Two hands clap and there is a sound. What is the sound of one hand?
—Hakuin Ekaku (Rinzai Zen Master, 1686-1768)
From my perspective, we are witnessing what might be called a crisis of faith in the realm of antidepressant treatment. Now, “faith” is a curious term. It may imply something like “confidence”—or something more akin to “deep religious conviction.” My experience with a few recent blog postings has convinced me that both interpretations apply to the state of antidepressant research and treatment.
First, there is the issue of confidence, with respect to the efficacy of antidepressants (ADs) in both acute treatment studies and studies of long-term prophylaxis. (“Efficacy” usually refers to benefit under controlled, clinical conditions; “effectiveness,” to naturalistic settings). It turns out that the reported efficacy of ADs in acute treatment studies is, to a large degree, in the eye of the beholder—or of the statistical analyst. A great deal depends, for example, on how much stock we put in a 2- or 3-point change on the Hamilton Depression Rating Scale (HDRS, commonly known as the HAM-D) and how we understand the nature of the “placebo response.”
Nearly all acute efficacy studies of ADs make use of the HAM-D, which is actually subject to a variety of confounds, such as the training of the HAM-D raters.1 And, depending on whether one considers a mean change of 2 or 3 points on the HAM-D to be clinically significant, meta-analyses of acute AD efficacy studies lead to different conclusions. Furthermore, placebo responses have been rising in recent years because of a variety of factors, such as multisite studies and use of less severely depressed patients.2 It is hard to apply the results of randomized, placebo-controlled studies to clinical practice. Thus, in a superb review, Prof Hans-Jürgen Möller3 concludes:
“In interpreting such mean [HAM-D] score differences, it has to be stated that the mean of the pre-post differences of the placebo groups and the verum [active medication] groups only give a global estimation of efficacy under the artificial conditions of placebo-controlled trials, in which, due to the principal characteristics of the design, the verum response is underestimated and the placebo response is overvalued. . . . One cannot conclude very much from this for everyday clinical practice, especially not on the efficacy for special patient subgroups or even for individual patients.”
Furthermore, as an extensive critique of the much-discussed Kirsch4 meta-analysis pointed out,
“Perhaps the most overlooked aspect of the data included in Kirsch et al. (2008) is that it is not about the effects of antidepressants upon depression, but rather about the effects of antidepressants on the HAMD . . . [and the HAM-D] describes a construct of depression which corresponds poorly to that found in DSM-IV . . .[underline added]”5
Having recently reviewed the extensive antidepressant literature for a paper in press,6 my own conclusion is that ADs show, overall, modest-to-moderate efficacy compared with the placebo condition in acute treatment studies—depending on the type and severity of depression. I use the term “placebo condition” advisedly, since inclusion in this condition involves far more than popping a “sugar pill” into the patient’s mouth, as the lay press likes to imagine. Dr. Sheldon Preskorn7 has noted that in a typical randomized placebo-controlled trial, subjects in the placebo condition often receive 8 to 12 hours of supportive and educative care—more than many depressed patients in “the real world” receive from their doctors! We simply don’t know how this supportive component may modify the “intrinsic” properties of a placebo tablet—but my guess is, there is considerable synergism.
The most robust AD effects in acute treatment studies seem to emerge in more severely depressed patients (although there is controversy on this point) and among those with melancholic features.8 However, a recent re-analysis of the Kirsch data by Ghaemi and Vöhringer9 found acute antidepressant efficacy in both moderate and severe—but not in mild—major depression. This should not really surprise us. All other factors being equal, the farther we move down the continuum from severe melancholic major depression toward normal sadness, the less pronounced the difference between specific and nonspecific interventions (like the placebo condition). In partial contrast, long-term maintenance studies of ADs seem to show a more robust drug-placebo difference. As Möller3 notes:
“If the results of placebo-controlled studies regarding a maintenance therapy with antidepressants (maintenance of the response for 6-12 months after the acute therapy) are considered . . . the conclusion regarding the clinical relevance of antidepressants is even strengthened. Geddes et al (Geddes et al. 2003) in their meta-analysis of 31 randomized, double-blind, placebo-controlled studies found a highly significant efficacy of continuation therapy, with relapse rates of 41% under placebo versus 18% under verum [active medication] . . .”
To be sure, there are methodological issues that arise with many long-term studies; eg, the use of “enriched” samples (known responders to the medication) during the maintenance phase, which tends to bias outcomes in favor of the active drug.9 Furthermore, several researchers have questioned whether most “maintenance” studies are demonstrating true prophylaxis, or mere “relapse prevention.”10 Arguably, one would need to see prevention of depressive recurrence 6 months or more after the index depressive episode in order to show a true prophylactic effect. Nevertheless, reducing the risk of relapse in severe major depression, even for 6 months, is not a trivial benefit in a potentially lethal disorder.
While I endorse the randomized, double-blind, placebo-controlled trial (RCT) as the “gold standard” of scientific research, we also need more naturalistic, “real-world” studies of antidepressants. After all, most RCTs select “depressed” patients the likes of whom most of us rarely see: patients without active suicidal intentions, substance abuse, or comorbid conditions on Axes II and III, many of whom are recruited from nonclinical settings.
One encouraging “real-world” study comes from Leon et al,11 who carried out a 27-year observational study of antidepressants.11 The authors concluded that “. . . antidepressants were associated with a significant reduction in the risk of suicidal behavior,” which was defined as a reduction in actual suicide attempts or completed suicides while taking ADs. This is consistent with most epidemiological studies and shows an association between AD prescription and stable or declining suicide rates.12 If the Leon et al findings are confirmed, we will have convincing evidence that ADs are providing a substantial benefit to depressed patients, in “real-world” conditions.
The other dimension in this crisis of faith involves the more religious connotation I alluded to earlier—and here we must depart the realm of “pure science.” It is clear to me that among the general public, there is a vociferous and sometimes vitriolic group who fervently and inflexibly believe that antidepressants are worse than useless—indeed, who see these medications as quite toxic and dangerous. My sense is that many of these individuals fit the description of the “true believer,” as described by philosopher Eric Hoffer13(p76):
“It is the true believer’s ability to “shut his eyes and stop his ears” to facts that do not deserve to be either seen or heard which is the source of his unequaled fortitude and constancy.”
(Of course, characteristically, these opponents of antidepressants view psychiatrists as having precisely the same blinkered obstinacy.) Hoffer13(p86) goes on to say that
“Mass movements can rise and spread without belief in a God, but never without belief in a devil. Usually the strength of a mass movement is proportionate to the vividness and tangibility of its devil.”
Indeed, this loosely knit community of naysayers sees psychiatry and psychiatrists in nearly satanic terms. For them, however, the “devil” is a Chimera-like creature, with the head of a fraudulent researcher; the body of a drug sales rep; and the tail of a psychiatrist. I do not intend to publicize the Web sites of these individuals, but you can read some of their comments by linking to the blog sites below.*13,14
Now, it is tempting to write these individuals off as fear-mongering cranks, possessed of only the faintest scientific knowledge. A handful certainly fit that description—the ones, for example, who insist that antidepressants are destroying the brains of millions; causing thousands of suicides; and turning scores of once-placid accountants into knife-wielding, psychotic killers. Oh, yes—and causing permanent sexual dysfunction in thousands of patients. (The actual prevalence of this last phenomenon is not known, though, to my knowledge, there are fewer than 30 such cases in the published literature—see references 15 and 16.) These individuals will probably never be persuaded, no matter the evidence, that antidepressants are usually effective and well-tolerated when properly prescribed and monitored.
And yet, dismissing all critics of psychiatric treatment as querulous crackpots would be a serious mistake. Some of those who wrote to me were both knowledgeable about psychiatric medications, and sophisticated in their grasp of medical research. Some spoke from painful personal experience with psychiatric medications—whether antidepressants, antipsychotics, or mood stabilizers. They spoke, for example, of becoming agitated or manic while taking antidepressants, and feeling depressed or “doped up” while taking mood stabilizers. They spoke of painful “withdrawal symptoms” lasting many months, after their antidepressant was stopped. They spoke of lethargy, blunted creativity, or impaired cognition while taking antidepressants or mood stabilizers. Perhaps most disheartening, they spoke of how little they felt understood, “listened to,” or respected by their physicians.