From measures to models: an evaluation of air pollution exposure assessment for epidemiological studies of pregnant women

Abstract

Objectives: To evaluate exposure estimation methods such as spatially resolved land-use regression models and ambient monitoring data in the context of epidemiological studies of the impact of air pollution on pregnancy outcomes.

Methods: The study measured personal 48 h exposures (NO, NO(2), PM(2.5) mass and absorbance) and mobility (time activity and GPS) for 62 pregnant women during 2005-2006 in Vancouver, Canada, one to three times during pregnancy. Measurements were compared to modelled (using land-use regression and interpolation of ambient monitors) outdoor concentrations at subjects' home and work locations.

Results: Personal NO and absorbance (ABS) measurements were moderately correlated (NO: r = 0.54, ABS: r = 0.29) with monitor interpolations and explained primarily within-subject (temporal) variability. Land-use regression estimates including work location improved correlations for NO over those based on home postal code (for NO: r = 0.49 changed to NO: r = 0.55) and explained more between-subject variance (4-20%); limiting to a subset of samples (n = 61) when subjects spent >65% time at home also improved correlations (NO: r = 0.72). Limitations of the GPS equipment precluded assessment of including complete GPS-based mobility information.

Conclusions: The study found moderate agreement between short-term personal measurements and estimates of ambient air pollution at home based on interpolation of ambient monitors and land-use regression. These results support the use of land-use regression models in epidemiological studies, as the ability of such models to characterise high resolution spatial variability is "reflected" in personal exposure measurements, especially when mobility is characterised.