Accounting for gestational age uncertainty in estimating associations between preterm birth and ambient air pollution

Abstract

PDS 69: Methods and statistics, Johan Friso Foyer, Floor 1, August 26, 2019, 4:30 PM - 5:30 PM BACKGROUND: Previous epidemiologic studies utilizing birth records have shown heterogeneous associations between air pollution exposure during pregnancy and the risk of preterm birth (PTB, gestational age <37 weeks). Uncertainty in gestational age at birth may contribute to this heterogeneity. METHODS: We first examined disagreement between clinical and last menstrual period-based (LMP) determination of PTB from individual-level birth certificate data for the 20-county Atlanta metropolitan area during 2002 to 2006. We then estimated associations between five trimester-averaged pollutant exposures and PTB, defined using various methods based on the clinical or LMP gestational age. We then used a multiple imputation approach to incorporate uncertainty in gestational age to quantify the impact of this variability on associations between pollutant exposures and PTB. Finally, a Bayesian hierarchical model was developed under the framework of outcome-misclassification without a gold standard, where both the LMP and the clinical estimates are treated as error-prone observations. Results: Odds ratios (OR) were most elevated when a more stringent definition of PTB was used. For example, defining PTB only when LMP and clinical diagnoses agree yielded an OR of 1.09 (95% confidence interval [CI] = 1.04, 1.14) per interquartile range increase in first trimester carbon monoxide exposure versus an OR of 1.04 95% CI = 1.01, 1.08) when PTB was defined as either an LMP or clinical diagnosis. Accounting for outcome uncertainty resulted in wider CIs: between 7.4% and 43.8% wider than those assuming the PTB outcome is without error. CONCLUSION: Despite discrepancies in PTB derived using either the clinical or LMP gestational age estimates, our analyses demonstrated robust positive associations between PTB and ambient air pollution exposures even when gestational age uncertainty is present.