[Restrictions and methodological considerations of multicenter trials (author's transl)].

Multicenter or cooperative trials should be initially evaluated for efficacy using criteria that in some way measure the degree to which the various centers are coordinated. The fundamental limitations of these studies can be traced to the inadequacy of coordination among the centers taking part in the trial. Whereas the structural homogeneity of the patient groups can be achieved by stratification and randomization within the centers, the organisational effort to prevent observer variation between centers is often underestimated. In this case the selection of the centers has a great bearing. More specifically their representativeness plays an important part in the interpretation of the apparent differences between the centers.