[Appraising the evidence for new treatments].

For new interventions, the results of sequential randomized or non-randomized trials and meta-analysis can differ significantly. Evaluation of the evidence for the effect of a new treatment is a complex interplay of several factors, including the methodological design, the risk of a coincidental finding and applicability in practice. For proper appraisal of the design of trials, the use of aggregate scores should be avoided and individual study limitations should be mentioned. With the use of additional analyses we can now test whether meta-analyses contain sufficient data to find potentially relevant differences. The new 'Grading of recommendations assessment, development and evaluation' (GRADE) system is a consensus guideline that combines multiple factors into an easily interpreted judgment of the strength of evidence. Concise presentation of important quality factors for each outcome significantly increases clarity. A more realistic assessment of the reliability of the evidence decreases the risk of major fluctuations in treatment policy.