The placebo effect has always been an intriguing topic in medicine, probably because of both its clinical implications and its philosophical correlates for the mind-body interaction . From a scientific point of view, our knowledge of the biological mechanisms for response to placebo has recently increased, thanks to more rigorous and systematic investigations . Accumulating evidence suggests that the placebo effect is a genuine psychobiological phenomenon attributable to the overall therapeutic context, which includes the interaction between the patient, the clinician and the treatment environment .
The placebo effect has been explored across a variety of diseases and medical conditions but it is in psychiatry - and most of all in depression - that the placebo response may play a major clinical role. This has generated concerns and a long-lasting debate around the real efficacy of antidepressants, with very conflicting opinions [4, 5]. In antidepressant trials with adults, the mean response rate for active treatment is 50%, while the mean placebo response is 31%, with an absolute difference of 19% . This difference is even smaller if unpublished data are included in the analysis  or if special populations, like children and adolescents, are considered . Interestingly, the placebo response has been increasing over the past 30 years and this increase does not appear to be directly explained by changes in study characteristics . The real question, then, is not whether the placebo effect can influence clinical outcome but rather the extent to which it does so. A study by Naudet et al. published in BMC Medicine investigates this issue within the frame of a network meta-analysis .
Accepting that randomized evidence shows both antidepressants and placebo to be effective in major depression, and that different antidepressants have different efficacy profiles, Naudet et al. wanted to investigate whether the effects of placebo may vary in different situations . As acknowledged by the authors, this is more an epistemological question than a pragmatic clinical issue. However, Naudet and colleagues approached it scientifically and performed a review of the literature comparing the placebos of two widely prescribed antidepressants: fluoxetine and venlafaxine.
Their primary aim was to compare placebo arms in fluoxetine and venlafaxine placebo-controlled studies, so the authors looked at three different types of placebo: fluoxetine placebo (in studies where placebo was compared to fluoxetine), venlafaxine placebo (where placebo was compared to venlafaxine), and fluoxetine/venlafaxine placebo (where placebo was compared in the same trial with both venlafaxine and fluoxetine). The authors find that the two antidepressant agents are more efficacious than the placebos and the three placebos do not differ in terms of response or remission. As has been previously reported, venlafaxine is more efficacious than fluoxetine, but the funnel plots show some evidence of publication bias. Emphasising this last point - and almost neglecting the primary finding of no difference between placebos - Naudet and colleagues argue that no one can be sure placebo really equals placebo in trials of major depressive disorder. Following this theoretical reasoning, they warn clinicians to ‘step back to take a more objective view when interpreting a scientific result’, because ‘Science can never be actually sure that “sucrose = sucrose” in the treatment of major depressive disorder’. For Naudet and colleagues, these clinical implications are more important than the differences in efficacy between antidepressants (which, in our view, is much more likely and clinically very meaningful).
Strengths and limitations of the study
The main strength of this paper is the authors’ attempt to look at this clinical scenario in a novel way. Scientific knowledge in medicine can progress only when people try to see things from a different perspective, pursuing the truth and avoiding easy answers .
The primary results from this study are consistent with evidence on antidepressants in moderate to severe acute major depression: some antidepressants are more effective than placebo  and material differences in efficacy exist between these drugs . The clinical interpretation of these findings may vary  and it is worth remembering that the practice of evidence-based medicine implies the integration of the best available research data with individual clinical expertise and patients’ preferences and values . Evidence-based practice is not ‘cookbook’ medicine and, similarly, ranking antidepressants is not aimed at providing clinicians and patients with a list of drugs to mechanically scroll down from top to bottom. The crucial point is whether to insist on a conservative approach that considers antidepressants as a group of equivalent compounds (thus favouring the prescription of the last one put on the market) or to try to make the best use of the available evidence. We still think that valid conclusions can be drawn from a critical and cautious use of the best available, if flawed, evidence.
Currently, placebos are not an alternative for our patients when an antidepressant is indicated for moderate to severe depression. In such cases, we all prescribe an active treatment. However, when discussing the treatment plan with our patients, it would be wrong and probably clinically irresponsible to say that antidepressants are all the same and to give our patients any antidepressant, just because it is licensed and the side-effect profile is more suitable for that individual. There is evidence supporting that some drugs licensed for depression are clearly less effective than others  and that some do not differ from placebo . Even though the decisional process varies case by case and we may end up making a completely different choice, as clinical researchers this is the best information we can give to help patients and clinicians when an antidepressant is to be prescribed for a first episode of acute major depression.