Do antidepressants work? A commentary on "Initial severity and antidepressant benefits: a meta-analysis of data submitted to the Food and Drug Administration" by Kirsch et al.

tial multicentre ones not funded by industry, which together have been cited over 800 times in the literature already, that have confirmed that there are indeed no special advantages of atypical antipsychotics in the treatment of schizophrenia and, compared with conventional anti-psychotic drugs, they are not costeffective. I deliberately put the word ‘‘confirmed’’ in italics because the wording conventionally is the wrong way round—meta-analyses with their higher level of evidence should do the confirming, not the individual trials. The earlier meta-analysis may have only had a small part to play in stimulating these two major studies but I judge that it did make an important one, because when a metaanalysis throws up a result that is against expectations everyone has to take notice. So although systematic reviews and meta-analyses have an important place I do not think they should be at the pinnacle of the evidence-based table. Their best function is to point to the unsatisfactory state of affairs with regard to the evidence base and indicate the sort of studies that would provide more definitive answers. A meta-analysis is therefore the precursor of the definitive trial; if researchers, funders and editors all remember this they will get these trials into proper perspective.