The recent editorial by Gavin Andrews (Reference Andrews2001) omitted some important considerations on the discussion of placebo response in clinical trials of antidepressants. These include the obligations of regulatory authorities in the appraisal of new treatments, numerous research-specific factors that contribute to the placebo response in a research environment, and the contribution of the scales currently employed as primary efficacy measures in depression trials to test the null hypothesis.

The demand by regulatory authorities for placebo-controlled trials in the evaluation of antidepressant therapies is supported by data published by Paul Leber, former Director of the Neuropharmacologic Products Division of the US Food and Drug Administration. Leber (Reference Leber1989, Reference Leber1991) cited research studies where the antidepressant test agent was as effective as an already-approved active control medication, but in five of the six studies he cited, both were inferior to placebo. Given the notoriously poor sensitivity and interrater reliability of the Hamilton Rating Scale for Depression (HRSD), the usual primary efficacy instrument in US antidepressant clinical trials, this is not surprising to investigators who use the HRSD on a day-to-day basis. However, based on Leber's evidence alone, if regulatory agencies move toward the Declaration of Helsinki (World Medical Association, 2000) mandated non-inferiority trials as a basis for approving new drugs, then it is only a question of time before regulatory approval is given to drugs that very well might be less effective than placebo. The regulatory approval of an-inferior-to-placebo drug would ultimately be harmful to the large number of patients who would take this ineffective drug, in part being persuaded to do so on the basis of regulatory agency assurance of its efficacy. As all drugs have some side-effects in some patients, regulatory agencies must be able to affirm that a drug has been demonstrated to be better than ‘nothing’ (i.e. placebo) in conditions where ‘nothing’ has demonstrated benefits.

Andrews' summary of the mechanisms of placebo response in antidepressant clinical trials also omits several important considerations. Spontaneous remission of depression with time, the natural fluctuations of a chronic illness, and the encouragement that comes with being treated were the only factors he cited as contributors to placebo response. He does not cite the anxiety-lowering effect of receiving of a definitive diagnosis from a trusted expert physician in a clinical trial, and the increased sense of mastery and control that comes from patients' greater understanding of their illness as a result of the more unhurried and usually very extensive evaluation that occurs in clinical research practice compared with standard medical practice. He does not consider that these aspects of placebo response might be inherent in the environment of clinical trials, but are not generalisable to day-to-day general or psychiatric practice (although these differences are difficult to assess quantitatively). Other factors of the placebo response in clinical research are potentially amenable to change. These include potentially unhelpful-to-research, overly encouraging behaviours of those conducting the clinical trial, and false, overly optimistic patient assumptions and expectations about their outcome in the research trial, leading to inaccurate reporting of symptoms and thereby excessive response in those patients randomised to placebo. It is likely that increased efforts in patient and investigator education about how both patients and study site staff can be helpful in forming a productive research alliance and not generate ‘wishful thinking’ and the overly positive responses that might accompany it, is needed to reduce the costly, wasteful number of failed trials caused by excessive placebo response. In response to this need, I have recently proposed PREECT (Patient and Rater Education about Expectations in Clinical Trials), a two-component approach to the problem (Reference ZimbroffZimbroff, 2001). Briefly, the first component focuses on ensuring that clinical trial participants understand that they are entering a research alliance — not receiving regular medical care. The second component involves education of study site staff. Both anecdotal feedback and data lend support to the contention that PREECT reduces placebo response rates and, thus, could ultimately reduce the numbers of patients needed in placebo-controlled trials to achieve sufficient power to test the null hypothesis. Further exploration and validation of the PREECT approach would benefit patients and researchers involved in antidepressant trials, and would reduce the likelihood of an excessive placebo response.

Finally, a critique of the scales used as primary efficacy measures needs to be considered. This in itself is a thesis. The very fact that regulatory authorities in the USA are collaborating with researchers on tackling this issue lends support to its importance as a consideration in appraising this topic.