The key papers on this are listed here: https://scholar.google.co.uk/scholar?q=kirsch+antidepressant+meta+analysis&hl=en&as_sdt=0&as_vis=1&oi=scholart&sa=X&ei=

The key papers on this are listed here: https://scholar.google.co.uk/scholar?q=kirsch+antidepressant+meta+analysis&hl=en&as_sdt=0&as_vis=1&oi=scholart&sa=X&ei=UMk4VZXnOJLfasv_gPgD&ved=0CB8QgQMwAA https://scholar.google.co.uk/scholar?q=kirsch+antidepressant... About 70-80% of the drug effect is replicated in the placebo arm. Other things to be worried about: - Unpublished data was much less likely to show a benefit for SSRI than published data. - SSRIs have side effects which may increase expectation of benefit in patients - Numerous studies show unblinding is very common in trials of psychoactive drugs. These studies are not a 'gold standard' by any stretch of the imagination.