Abstract:
|
The restricted mean survival time (RMST) is a summary measure for possibly censored survival time distribution, which is defined as the expectation of min(T, tau), where T and tau are event time and fixed truncation time, respectively. RMST and its contrast provide easy and straightforward interpretation compared with conventional summary measures such as the t-year survival probability and the hazard ratio. For the estimation of RMST, the pseudo-value (PV) and the inverse probability of censoring weighting (IPCW) approaches can be used to account for right censoring, both of which require the independent censoring assumptions for validity. Although it would be difficult to fully satisfy in practice, evaluations of these approaches have not been conducted when the independent censoring is violated. With this background, the aim of this research is to assess the robustness of PV and IPCW against the violation of the independent censoring through simulation studies. In the simple simulation setting where both event and censoring time depended only on the assigned group, IPCW showed biased results. The PV approach did not show such bias.
|