Latent variable modelling of risk factors associated with childhood diseases: Case study for Nigeria

Objective: To investigate the impact of various bio-demographic and socio-economic variables on joint childhood diseases in Nigeria with flexible geoadditive probit models. Methods: Geoadditive latent variable model (LVM) was applied where the three observable disease (diarrhea, cough, fever) variables were modelled as indicators for the latent individual variable "health status" or "frailty" of a child. This modelling approach allowed us to investigate the common influence of risk factors on individual frailties of children, thereby automatically accounting for association between diseases as indicators for health status. The LVM extended to analyze the impact of risk factors and the spatial effects on the unobservable variable "health status" of a child less than 5 years of age using the 2003 Demographic and Health Surveys (DHS) data for Nigeria. Results: The results suggest some strong underlying spatial patterns of the three ailments with a clear southeastern divide of childhood morbidities and this might be the results in the overlapping of the various risk factors. Conclusions: Comorbidity with conditions such as cough, diarrhoea and fever is common in Nigeria. However, little is known about common risk factors and geographical overlaps in these illnesses. The search for overlapping common risk factors and their spatial effects may improve our understanding of the etiology of diseases for efficient and cost-effective control and planning of the three ailments. © 2011 Asian Pacific Tropical Medicine Press.