[The ABC's of medical statistics. Reading and understanding clinical trials].

Clinical trials test hypotheses that are accepted or rejected according to a predetermined probability of error (level of significance). Significance does not however mean relevance. Good parameters of relevance are absolute risk reduction and based on this the calculation of the number of patients who need to be treated for one additional patient to benefit. The randomized controlled trial is the gold standard for comparative evaluation of effects. In the ideal scenario it is designed so that a difference established by statistical methods becomes probable. In non-inferiority studies care should be taken that no equivalence is shown but rather that the difference is not greater than a predefined margin of error for differences. Meta-analyses of studies with similar endpoints have the potential to improve the level of evidence. Since the findings of meta-analyses depend on the studies included, critical assessment of the results is essential.