A group sequential procedure for clinical trials in multiple sclerosis with count data

摘要

Group sequential designs are widely used in clinical trials to sequentially monitor safety and efficacy data of an experimental therapy in comparison to the control. This paper develops a group sequential procedure for count data with or without overdispersions, which are commonly used as the primary endpoint in multiple sclerosis trials. The usual stopping boundaries are no longer valid since the independent increment structure for the sequentially calculated test statistic does not hold. We investigate a Jackknife method to estimate the true correlation of the sequential test statistic and derive the stopping boundaries stage by stage from the data. Examples are provided to illustrate the method and the impact of dependent increments. Count data, Group sequential, Independent increment, Interim analysis, Jackknife.