Mixed-Effects Models for the Effects of Partially Observed Covariates on Dependent Recurrent Events Data

• Activity Number: 461
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Stephen Rathbun*+
• Companies: University of Georgia
• Keywords: point process ; ecological momentary assessment ; estimating equations ; smoking data

Abstract

In an Ecological Momentary Assessment of smoking, volunteer smokers were asked to record each cigarette smoked on a PDA over a 16-day period prior to a designated quit date. In addition, the PDA prompted smokers to answer questions regarding mood and environment at times of randomly sampled cigarettes and at times selected according to a probability-based sampling design. A mixed effects version of a point process model is constructed to describe the relationship between the repeated cigarette smoking events and the partially-observed time-varying covariates, allowing for variation among smokers in their responses to mood and environment. Estimating equations are proposed that yield consistent estimators of model parameters without making assumptions regarding the distribution of time-varying covariates. Moreover, statistical inference remains valid even when, conditional on the random effects, event times are not realized from a Poisson point process.