Key points are not available for this paper at this time.
Sequential procedures for the comparison of two groups of survival times are considered. The assumption that the two survival curves are members of a Lehmann family, i.e. the proportional hazards model, leads to a sequential version of the log rank test. The alternative assumption of a logistic family leads to a sequential modified Wilcoxon test. Theoretical large-sample properties of both methods are assessed for trials of realistic dimensions by computer simulation.
Jones et al. (Mon,) studied this question.