Semiparametric Models for Clustered Survival Data with Random Cluster Size

• Activity Number: 680
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Amita Manatunga and Shuling Liu*+ and Limin Peng
• Companies: Emory University and Emory University and Emory University
• Keywords: clustered survival data ; Clayton-Oakes model ; discrete survival model ; joint modeling ; menstrual cycle length ; time-to-pregnancy

Abstract

We consider semi-parametric regression analysis of the clustered survival data with random cluster size. Our method is motivated by the Mount Sinai Study of Women Office Workers in which menstrual cycle lengths are recorded until time-to-pregnancy(TTP) or the end of study for each woman. TTP is defined as the number of cycles to conceive, a discrete random variable. Therefore, it is natural to view the data as clustered menstrual cycle lengths with the random cluster size equal to TTP, which involves several challenging complications. To address these features , we consider a semi-parametric framework where repeated menstrual cycle lengths are modeled by the Clayton-Oakes model with the dimension indexed by TTP. We parameterize the covariate effects on menstrual lengths using a proportional odds model. A hazard regression model is assumed for TTP to incorporate potential risk factors. We propose an estimation procedure that can appropriately accommodate missing and censoring in menstrual cycle lengths as well as truncation and censoring in TTP. Simulation studies are conducted to illustrate the performance of the proposed method. Finally we apply our method to the MSSWOW study.