Spatial Hurdle Models for Predicting the Number of Children with Lead Poisoning
Date:2025-06-02 clicks:
Impact Factor:3.39
DOI number:10.3390/ijerph15091792
Journal:International Journal of Environmental Research and Public Health
Key Words:overdispersion; zero-inflated count data; negative binomial Hurdle model; generalized linear mixed models; random effects; spatial effects
Abstract:Objective The purpose of this study is to identify the high-risk areas of children’s lead poisoning in Syracuse, NY, USA, using spatial modeling techniques. The relationships between the number of children’s lead poisoning cases and three socio-economic and environmental factors (i.e., building year and town taxable value of houses, and soil lead concentration) were investigated. Methods Spatial generalized linear models (including Poisson, negative binomial, Poisson Hurdle, and negative binomial Hurdle models) were used to model the number of children’s lead poisoning cases using the three predictor variables at the census block level in the inner city of Syracuse. Results The building year and town taxable value were strongly and positively associated with the elevated risk for lead poisoning, while soil lead concentration showed a weak relationship with lead poisoning. The negative binomial Hurdle model with spatial random effects was the appropriate model for the disease count data across the city neighborhood. Conclusions The spatial negative binomial Hurdle model best fitted the number of children with lead poisoning and provided better predictions over other models. It could be used to deal with complex spatial data of children with lead poisoning, and may be generalized to other cities.
Co-author:Liyang Shao
First Author:Q1, Zhen Zhen
Indexed by:Journal paper
Correspondence Author:Lianjun Zhang*
Volume:15
Issue:9
Page Number:1792
ISSN No.:1660-4601
Translation or Not:no
Date of Publication:2018-01-01
Included Journals:SCI、SSCI