|
Abstract:
|
In cardiovascular (CV) trials, the endpoint of interest is often a composite of two or more types of related clinical outcomes. Traditional analyses on time to the first event will ignore more important subsequent event such as CV death. To account for the order of clinical priorities, Pocock et al. (2012, European Heart Journal) proposed a win ratio approach to analyze composite endpoints. Luo et al. (2015, Biometrics) used U-statistics technique for hypothesis testing on win ratio in the "unmatched" situation. An explicit variance estimator for the win ratio statistic was derived with computational efficiency and a relatively simple form. Bebu and Lachin (2015, Biostatistics) generated the win ratio approach to more general settings. However, the current win ratio approach can be improved via weighting. Currently, the win ratio for the terminal event part is Gehan statistic which may be replaced by log-rank statistic. For the non-terminal event, a similar but a bit more complex weighting strategy can be implemented. We will examine the effects of different weighting mechanisms on the win ratio statistic through simulation studies and theoretical justification.
|
