Verified by 
When there’s a gap between what practicing doctors observe and what formal research demonstrates, we tend to doubt the doctors. Caregivers, so we think, overestimate the effects of their treatments.
But what if—sometimes—it’s the research that misleads, and in consistent fashion, in study after study? A recent small drug trial suggests that the testing of antidepressants is liable to a distortion that can cause experimenters to miss realities that practitioners see every day.
The trial, lead by a team at Columbia University and elsewhere, was designed to explore the effects of hopeful expectations on responses to antidepressants. The patient sample was small—50 people in all, arguably too few to allow us to consider the trial randomized—but the set-up was interesting.
Four patients got placebo. Twenty received Celexa (an antidepressant) and were told that they might be on placebo. Twenty-six were given Celexa “openly”; they knew that they were getting a helpful medication.
Scientists who believe that mood disorder responds to placebo often claim that the effects of hopeful expectations are especially strong for patients with mild to moderate major depression. That’s who were admitted to this study.
Two months out, only one patient (of the four) on placebo had shown a response. For the patients given an antidepressant but left in doubt, the response rate was 45 percent. The “open trial” of the antidepressant produced a higher response rate, 54 percent.
Knowing for sure that you were on medication was helpful. The publication did not express outcomes via the most commonly used questionnaire, the 17-item Hamilton depression rating scale, but it looked as though certainty that you were on a real drug resulted in one or two symptoms’ worth of additional relief among the items that the standard measure covers. That difference is considered important. The result should lead us to suspect that observers of drug trials (where participants know that they may be taking placebo) will see antidepressant responses that underperform substantially, relative to what office practitioners (who prescribe openly) will notice in their patients.
In the Celexa trial, the course of recovery was telling. If you believe in placebo effects, ones that operate through hopeful expectancy, you might imagine that people told that they were on Celexa would do well right out of the gate. They did not. For four weeks or more, Celexa’s beneficial effects were uniform—more or less identical for those who were and were not sure about what they had been offered.
The report’s sole graph indicates that taking Celexa openly first gave clear extra benefit at the six-week mark. In the text, the authors report a statistically significant outperformance—the loss of additional symptoms among those not cast in doubt—at eight weeks. How shall we interpret this pattern?
The traditional take on antidepressants is that their major effect—a notable drop in the level of depression—occurs in this time frame, starting at or after four weeks. It looks as if, for people with non-severe major depression, awareness of being on medication—knowing that they are getting solid help—only makes a difference once the antidepressant has brought about considerable improvement on its own.
Starting in the 1990s, researchers noted that patients in trials comparing two effective drugs—trials in which patients were certain that they were on useful medication—fared better than patients in trials comparing a drug and a placebo pill. The phenomenon was observed in studies of depression and studies of Parkinson’s disease. The findings inspired the coinage “lessebo effect,” to encapsulate the corrosive influence that doubt has in muting the impact of effective medication.
In my current book, Ordinarily Well, I pair a discussion of lessebo effects with an introduction to the theories of an Oxford psychiatrist, Philip Cowen, about recovery from depression. Cowen believes that antidepressants create the context for progress. In early going, they begin to relieve patients of hopelessness—of a fixed negative perspective on the self and their surroundings. In response, patients behave differently. They re-enter the social realm, function better at work, receive encouragement, make note of the progress, and so on—amplifying and completing the effects of medication.
Cowen writes, “According to this view, [favorable] effects of antidepressants on emotional bias are seen rapidly, but the translation of these changes into improved subjective mood takes time as the patient learns to respond to this new, more positive social and emotional perspective of the world.”
But what if patients doubt that they’re getting genuine help? The lessebo effect will disincline them to elaborate on early improvement due directly to medication. Doubting patients will be less apt to phone friends and so on.
In their report, the Columbia-led authors acknowledge this sort of interpretation: “[E]xpectancy may lead to more gradual depression improvement by mean of indirect mechanisms, such as increasing behavioral activation, improving medication compliance, or enhancing the therapeutic alliance between the patient and doctor. Given the delayed time course, data from this study appear to suggest that indirect mechanisms are likely to play a role in mediating expectancy effects on depression.”
The researchers—the team includes some insistent exponents of the placebo effect—offer alternative explanations that are friendlier to the notion that pills works via hopeful expectancy. But if Cowen-style elaboration and (interfering with it) lessebo effects are in play, there is no need to invoke placebo action to explain the trial’s results. The Columbia experiment’s main finding is that openness about who’s on what makes a difference only in the time frame usually associated with prior antidepressant action.
If uncertainty interferes with patients’ propensity to translate early progress into substantial change, then traditional randomized trials, ones that employ dummy pills, will produce estimates of efficacy that run low. It’s not that practitioners are fooled, it’s that experimenters may be, because the design of standard drug trials puts an artificial drag on patients in their course of recovery.
One further thought: the new study suggests that traditional research may distort our overall understanding of antidepressants. In Ordinarily Well, after presenting Cowen’s theories, I open a discussion of “inherent efficacy.” When an antibiotic begins the process of killing bacteria and the patient’s immune system finishes the job, how much of the resulting recovery will we attribute to the inherent efficacy of the drug? Usually, all of it. When an antidepressant restores resilience to the brain or mind, freeing a patient to behave (at home and on the job) in ways that further combat a mood disorder, how much of any resulting remission should we attribute to the medication’s power to heal? The question is difficult, but a case can be made for the simplest answer: all of it.
To be sure, by this account an intact immune system and a reasonably benign social environment are required to speed healing. But we measure treatments according to how they act under expectable conditions.
Randomized trials are meant to test a treatment’s inherent efficacy. When it comes to antidepressant medication, they may fail at that job. The research design, which includes the use of dummy pills, induces doubt, a confounding factor that blinds experimenters to the impact of a key factor in inherent efficacy, patients’ more confident re-entry into the world.
Postscript: A reader has correctly commented that a 54 percent response rate is nothing to write home about. This posting is not about the absolute effectiveness of antidepressants, but, reading the numbers, the concern is a natural one. The low figure relates to the current state of affairs in drug testing, which I discuss at length in Ordinarily Well. (I take a visit to a commercial drug testing site in the suburbs and go for a ride in the van that fetches study participants from the inner city.)
Because people with access to health care also have reliable access to effective generic antidepressants, it has become hard to recruit representative patients into studies where there is a chance of receiving placebo. An earlier open trial—no placebo—in primary care settings gives what is arguably a fairer account of Celexa's clinical efficacy. I don't want to make the case for "completer" data (calculations based on patients who participate for the entire study), but by the six-month mark, 93% of patients who stayed with the program had responded--so that's the high-end figure for response rates to Celexa.
