashion, to continue antidepressant medication or to be switched to placebo. Unlike the short-term response studies, these relapse-prevention studies typically last six months or longer and measure, not improvements in depressive symptoms, but the proportion of patients in each treatment that experience a depressive episode relapse (objectively defined beforehand). These studies are more complex and require more subjects to determine statistical significance and are significantly more costly than short-term studies as a result. Not surprisingly, therefore, these studies are performed much less frequently than the short-term response studies, especially since the short-term studies are sufficient to obtain FDA approval for a putative antidepressant. Nevertheless, maintenance studies produce a much stronger effect (larger effect size) than short-term studies. One of the reasons, no doubt, for this is that placebo response is greatly reduced in these studies since subjects entering the double-blind phase are already in a depression-free state.

So, while antidepressants have recently taken some heavy blows recently (e.g., their association with increased suicidal ideation in younger patients and the New England Journal of Medicine paper impugning their efficacy), a well-informed read of all the prevailing data reaffirms their efficacy and supports their continued use as the main weapon against depression.

References
1. Turner, EH, Matthews, AM, LInardatos E, et al. Selective publication of antidepressant trials and its influence on apparent efficacy. N Engl J Med 2008;358;3:252–60.
2. Abaid, LN, Grimes, DA, Schulz, KF. Reducing publication bias through trial registration. Obstetr Gynecol 2007;109(6):1434.
3. Berk M, Dodd S. Antidepressants and the placebo response. Hum psychopharmacol 2005;20(5);305–7.

Disclosures

Dr. Feifel has received research funding, consulting, or speaking fees from the following pharmaceutical companies: Abbott Laboratories, AstraZeneca, Argolyn Biosciences, Eli Lilly and Co., Bristol-Myers Squibb, Solvay, Janssen, Wyeth, McNeil, and Shire.