“Semiparametric Regression Analysis for a Shared-Frailty Progressive Multistate Model”
Chen Hu, PhD
Statistical Center of Radiation Therapy Oncology Group (RTOG)
Abstract: In advanced or adjuvant cancer studies, progression-related events (e.g., progression-free or recurrence-free survival) and cancer death are common endpoints that are sequentially observed.The relationship between covariate (e.g., therapeutic intervention), progression, and death is often of interest, as it may provide a key to optimal treatment decisions.The evaluation of this relationship is often complicated by the latency of disease progression leading to undetected or missing progression-related events. We consider a progressive multistate model with a frailty modeling the association between progression and death, and propose a semiparametric regression model for the joint distribution.An Expectation-Maximization (EM) approach is used to derive the maximum likelihood estimators of covariate effects on both endpoints, the probability of missing progression event, as well as the parameters involved in the association.The asymptotic properties of the estimators are studied.We illustrate the proposed method with Monte Carlo simulation and data analysis of a randomized clinical trial of adjuvant therapy for colorectal cancer.
“Estimating the Effect of a Time-Dependent Factor on Pre-Treatment Survival”
Douglas Schaubel, PhD Professor of Biostatistics Department of Biostatistics School of Public Health University of Michigan
Abstract: We propose semiparametric methods for estimating the effect of a time-dependent covariate on pre-treatment survival. The observed data consist of a longitudinal sequence of measurements and a potentially censored survival time. The factor of interest is time-dependent and affects both survival and treatment assignment. Survival in the absence of treatment is of interest and is dependently censored by the receipt of treatment. Patients may be removed from consideration for treatment, temporarily or permanently. The proposed methods combine landmark analysis, partly conditional hazard regression, and Inverse Probability of Censoring Weighting. The resulting estimators are consistent and asymptotically normal. We evaluate finite-sample properties through simulation, then use the proposed procedures to model pre-transplant mortality among End-stage Liver Disease patients. This is joint work with Qi Gong.
Elizabeth Hauser, PhD
Department of Medicine, Division of Medical Genetics
Duke University Medical Center, Durham, NC