[Clinical trials in psychiatry. Recent evolution of the methodology in France (author's transl)].

The comparison, with the help of a check list of criteria, of psychotropic clinical trials published in France, with an interval of five years, covered 80 trials in 1970 and 75 trials in 1975. The differences found between these two years all indicate a stricter methodological approach in 1975. Certain gaps persist however, such as the extreme rarity of codified pretherapeutic evaluation, preliminary wash out periods, references to an official nosography and a sufficient homogeneity of treatment groups. On the other hand, in the majority of trials, the exclusion criteria are not given and a global clinical assessment is the only criterium of change used. Finally the conclusions proposed coincide with the data in only two thirds of the publications. The comparative revue of 120 controlled psychotropic trials so far published in French show above all the extreme variability of the type of control utilized. Only 58 trials out of 120 have been randomized and a minority of these were carried out according to a strict methodology, with quantitative evaluation of the results and an appropriate statistical analysis. The most frequent methodological insufficiences are the obvious heterogeneity of the treatment groups, the insufficient use of standardized evaluation instruments and the inadapted ness of certain experimental protocols to the goals being pursued. The points most frequently omitted in the presentation of the results are the drop-outs of treatment, information relating to associated and previous treatment, the dates of evaluation of the therapeutic effects as well as the approach used to established side effects. The recent application of the 1975 French Law has deeply changed the habits of French experimentators and it is necessary to correctly interpret the term of "control". The control of non-specific variables intervening in the therapeutical results is the only means of identifying the pharmacological effects of drugs. The best control of these variables is the randomized comparative trial of simultaneous treatment groups using double-blind prescriptions. In order to further comparisons of trials from different sources, it is absolutely necesary that evaluation studies in French be carried out and transmitted according to internationally recognized recommendations based on general rules of protocols and on standardized evaluation instruments.