Latent Class Analysis of the Immune Phenotypes in ChildrenView Presentation *Leila D Amorim, UFBA/UNC
Maurício L Barreto, UFBA
Camila A Figueiredo, UFBA
Keywords: latent variable modeling, categorical data, cytokines
A growing number of childhood diseases have been linked to environmental exposures. Cytokines denote a large class of molecules involved in the process of triggering immune responses. Description of immune profiles, and the influence of environment and infections on them may have important implications to understand the development of inflammatory diseases. Latent class analysis (LCA) is a statistical method used to identify distinct subsets (classes) underlying the observed heterogeneity in a population. We used LCA to determine whether an unobserved categorical variable adequately explains its observed features. Data on cytokines of 1127 children from a study conducted in Brazil is used to determine and describe the immune phenotypes, which were associated to environmental exposures, infections, allergy and wheezing. The results are consistent with a three-class unobserved factor explaining associations among the immunological observed components. This is to our knowledge the first attempt to apply LCA to better understand the immune phenotypes and seems promising in furthering description of their relationships with environmental exposures and infections.