[Definitions and recommendations for studying the delay of antidepressant action. Part 3: methodological aspects]

To provide a journal publishing all literature relevant to the time of response to antidepressants, in order to find out any existing consensus which might bring in fine recommendations for a thorough study of the concept. Three sections corresponding to 3 articles surveyed by the Encephale deal with the pharmacological, clinical, and metho-dological aspects.

A group of experts (see conclusion) was trained on the initiative of the authors, following a think-talk at a meeting of scientific learned societies (FUAG and AFPB in particular). Dr Sophie Banzet, from the Boehringher-Ingelheim Laboratories, attended the meeting as a partner supporting this project. The article tries to give the most faithfull coverage of the main topics discussed by the group of experts, allowing each author to highlight his own contribution after the rereading of his text by the two other authors of the group. Are also included in the analysis the major studies published in the literature at the date when the present article was submitted.

There are no unequivocal criteria to assess or identify the time of response to antidepressants on a pharmacological level. In fact, the criteria depend on the symptomatic impact of the antidepressant while depression cannot be defined as a single-symptom pathology. The criteria retained for assessment depend on the chosen definition which is: either quantitative (patients profiles global method), or qualitative (sympto-matic approach). One should discriminate between the assessment of the time of response and the stage proving the efficacy of the antidepressant. The threshold proposed for efficacy is based on the values of the thresholds used in clinical tests and in literature, as well as those commonly accepted by the various drug agencies. Choosing the variable for the response to antidepressants must take into account the popu-lation of patients responding to treatment. A consensus on the choice of comparators can be observed in studies on the time of response to antidepressants: using a placebo is necessary. The active reference product depends on the population concerned. The statistical methods which have been used, in particular the survival analysis methods are very useful indeed but they entails loss of information and must be completed by tests which, though simpler, allow simpler conclusions (such as c(2)). Lastly, the experimental level, which aims to assess the time of response, remains by far the most difficult part. It is still unsolved and the study aiming to prove efficacy. We have not included the considerations on this particular point as this would require an article in its own right. The cognitive symptoms belonging to the DSM IV diagnosis criteria in major depressive periods can help to build new tools.