In her editorial, ‘The antidepressant debate’, Moncrieff (Reference Moncrieff2002) provocatively questioned the orthodox view that antidepressants are efficacious (i.e. work under clinical trial conditions) in the treatment of depressive illness. Questioning accepted views is valuable but Moncrieff missed the real question, which relates to effectiveness, that is when are antidepressants useful clinically? The efficacy argument at the head of her critique, based on individual, often old and poor-quality, studies flies in the face of consistent findings of antidepressant efficacy in systematic reviews and meta-analyses (e.g. Reference Anderson, Nutt and DeakinAnderson et al, 2000). Even the argument of bias due to unblinding because of side-effects is contradicted by her own meta-analysis, which showed a significant benefit for antidepressants over ‘active’ placebo (Reference Moncrieff, Wessely and HardyMoncrieff et al, 1998). Even more compelling is the evidence from continuation/maintenance studies which show that antidepressants have a robust effect in reducing rates of relapse and recurrence (Reference Carney, Geddes and DaviesCarney et al, 2001), a cumulative effect over months or years. Explaining this by a placebo effect is difficult to accept, or else demands re-evaluation of the nature of placebo.

This is not to say that ‘negative’ studies, where antidepressants are no better than placebo, should be ignored. An important factor is probably related to severity of depression. Khan et al (Reference Khan, Leventhal and Khan2002) found that the proportion of studies favouring antidepressants over placebo increased with the severity of depression; the response to placebo declined with increasing severity whereas that to antidepressants increased. This raises the fundamental question of when (i.e. at what severity) in real life practice does someone with depression clearly benefit from antidepressant drug treatment. Put another way, is the current trend to wider use of antidepressants for milder depression justified? This can only be answered empirically in appropriate naturalistic trials, and even then will require value judgement about the size of the benefit.

Perhaps the most worrying aspect of Moncrieff's editorial was the implication that we should take either a psychosocial or a physical approach to the treatment of depression. Surely we should have put this rather tired dualist view of psychiatry behind us by now? A holistic view combining drug and psychological treatments is to be preferred and evidence is accumulating that this leads to better outcomes. To conclude, a balanced view of the evidence for antidepressants firmly places them as an established and important therapeutic option (alongside others) in the treatment of depression, with their role becoming more central with increasing severity. The true debate is about the best way to use them.