Analysis of childhood morbidity with geoadditive probit and latent variable model: a case study for Egypt

This work applies geoadditive latent variable models to analyze the impact of risk factors and the spatial effects on the latent, unobservable variable “health status” or “frailty” of a child less than 5 years of age using the 2003 Demographic and Health survey (DHS) data from Egypt. Childhood diseases are a major cause of death of children in the developing world. In developing countries a quarter of infant and childhood mortality is related to childhood disease, particularly to diarrhea. Our case study is based on the 2003 Demographic and Health Survey for Egypt (EDHS). It provided data on the prevalence and treatment of common childhood disease such as diarrhea, cough, and fever, which are seen as symptoms or indicators of children’s health status, causing increased morbidity and mortality. These causes are often associated with a number of risk factors, including inadequate antenatal care, lack of or inadequate vaccination, and environmental factors that affected the health of the child in early years, various bio-demographic and socioeconomic variables. In this work, we investigate the impact of such factors on childhood disease with flexible geoadditive models. These models allow us to analyze usual linear effects of covariates, nonlinear effects of continuous covariates, and small-area regional effects within a unified, semi-parametric Bayesian framework for modeling and inference. As a first step, we use separate geoadditive probit models the binary target variables for diarrhea, cough, and fever using covariate information from the EDHS. Based on these results, we then apply recently developed geoadditive latent variable models where the three observable disease variables are taken as indicators for the latent individual variable “health status” or “frailty” of a child. This modeling approach allows us to study the common influence of risk factors on individual frailties of children, thereby automatically accounting for association between diseases as indicators for health status.