[Comparative risks and benefits of drugs. Methodological problems]

A focalizing toxicity is defined here as a specific toxicity which attracts much of the attention devoted to the safety of a drug: the hematotoxicity of "second generation" antidepressants is a typical example of such toxicity. The way in which it is handled is frequently biased, whether this be related to its perception or its reporting; as a result, it hampers the process of assessing the risk/benefit ratio of drugs. An attempt is made here to sketch the principles and methods of the validation of safety data; antidepressants furnish various examples of the different points that we make. Finally, we insist upon the fact that assessment of drug risk/benefit ratio must be global and comparative.