Abdus S. Wahed
Doctoral Student
Department of Statistics
Efficient estimation of the survival distribution and related quantities for treatment policies in two-stage randomization designs in clinical trials
Two-stage designs that are common in clinical investigations of cancer and other pathologies involve the initial randomization of patients to an induction therapy, followed by their randomization to a maintenance therapy, depending upon their initial therapeutic response and consent. The goal is to compare different combinations of primary and maintenance therapies to find the combination that is most beneficial. In practice, the analysis is usually conducted in two separate stages, which does not directly address the major objective of finding the best treatment combination. Recently, Lunceford et al., (Biometrics 2002; 58:48-57) introduced ad hoc estimators for the survival distribution and mean restricted survival time under different treatment policies. These estimators are consistent but not efficient, and do not include information from auxiliary covariates. In this paper we derive estimators that are both easy to compute and more efficient than estimators previously derived. We also show how to further improve efficiency by taking into account additional information from auxiliary variables. Large sample properties of these estimators are derived, and comparisons with other estimators are made using simulation. We apply our estimators to a data set from a leukemia clinical trial, which provided the motivation for this study.