|
Abstract:
|
The stochastic ordering of recurrent event data presents many challenges for statistical analysis, as it results in difficult problems such as induced dependence censoring and sampling bias for recurrence times. This article aims to provide a unified approach for the analysis of recurrence time data. We suggest to apply the within-cluster resampling technique by properly selecting a recurrence time at random from each subject and analyzing the reduced data set with standard methods for independent survival times. By repeating the procedure multiple times and averaging the resulting estimates, we show that most existing methods for recurrence time data analysis can be approximated by the proposed WCR method. With minor modifications to the standard software, we are able to apply the WCR method to a broad range of settings, including one-sample estimation, regression, and model checking.
|
