|
Abstract:
|
In observational studies, clustered right-censored data arise often. For example, to determine whether treatment extends the lifetime of ventilating tubes, the children with otitis media are randomly assigned into two groups. The lifetimes of tubes in both ears for a child (cluster) are correlated. For testing the equality of two survival distributions with such data, the weighted logrank test (WLR) and the weighted Kaplan-Meier test (WKM) have been extended, respectively. The weights of both tests can be selected to be more sensitive to detect a certain alternative. However, one seldom knows what the exact alternative is. Therefore, this paper develop a class of versatile test based on the simultaneous use of WLR and WKM tests for clustered survival data. Furthermore, simulation studies show that the versatile tests maintain better power across a broad range of alternatives.
|
