Characterization of source-specific air pollution exposure for a large population-based Swiss cohort (SAPALDIA)

Abstract

Background: Although the dispersion model approach has been used in some epidemiologic studies to examine health effects of traffic-specific air pollution, no study has evaluated the model predictions vigorously.

Methods: We evaluated total and traffic-specific particulate matter < 10 and < 2.5 microm in aero-dynamic diameter (PM(10), PM(2.5)), nitrogren dioxide, and nitrogen oxide concentrations predicted by Gaussian dispersion models against fixed-site measurements at different locations, including traffic-impacted, urban-background, and alpine settings between and across cities. The model predictions were then used to estimate individual subjects' historical and cumulative exposures with a temporal trend model.

Results: Modeled PM(10) and NO(2) predicted at least 55% and 72% of the variability of the measured PM(10) and NO(2), respectively. Traffic-specific pollution estimates correlated with the NO(x) measurements (R(2) >or=0.77) for background sites but not for traffic sites. Regional background PM(10) accounted for most PM(10) mass in all cities. Whereas traffic PM(10) accounted for < 20% of the total PM(10), it varied significantly within cities. The modeling error for PM(10) was similar within and between cities. Traffic NO(x) accounted for the majority of NO(x) mass in urban areas, whereas background NO(x) accounted for the majority of NO(x) in rural areas. The within-city NO(2) modeling error was larger than that between cities.

Conclusions: The dispersion model predicted well the total PM(10), NO(x), and NO(2) and traffic-specific pollution at background sites. However, the model underpredicted traffic NO(x) and NO(2) at traffic sites and needs refinement to reflect local conditions. The dispersion model predictions for PM(10) are suitable for examining individual exposures and health effects within and between cities.

Keywords: cohort study; cumulative exposure; dispersion model; exposure assessment; long-term exposure.

Figure 1

Measured versus modeled annual PM₁₀ concentrations for all Swiss sites including those in the SAPALDIA areas (in white) in (A) 2000 [SAPALDIA areas: y = 1.9 + 0.87x (R² = 0.68, n = 15); all sites: y = 5.2 + 0.72x (R² = 0.55, n = 57] and (B) 1990 [SAPALDIA areas: y = 8.6 + 0.6x (R² = 0.45, n = 8); all sites: y = 8.6 + 0.61x (R² = 0.63, n = 25)]. Traffic sites are within 20 m of a major road.

Figure 2

Measured vs. modeled annual NO_x concentrations for all Swiss sites including those in the SAPALDIA areas (in white) in (A) 2000 [SAPALDIA: y = 17.1 + 0.68x (R² = 0.43, n = 10); all sites: y = 18.6 + 0.67x (R² = 0.48, n = 17)] and (B) 1990 [SAPALDIA areas: y = −39 + 2.4x (R² = 0.79, n = 6); all sites: y = −0.1 + 1.2x (R² = 0.66, n = 11)]. Traffic sites are shown as triangles.

Figure 3

Measured vs. modeled annual NO₂ concentrations for all Swiss sites including those in the SAPALDIA areas (in white) in (A) 2000 [SAPALDIA areas: y = 3.3 + 0.81x (R² = 0.79, n = 24); all sites: y = 5.8 + 0.73x (R² = 0.72, n = 103)] and (B) 1990 [SAPALDIA areas: y = 1.8 + 0.78x (R² = 0.80, n = 12); all sites: y = 6.0 + 0.69x (R² = 0.75, n = 38)]. Two blue triangles are airport sites.

Figure 4

Relationship in 2000 between measured NO_x and modeled traffic-related (A) NO_x [background sites: y = −14 + 0.98x (R² = 0.77, n = 11); traffic sites: y = 41 + 0.13x (R² = 0.01, n = 6); all sites: y = 2 + 0.50x (R² = 0.41, n = 17)]; (B) PM₁₀ [background: y = −2 + 0.16x (R ² = 0.84, n = 11); traffic sites: y = 0.4 + 0.06x (R² = 0.35, n = 6); all sites: y = 1 + 0.06x (R² = 0.54, n = 17)]; (C) NO₂ [background: y = −7 + 0.59x (R ² = 0.85, n = 11); traffic sites: y = 20 + 0.08x (R² = 0.04, n = 6); all sites: y = 4 + 0.24x (R² = 0.50, n = 17)]; and (D) PM_2.5 [background: y = −0.7 + 0.07x (R² = 0.86, n = 11); traffic sites: y = 0.4 + 0.03x (R² = 0.34, n = 6); all sites: y = 0.6 + 0.03x (R² = 0.53, n = 17)]. SAPALDIA sites are in white.

Figure 5

Dispersion modeling predictions for PM₁₀ concentrations outside residences of the SAPALDIA 2 cohort.

Figure 6

(A) Two principal components of sources which are presented by their loadings on these two components. (B) Source characteristics of the SAPALDIA areas given by the two component scores. “Other” includes subjects who have moved out of the SAPALDIA areas.