Questions in Depression: Does the High Rate of Placebo Response in Antidepressants Matter?

Gertrude Conaway Vanderbilt Professor of Psychology, Vanderbilt University

First published in Psychiatry Weekly, Volume 6, Issue 18, September 5, 2011

This interview was conducted on July 7, 2011 by Lonnie Stoltzfoos

Introduction

People who seek treatment for depression often present with a broad, albeit well-packaged, variety of symptoms at baseline. These moderately to severely depressed patients usually experience sad affect or negative self-concept, loss of appetite or interest in sex, and difficulty with sleep and getting things done the way they used to.

Most patients being treated for depression receive pharmacological treatment. A widely reported study in 2009, by Olfson and Marcus, found that 10% of Americans over 6 years of age receive antidepressants over the course of 1 year, nearly double the rate from a decade earlier. Only 27% of those prescriptions were written specifically to treat depression (usually by primary care physicians), and 20% of patients treated with antidepressants for depression also receive some form of psychotherapy. With their widely prevalent use for treating a great variety of health conditions, antidepressants could appear to be a modern pharmacological miracle. Yet, a number of studies have found that the placebo response rates of antidepressants remain notoriously high, at least with respect to depression. Is the SSRI class suffering from its overachievements, or has it been stretched unfairly to a breaking point?

Depression Severity and Specificity of Treatment

“We know that antidepressant treatments, in general, work for 60% to 70% of patients,” says Dr. Steven Hollon, “and many of those patients will improve on any given medication. However, for around half of the people who meet the criteria for MDD, you won’t get the medication to separate from placebo; greater severity of depression does predict the specificity of response. Most patients will improve, mostly for non-pharmacological reasons. The same holds true for psychotherapy, in that many improvements occur for non-specific reasons, although that body of literature is much smaller. This does not mean that patients don’t get better if treated; only that, until you get farther up the severity continuum, non-specific aspects of treatment carry a big chunk of the weight of clinical improvements.”

Indeed, a 2010 JAMA report that Dr. Hollon co-authored compared the placebo response with medication response in 6 placebo-controlled antidepressant trials with a total of over 700 patients. The review concentrated on the medication response rates relative to baseline depression severity, as measured by the Hamilton Rating Scale for depression (HAM-D). All told, antidepressants had a Cohen d effect size of 0.11 in mild to moderate depression (HAM-D score ≤18). Severe depression had a 0.17 effect size (HAM-D, 19–22). In very severe depression, however, antidepressants had an effect size of 0.47 (HAM-D, ≥23), just below the 0.50 cutoff for a medium effect size.

"The specificity of medication effects is not as great as many believe, but that doesn’t mean the treatments don’t work—they often work for non-specific reasons."

If these perhaps unflattering numbers were a surprise to many people, Dr. Hollon points out that publication bias has given way to falsely high expectations of the measurable and specific benefits of antidepressants.

“If you refer to the published literature, ~95% of the trials for medications are positive and supportive of the medications,” he says. “If you refer to the FDA database where those trials come from, only ~50% of the trials are positive. It turns out that most of the trials in which medications do well end up in the literature, and most of the trials in which medications don’t do well, don’t. Two important reviews on publication bias, by Kirsch and colleagues, and Turner and colleagues, both from 2008, found an actual medication d effect size of 0.32 and 0.31, respectively, for antidepressants when compared to placebo.”

Clinical Implications, Conclusion

“Depression is probably a chronic and recurrent disorder, and it’s important for people to get treatment if they are depressed,” says Dr. Hollon. “The specificity of medication effects is not as great as many people think it is, but that doesn’t mean the treatments don’t work—they often work for non-specific reasons. Most good psychiatrists come to believe, anecdotally, that certain people will respond better to some medications than others, in terms of acute treatment. Over the long term, however, you start to get more lasting, enduring effects with some cognitive and behavioral therapies.

“I would say that most people are probably better off if they start off on a psychological intervention and add medication if they need it, although it usually works the other way around,” continues Dr. Hollon. “Although concomitant psychotherapy and medication has not demonstrated a striking benefit over either treatment modality alone, this scenario increases the patient’s odds for improvement, with a response rate of ~75%, compared to 60%–70% with a psychotherapy or medication alone. Let’s say that psychotherapy isn’t competently done, or the medication is poorly chosen—there is still a likelihood of one treatment modality compensating for any inadequacies in the other.”

Disclosure: Dr. Hollon has received research funding from the NIMH (MH60173).

 

 Tweet