General Departmental Seminar Series
Estimation of the Effects on Survival of Treatment Sequences for Recurrent Diseases
Xuelin Huang, PhD,
Department of Biostatistics and Applied Mathematics,
MD Anderson Cancer Center
Friday, September 10, 2004, 12-1 p.m.
G5/113 Clinical Sciences Center, 600 Highland Ave.
Many diseases may recur after initial treatments. Patients commonly receive a variety of sequential treatments. The initial treatments administered following diagnosis can vary as well as subsequent salvage regimens given after disease recurrences. The number of recurrences, and thus the number of treatments received, can vary from patient to patient. The treatment sequences are dynamic in the sense that they depend on the severity and other prognostic information of the disease. The goal is to find the optimal treatment strategy that results in the longest patient survival. In this talk, I will present a method to evaluate different treatment strategies, and show that why the Cox proportional hazards model with time-dependent covariates is not appropriate for this task. To accomplish this goal, we use estimating equations to estimate the marginal mean survival time, the survival distribution function, and the covariate-adjusted mean survival time. This is done for each individual treatment strategy under consideration. Different treatment strategies are then compared using the estimated marginal and covariate-adjusted mean survival times. The estimating equations are weighted by the inverse of the probability of "compliance" with a treatment strategy. The covariance matrices of the estimated parameters are obtained by the sandwich formula. Simulation studies are conducted to evaluate the method. A retrospective study of patients with soft tissue sarcoma is presented for motivation and illustration.
Return to seminar list