Why antidepressant clinical trials fail: the role of expectations

The final decades of the 20th century may be considered the golden age of anti­ depressant medications. There was a boom in their development and clinical use as well as a sense that they might represent a major advance towards more targeted treatment for patients suffering from depression. However, recent evidence point­ ing to the failure of many antidepressant randomized controlled trials (RCTs), particularly for depression of mild­to­moderate severity [1–3], has shaken the field. Multiple factors may explain these less than promising findings. First, we now know that the antidepressant literature has been biased as a result of selective publication of positive trials [4]. There is also an increasing awareness of the het­ erogeneity of what we understand to be major depressive disorder. Culture and ethnicity, for example, have an important impact on response to psychopharma­ cology [5], to the extent that trials conducted in different places may yield different results. Indeed, there is even a question of whether our current definition of major depression represents a distinct entity or rather multiple disorders with only some responding to antidepressants [6]. However, the core issue in failed antidepres­ sant trials may not be the lack of response to active treatment, but the narrowing separation between drug and placebo, accounted for in large part by rising placebo response rates [7]. One important, yet underappreciated factor, contributing to placebo response rates is the manner in which patient expectations significantly influence study results. The reason expectations are so important for psychiatric conditions such as depression and anxiety, as well as a variety of neurological conditions such as pain syndromes, is that people with these conditions are much more likely to improve if they expect to do so. For example, one study that measured the relationship between pretreatment expectations and outcome in a group of patients treated with reboxetine for major depressive disorder, found that 90% of patients who initially expected reboxetine to be ‘very effective’ responded by the end of the trial compared with only 33% of patients who only expected it to be ‘somewhat effective’ [8]. Therefore, expectations are extremely important, and researchers and physicians are modulating them all the time, often without being aware of it. A few years ago, researchers at the California Institute of Technology studied sub­ jects tasting wine and found that both their subjective and objective experience (the latter demonstrated by functional MRI) were more positive when they were told the wine they were drinking was more expensive [9]. Subjects do a similar kind of unconscious calculation in randomized clinical trials. There is a broad literature in medicine demonstrating that research subjects generally expect drugs to be stronger and therefore respond more often to larger pills, colored rather than white pills, capsules rather than tablets and injections rather than pills [10,11].