Air Pollution and Lung Function in Dutch Children: A Comparison of Exposure Estimates and Associations Based on Land Use Regression and Dispersion Exposure Modeling Approaches

Abstract

Background: There is limited knowledge about the extent to which estimates of air pollution effects on health are affected by the choice for a specific exposure model.

Objectives: We aimed to evaluate the correlation between long-term air pollution exposure estimates using two commonly used exposure modeling techniques [dispersion and land use regression (LUR) models] and, in addition, to compare the estimates of the association between long-term exposure to air pollution and lung function in children using these exposure modeling techniques.

Methods: We used data of 1,058 participants of a Dutch birth cohort study with measured forced expiratory volume in 1 sec (FEV1), forced vital capacity (FVC), and peak expiratory flow (PEF) measurements at 8 years of age. For each child, annual average outdoor air pollution exposure [nitrogen dioxide (NO2), mass concentration of particulate matter with diameters ≤ 2.5 and ≤ 10 μm (PM2.5, PM10), and PM2.5 soot] was estimated for the current addresses of the participants by a dispersion and a LUR model. Associations between exposures to air pollution and lung function parameters were estimated using linear regression analysis with confounder adjustment.

Results: Correlations between LUR- and dispersion-modeled pollution concentrations were high for NO2, PM2.5, and PM2.5 soot (R = 0.86-0.90) but low for PM10 (R = 0.57). Associations with lung function were similar for air pollutant exposures estimated using LUR and dispersion modeling, except for associations of PM2.5 with FEV1 and FVC, which were stronger but less precise for exposures based on LUR compared with dispersion model.

Conclusions: Predictions from LUR and dispersion models correlated very well for PM2.5, NO2, and PM2.5 soot but not for PM10. Health effect estimates did not depend on the type of model used to estimate exposure in a population of Dutch children.

Figure 1

Pearson correlation coefficients of dispersion-modeled NO₂ (n = 80), PM_2.5, PM_2.5 soot, PM₁₀ (n = 40) with the same pollutants measured at the ESCAPE sites.

Figure 2

Pearson correlation coefficients of air pollution estimates between localized dispersion and LUR models at the PIAMA addresses (n = 1,058).

Figure 3

Adjusted associations of annual levels of air pollutants estimated by dispersion and LUR modeling approaches with FEV₁, FVC, and PEF level (n = 1,058) at the PIAMA current addresses. The increment of each pollutant is calculated by 10 μg/m³ for NO₂ and PM₁₀, 1 × 10^–5/m for PM_2.5 soot, and 5 μg/m³ for PM_2.5.