The Causal Effects of Prenatal Air Pollution on Pulmonary Function in Children with Asthma: An Example of Causal Inference Methods in the FACES LiTE cohort

Amy Padula (1) Kathleen Mortimer (1) Ira Tager (1)

(1) University of California, Berkeley

     Abstract Number: 131
     Last modified: November 5, 2009

Abstract
Background
Given recent evidence related to the association of prenatal air pollution and children’s pulmonary function, application of causal inference methods to this question assumes greater importance. The impact of these exposures, particularly among susceptible groups such as asthmatic children, is a highly relevant public health question. The Fresno Asthmatic Children’s Environment Study (FACES) – Lifetime Exposure (LiTE) examines the influence of prenatal exposure to a number of ambient air pollutants on the growth of lung function in childhood and early teen years in a high pollution area.

Methods
Based on maternal self-report, we geocoded all residences during pregnancy with Tele-Atlas. Pollutant concentrations were obtained from the Aerometric Information Retrieval System supported by the US Environmental Protection Agency. Monthly average pollutant concentrations were assigned from 24-hour averages obtained at a central site monitor and summaries of the entire pregnancy and each trimester were calculated. Spirometry was performed by trained staff as part of the FACES Study. We used targeted maximum likelihood estimation (TMLE), a recently developed method for causal inference, to examine the effects of prenatal air pollution exposure on repeated measures of pulmonary function as the cohort aged over a 4 year period. We calculated the difference in mean pulmonary function (i.e., FVC, FEV$_1, FEF$_(25-75), FEV$_1/FVC, FEF$_(25-75)/FVC, FEF$_(75)) had all children been exposed to “above the median concentration” versus “below the median concentration” of ozone (O$_3) and a series of traffic-related pollutants (nitrogen dioxide (NO$_2), particle matter <10 micro-grams/m$^3 (PM$_(10)), carbon monoxide (CO)). We stratified the data into three age groups (6-8, 9-10, and 11-13 years old) to examine separately the effect across ages.

Results
Our analysis included 162 children with a total of 947 observations over the years of 2000-2005. In general, higher exposure to NO$_2 during the first and second trimesters had a negative effect on most pulmonary function tests (PFTs). For example, among the 9-10 year olds, higher exposure to NO$_2 during the 2nd trimester resulted in a population-level decrease in FEF$_(25-75) of 0.148 L/sec (~7.5% decrease). Similarly, higher exposure to CO during each trimester had a negative impact on most PFTs across all age groups, with especially strong effects for the second trimester exposures. For both pollutants and most PFT parameters, the effects were greater in the 11-13 year old age groups. Results for PM$_(10) and O$_3 were less consistent.

Conclusion
We found that prenatal exposures to NO$_2 and CO, particularly during the second trimester, adversely affect most metrics of pulmonary function among asthmatic children ages 6-13 years. The use of TMLE provides for a more causal interpretation than traditional regression methods and effect sizes are not directly comparable. This recent analysis, however, adds to the evidence that maternal exposure to ambient traffic-related air pollutants can have persistent effects on lung function development in children with asthma. Funded by: American Lung Association and California Air Resources Board.