Authors' reply

Huneke & Baldwin raise important points regarding the interpretation of our study results and the relationship of our findings to the broader placebo literature. It is challenging to compare the results from our study with the literature cited by them. As they note, studies of placebo analgesia generally are performed in healthy volunteers not being treated for a chronic illness. Such studies examine the placebo effect, namely the relief of transient, experimentally induced symptoms during manipulation of expectations. By contrast, our study examined placebo response, which involves relief of naturally occurring symptoms of a chronic illness (in this case major depressive disorder, or MDD) within the context of a clinical trial. Because patients with MDD have long courses of illness and treatment, they commonly enter treatment studies with pre-existing expectations and beliefs, and our participants had indeed formed expectations about medications at the time of study enrolment. We concluded that these expectations were probably formed by factors external to the study, and speculated on the role that external social factors (such as direct-to-consumer advertising) might have had in forming positive medication expectations.

Huneke & Baldwin raise the point that our participants' medication expectations, which predicted placebo response, may have been formed, at least in part, by the consent process and initial exposure to the study environment. Participants consented and had their introduction to study personnel prior to rating their expectations of improvement. Although we do not know to what extent medication expectations might have been influenced by this initial exposure, a significant effect is unlikely. Participants rated the degree to which they expected that treatment in general, and medication in particular, would be helpful in relieving their depression. If participants' initial exposure to the study milieu shaped expectations, it would be expected to influence both medication and general treatment expectations. Yet, only participants' ratings of medication expectations predicted response to placebo. The selective relationship between medication, but not general, treatment expectations and placebo outcome suggests the influence of a process outside of the study milieu.

We agree with Huneke & Baldwin that it would be instructive to learn more about participants' previous experiences with antidepressant treatment and how this might affect current medication expectations, as well as the likelihood of placebo response. In this regard, we recently examined the potential role of prior antidepressant treatment and placebo treatment response in these same participants. ¹ Self-report data collected from a subset of participants from the parent study revealed that previous experience with antidepressant medication was significantly associated with poorer response to placebo. Interestingly, among those who had received prior antidepressant treatment, their self-report of response to prior treatment was not significantly related to expectations in the current trial or to placebo outcome. This finding suggests that antidepressant-experienced participants may show classic conditioning effects, consistent with our previously reported findings. ² The finding that prior antidepressant exposure, regardless of response, predicts placebo outcome is worthy of future study.