Ever imagine that the Freedom of Information Act would be used to gain access to randomized controlled clinical trials information about the efficacy of antidepressant medication? Well, I didn't—and neither, I suspect, did the pharmaceutical manufacturers of the six most widely prescribed antidepressants approved by the Food and Drug Administration (FDA) between 1987 and 1999. Nevertheless, in the article “The Emperor's New Drugs: An Analysis of Antidepressant Medication Data Submitted to the U.S. Food and Drug Administration,” an enterprising group of scientists headed by Irving Kirsch with colleagues Thomas Moore, Alan Scoboria, and Sarah Nicholls reported in the July 2002 issue of Prevention and Treatment that they petitioned the FDA and were granted access to the medical and statistical evaluations of “every placebo[-]controlled clinical trial for depression … submitted for approval.” Kirsch et al. note that the only antidepressant medications that made the “most frequently prescribed list” were six selective serotonin reuptake inhibitors (SSRIs), including citalopram (Celexa), fluoxetine (Prozac), nefazodone (Serzone), paroxetine (Paxil), sertraline (Zoloft), and venlafaxine (Effexor).
We are left to wonder why the authors were obliged to fetch the antidepressant effectiveness data in this unusual way. What their straightforward re-analysis of the pharmaceutically sponsored and FDA-approved efficacy data revealed about the pharmacological treatment of depressive illness makes their research efforts more than worth the trouble to which they must have gone.
The Kirsch team reported that the FDA reviews were based on findings from forty-seven randomized, placebo-controlled trials of the six antidepressants. Each clinical trial had been conducted to support the various drug manufacturers' efficacy claims and to gain FDA approval for the pharmaceutical treatment of depression. Recall that evaluating a drug's consequences in a controlled clinical trial involves (1) prescribing an experimental pharmaceutical to a group of depressed volunteer participants, (2) prescribing a standard pharmaceutical or a chemically inert substance (the placebo) to a second participant group, and then (3) comparing their effects on a validated depression measure. In the typical FDA-approved “double-blind” controlled clinical trial, neither the prescribing physician nor the participant knows which substance is being administered.
The implications of the re-analysis of the pharmaceutical industries' controlled clinical trial data for the pharmacological treatment of depressive illness will be debated for some time to come.
Since the same standardized clinician-administered depression measure, the Hamilton Depression Rating Scale, had been used in all studies to estimate severity of depressive illness, improvement rates could be computed and compared across clinical trials in the Kirsch re-analysis.
The research team's evaluation of the pharmaceutical industries' controlled clinical trials data indicated that improvement among the placebo-treated depressed patients was 82 percent of that found among patients who actually received one of the six SSRIs. They interpreted this finding to mean that only“ 18% of the drug response [was] due to the pharmacological effects of the medication.” While the magnitude of depressive symptom improvement attributable to the active drug agent was large enough to achieve statistical significance, the average measured improvement for patients receiving an SSRI was just a couple of points greater on the Hamilton Depression Rating Scale than for those taking the placebo. This fact led Kirsch et al. to question the clinical utility of adding an SSRI in the treatment of depression and to assert that “by far, the greatest part of the change is also observed among patients treated with inert placebo. The active agent enhances this effect, but to a degree, that may be clinically meaningless.”
The implications of the re-analysis of the pharmaceutical industries' controlled clinical trial data for the pharmacological treatment of depressive illness will be debated for some time to come. The editors of Prevention and Treatment invited nine prominent psychopharmacologists to comment on the methods and findings of the Kirsch research team. While there were various interpretations as to why the magnitude of the placebo effect was so large and the antidepressant drug effect so small, there was no debate about the major finding itself. In their reply to the commentators, Kirsch et al. emphasize the difficulty the drug companies encountered in attempting to demonstrate an antidepressant drug's effect:
57% of the trials funded by the pharmaceutical industry failed to show a [statistically] significant difference between drug and placebo. Most of these negative data were not published … and were accessible only by gaining access to U.S. Food and Drug Administration … documents. [italics added]
“Placebo” is the first person singular future indicative of the Latin verb placere. It is translated as “I shall please” or “I shall be acceptable.” Etymologically speaking, placebo effects are produced by the patient's wish to please or to be acceptable to the doctor and by the doctor's evoked desire to relieve the patient's pain. Kirsch's analysis showed that in the brief pharmacological treatment of depressive illness, the placebo effect was remarkably strong. In considering the meanings of these findings, it is instructive to recall that Freud's procedural transition from hypnosis to free association was guided by his understanding of the role of placebo effects in symptom development, maintenance, and remission. He gave the placebo phenomenon a more functional name by calling it transference. He never doubted for a moment the power of its effects.