Risk Prediction in Consecutive Time-to-Event Outcome Subject to a Competing Event

• Activity Number: 240
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Joanna Shih*+ and Paul Albert
• Companies: National Cancer Institute and NICHD
• Keywords: risk prediction ; adverse pregnancy outcome ; normal copula ; repeated pregnancies ; polychotomous random effects model

Abstract

Adverse pregnancy outcomes, such as preeclampsia-related pre-term birth, may recur, and its recurrence may depend on previous history of adverse pregnancy outcomes and gestation age. It is clinically important to quantify the risk of recurrence of the adverse pregnancy outcome of interest and to identify the associated risk factors. To this end, a retrospective observational study in an obstetric population for women with more than one pregnancy was conducted. The focus of this article is on risk prediction subject to a competing event. To acheive this, we form a joint model,where adverse pregnancy outcomes are modeled with a polychotomous random effects model,and gestation ages are modeled conditional on the adverse pregnancy outcomes. The correlation between repeated adverse pregnancy outcomes is induced by random effects, and the correlation between gestation ages conditional on the adverse outcomes is modeled by the semi-parametric normal copula function. We present an estimation method and develop the asymptotic theory for the proposed estimators. The proposed model and estimation procedure are applied to the NICHD Consecutive Pregnancy Study data and evaluated by simulations.