Estimating Time-Varying Effects with Penalized Splines for Recurrent Event Data

• Activity Number: 648
• Type: Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Leila Amorim*+ and Jianwen Cai and Donglin Zeng
• Companies: Universidade Federal da Bahia and The University of North Carolina at Chapel Hill and The University of North Carolina
• Keywords: penalized splines ; marginal rates models ; recurrent events ; time-varying effects ; survival models

Abstract

Methods have been proposed to explore the functional form of effects using splines with moderate number of knots, with parameters being estimated using penalized partial likelihoods in survival models. Research has also been conducted for development of estimation methods of means/rates of recurrent events. We propose a method that uses B-splines with moderate number of knots to estimate time-varying effects in the marginal rates model with estimation based on penalized partial likelihood. The proposed method is used to describe the functional form of the effect of vitamin A supplementation on the rate of recurrent diarrheal episodes and acute respiratory infections in small children from Brazil. Simulation studies examined the finite-sample properties of the proposed parameter estimators.