Restricted mean survival time and confidence intervals by empirical likelihood ratio
Mai Zhou
Abstract
Recently, an increasing number of researchers propose to use restricted mean survival time (RMST) to evaluate medical treatment outcomes when the proportional hazards assumption is in doubt. Using the empirical likelihood ratio for the Kaplan-Meier mean, we show how the statistical analysis for the RMST and difference/ratio of two RMSTs can be obtained. Examples are given using R software and package emplik. We argue that the confidence intervals produced with this new method has several intrinsic advantages over the current method implemented in R package survRM2. A simulation is also included.
Topics & Concepts
Confidence intervalStatisticsMathematicsR packageEmpirical likelihoodProportional hazards modelEconometricsComputer scienceStatistical Methods and InferenceAdvanced Causal Inference TechniquesStatistical Methods in Clinical Trials