Abstract
Research into what works in depression is in trouble. All treatments appear equally effective and what happens in research trials does not seem to translate into clinical practice or, to put it another way, clinical practice is not reflected in randomized controlled trials. We run the risk of designing cosmetic trials with no real relationship to clinical services that only serve to demonstrate that new treatments are effective in artificial settings. This article outlines possible reasons for this, which include how depression is diagnosed, how people are recruited into trials, the design of trials and how outcomes are assessed. The solutions to these problems involve changing how patients are recruited, considering other trial designs and using different outcome measures.