[The evaluation of the quality of clinical trials: the rationale, usefulness and drawbacks]

The quality of a randomised controlled trial should be understood as its internal validity, that is, the extent in which the study conclusions show correctly what actually happened in the design and conduct phases. This validity may be compromised by multiple biases coming from the allocation of subjects to study groups, treatment administration, observation of outcome variables, patient follow-up and statistical analysis. The presence of bias is assessed through three kinds of measurements: individual components, checklists and numerical scales. In systematic reviews some kind of quality assessment is necessary, which evaluates the quality of individual trials by means of several methods: a) a threshold that sets the minimal level of quality required; b) the relationship between quality and variability in trial results; c) sensitivity analysis, and d) quality scores as weights of trials. However, there exists considerable discrepancy among the published scales whereas their usefulness to assess the quality of trials to be included in the systematic review is a matter of controversy. We hope that the Consolidated Standards of Reporting Trials (CONSORT) initiative to standardize the report of randomized controlled trials and other proposals will contribute to obtain a more realistic view for each trial and to improve their internal validity.