[Risk-benefit ratio of drugs. Antidepressive agents as an example].

Medical judgments in therapeutics must be comparative. The toxicity of a drug cannot be considered independently of that of other treatments for the same condition. Such comparison requires new specific methods in order to assess the quantity, quality and validity of available data. A comprehensive analysis is of paramount importance: ideally, all the advantages and disadvantages of different treatments should be simultaneously compared. As far as antidepressants are concerned, the quality of life and the risk of overdose should be borne in mind. Such metaanalyses lead to the identification of a variety of biases likely to distort our judgments related to the risk/benefit ratio of drugs: the bias of perception and of reporting are amongst the most important.