Multivariate two‐sample permutation tests for trials with multiple time‐to‐event outcomes

• DOI: 10.1002/pst.1938
• Publication Date: 2019-03-26T10:51:58Z
• Authors: Inger Persson; Lukas Arnroth; Måns Thulin
• Keywords: Censoring (clinical trials); Sample (material); Resampling

Clinical trials involving multiple time-to-event outcomes are increasingly common. In this paper, permutation tests for testing group differences in multivariate data proposed. Unlike other two-sample survival data, the proposed attain nominal type I error rate. A simulation study shows that outperform their competitors when degree of censored observations is sufficiently high. When censoring low, it seen naive such as Hotelling's T2 tailored to data. Computational and practical aspects discussed, use illustrated by analyses three publicly available datasets. Implementations an accompanying R package.