Analysis of randomized dose-finding-studies: Closure test modifications based on multiple contrast tests

The power of single contrast tests strongly depends on the a priori unknown shape of the dose response relationship. Two approaches can be used to overcome this major disadvantage: the likelihood ratio test for order restriction or multiple contrast tests. Within the closure principle, the shape for the different hypotheses varies. Therefore multiple contrast tests are used as trend tests. Based on a simulation study the power advantage of this approach in comparison with single contrast tests is demonstrated.