Rank tests for matched survival data.


Journal Article

In a clinical trial with the time to an event as the outcome of interest, we may randomize a number of matched subjects, such as litters, to different treatments. The number of treatments equals the number of subjects per litter, two in the case of twins. In this case, the survival times of matched subjects could be dependent. Although the standard rank tests, such as the logrank and Wilcoxon tests, for independent samples may be used to test the equality of marginal survival distributions, their standard error should be modified to accommodate the possible dependence of survival times between matched subjects. In this paper we propose a method of calculating the standard error of the rank tests for paired two sample survival data. the method is naturally extended to that for K-sample tests under dependence.

Full Text

Duke Authors

Cited Authors

  • Jung, SH

Published Date

  • 1999

Published In

Volume / Issue

  • 5 / 1

Start / End Page

  • 67 - 79

PubMed ID

  • 10214003

Pubmed Central ID

  • 10214003

International Standard Serial Number (ISSN)

  • 1380-7870

Digital Object Identifier (DOI)

  • 10.1023/a:1009635201363


  • eng

Conference Location

  • United States