[Critical examination of scoring systems in therapeutic trials]

Scoring systems give a check-list and methodological informations which have to be found in controlled therapeutic trials reports and papers. These systems try to quantify each item to give a global score. The Chalmer's list is the most wellknown. It allows a balance in scoring taking in account the quality of the endpoints. Other lists are more simple. Many check-lists allow the scoring of the methodological design or the statistical analysis. In all systems the major methodological points are: the randomization, the description of the population, the double blind, the estimation of the sample size, the handling of withdrawal and drop out, the major endpoint, the patients follow-up, the statistical analysis and the data presentation. All these scoring systems have several limits: the quantitative evaluation of each item is subjective and the point scoring has never been validated, some scoring systems are old and don't integrate new methodological methods, the scores never included the clinical interest of the trial, some items are questionable, others are forgotten (intention to treat analysis, steering comity...). Scoring systems allow a control of the methodological quality of clinical trials but don't include the clinical or scientific interest of the study. These systems are a useful methodological tool for publication process in medical journals and for new drugs authorization. The evaluation by authors themselves of the quality of their papers using a standardized scoring system could clarify the reviewers decisions.