Extended Rank Tests for Analyzing Recurrent Event Data

Abstract

Wang and Chang studied the bias in the estimation of the marginal survival curve of recurrent event data and came up with an unbiased Kaplan–Meier (KM)-like estimator. However, there were no corresponding hypothesis tests to compare Wang and Chang’s survival estimates among different groups. In this article, we extended three commonly used rank tests to compare Wang and Chang’s KM-like survival estimates. Intra-subject correlation (ISC) issue is handled by using a robust variance estimator. We also studied the empirical power difference between our new method and Jung and Jeong’s method which was developed for clustered survival data and explored the relationship between ISC and the power of different rank tests. Supplementary materials for this article are available online.

Keywords:

• Bias
• Informative number of events
• Intra-subject correlation
• Rank tests
• Recurrent events