Utility and application of SEM in delineating cause–effect relationship between ambient pollutant and fetal birth outcome

The US Environmental Protection Agency has set air quality standards for six criteria pollutants based on threshold levels that could be hazardous to human health. Interest in the potential impact of prenatal exposure to these pollutants has increased; however, validity of modeling approaches has hindered research in this area. To address the methodological concerns of the previous studies, we introduce the utility of modeling exposure–outcome relationships in maternal and child health environmental research using the structural equation modeling framework. The dataset for this study comprised all live births in Hillsborough County, FL, covering the period from 1998 to 2007 inclusive. Vital records data were linked to the hospital discharge and vital statistic birth data for analysis. Additionally, local air pollution data were utilized. Out of 104,967 births, 8813 (rate = 8.4%) were born with low birth weight (LBW), 6757 infants were preterm birth (PTB) (rate = 6.4%), and 9575 (rate = 9.1%) were born small for gestational age (SGA). Our structural equation modeling approach indicated the following: (1) five of the six criteria pollutants had a positive contribution to the latent variable and (2) all the outcome variables (SGA, LBW, and PTB) had a positive relationship with the latent variable, thereby, suggesting an association between higher pollutant values and adverse birth outcomes. Copyright © 2014 John Wiley & Sons, Ltd.