Metaanalyses in evidence based medicine

Metaanalyses are considered as the highest level of evidence by some leading EBM-Committees. This position might have adverse consequences, as it became recently obvious in discussions of the efficacy of antidepressants. Several arguments in favor of a more cautions interpretation of metaanalyses will be given:

Metaanalyses consider each study equivalent and ignore the individual conditions, strengths and insufficiencies, of included studies; variation across studies is informative and not just random noise; metaanalyses may produce inconsistent conclusions even if an identical body of evidence is available.

We propose that metaanalyses are useful for identification of determinants of response to treatments, but not as a maximal level of evidence.