Two-part mixture models for zero-inflated longitudinal measurements with heterogeneous random effects and time to event data

Huirong Zhu, The University of Texas School of Public Health

Abstract

Longitudinal zero-inflated count data arise frequently in substance use research when assessing the effects of behavioral and pharmacological interventions or the effects of covariates and risk factors on outcomes. The first part of this work is extending zero-inflated count models to account for random effects heterogeneity by modeling their variance as a function of covariates. We showed via simulation that ignoring intervention and covariate-specific heterogeneity can produce biased estimates of covariate effects and random effect estimates. The methodological development was motivated by and applied to the Combined Pharmacotherapies and Behavioral Interventions for Alcohol Dependence (COMBINE) study. The second part is expanding the class of joint models for longitudinal and survival data to accommodate zero-inflated counts and time-to-event data using Cox proportional hazard model with piecewise constant baseline hazard. Via an extensive simulation study, we demonstrated more accurate estimates from applying the joint model versus the corresponding independence model. We applied the method to a Alpha-Tocopherol, Beta-Carotene (ATBC) Lung Cancer Prevention study.^

Subject Area

Biostatistics|Statistics|Public health

Recommended Citation

Zhu, Huirong, "Two-part mixture models for zero-inflated longitudinal measurements with heterogeneous random effects and time to event data" (2015). Texas Medical Center Dissertations (via ProQuest). AAI10027728.
http://digitalcommons.library.tmc.edu/dissertations/AAI10027728

Share

COinS