|
Abstract:
|
The proper understanding of the joint and inter-related long-term effects of air pollutants on pulmonary functions and respiratory symptoms (e.g. bronchitis) is of the interest in protecting children's respiratory health. To this end, we propose a flexible multi-level modeling technique that uses splines, whenever appropriate, to model bivariate mixed longitudinal outcomes. A latent variable approach is used to connect both outcomes within a subject. This technique allows for the examination of the inter-relationship between pulmonary functions and respiratory symptoms, after adjusting for subject-specific growth-curve parameters. A Gibbs sampling approach is implemented to estimate the posterior distribution of the parameters and latent variables. Ecologic inference is then conducted in a multi-level setting. We illustrate the technique via analysis of data from the Southern California.
|
