[Group sequential experiments. Short introduction and review].

Group sequential experiments are terminated when an interim statistical test turns out to be "significant", or when maximum study size is attained. To ensure a stipulated total level of significance the interim tests have to be performed at a stricter level of significance determined by the maximum number of interim tests. As for fixed size studies, it is possible to determine a study size leading to a stipulated power. This study size depends on expected effect of intervention, number of interim tests and level of significance. Tables and computer simulation programs have been developed to facilitate the design of such plans. If the intervention turns out to be about as effective as expected, a saving in study size of about 30% can be expected in the long run, relative to studies of fixed size. If the intervention is much more effective than expected, larger savings can be achieved.