[Interim analyses--a scientific dilemma].

Interim analyses are nowadays an important part of clinical trials. Unfortunately, there is no consensus about when and how to conduct these analyses. In the field of statistics different schools promote different statistical methods. Since the choice of method can have a major impact on the results it is essential to be well informed about the problems, associated with interim analyses. One such problem concerns communication between statisticians and physicians. Close collaboration is required to facilitate good clinical trials. This paper highlights some of the statistical difficulties and relates these to the problem of collaboration between statisticians and physicians.