|
Abstract:
|
Group sequential methods have been used in clinical trials extensively and some have advocated their use in preclinical studies. In this presentation, multiple group sequential methods are applied to a preclinical model, in which multiple compounds are compared to a control in repeated experiments for a quantitative response. Currently, the statistical approach to this preclinical model is to analyze the data by analysis of variance after data of each experiment are collected; the results from repeated experiments are combined by applying a decision rule without formally controlling for overall Type I error rate due to repeated testing. The group sequential methods are compared to the current approach in terms of Type I error, power, benefits and drawbacks.
|
