This article reviews how data on the benefits and hazards of antidepressants have been analysed, and how conclusions drawn from these analyses conflict with the data. Randomised trials of antidepressants have for two decades consistently shown evidence of an increased risk of suicidal acts on active treatment compared with placebo, but an inappropriate application of significance testing has led to this evidence being dismissed. During the same period a minority of antidepressant trials have produced data indicative of benefits that have reached statistical significance at a 95% level. In this case significance testing appears to have led to an unrealistic impression of the likely benefits of treatment in practice. Current approaches to evidence-based medicine risk perpetuating misunderstandings of this type. Against a background of current developments in healthcare delivery, clinicians might need to reconsider how they handle and present clinical trial data.
Posted by gruwell at September 4, 2006 6:51 AM | TrackBack