Modeling of residential outdoor exposure to traffic air pollution and assessment of associated health effects

Traffic air pollution is known to affect cardiopulmonary health in the population. Children with asthma are amongst the most susceptible groups. Several epidemiological studies linked traffic air pollution with increased reporting of asthmatic symptoms and decreased lung function. New approaches with pulmonary inflammation biomarkers allow assessment of acute effects induced by air pollution.
Populations are usually exposed to a mixture of pollutants emitted by various sources. Also, epidemiological studies using central site measurements are not able to capture different spatiotemporal distributions of the pollutants. Therefore different modeling approaches are in use to refine the spatiotemporal and the source component in exposure assessment.
The aim of this thesis was to build models for estimating short-term residential outdoor exposure to traffic-related air pollution, to find and apportion source contributions to particulate matter smaller 10µm (PM10) and to examine the relationship between spatially refined exposure estimates and respiratory health effects in children with asthma.
Methods
This thesis was conducted within the framework of two pediatric asthma panel studies: a Southern California study in the greater Los Angeles area, and the MfM-U (Monitoring flankierende Massnahmen – Umwelt) study in a Swiss Alpine valley.
In the Southern California study measurements of personal particulate matter smaller 2.5µm (PM2.5), elemental carbon (EC), and organic carbon (OC) were collected in 63 children living in Riverside (Aug to Dec 2003) and Whittier (July to Nov 2004). Concurrently one home site and a fixed central site were monitored. Home site measurements were used to build city-specific and pooled models for estimating PM2.5, EC, and OC levels at all other participating children’s homes by using land-use regression methods including fixed site measurements and CALINE4 dispersion estimates (local traffic). We compared the home outdoor estimates with the personal measurements.
The MfM-U panel study was conducted in Erstfeld located in a highway impacted Swiss Alpine valley. From November 2007 to June 2009, thirteen children with asthma had monthly monitoring of pulmonary inflammation (i.e. fractional exhaled nitric oxide (FeNO)) and oxidative stress markers in exhaled breath condensate (eBC) (i.e. nitrite, pH). Concurrently levels of PM10, nitrogen dioxide (NO2), EC, OC, and particle numbers (PN) were monitored at one background, one highway and seven mobile sites. NO2 measurements were used to build a model estimating outdoor concentrations at the participating children’s homes with a similar approach as in the Southern California study. Chemically speciated data was used in receptor modeling to apportion the source contributions to PM10. NO2 model estimates and source-specific PM10 were then used to investigate associations to pulmonary inflammation and oxidative stress marker levels in the children.
Results
In the Southern California study, all models could explain a large part of variation for home outdoor PM2.5, OC and EC (adj R2 = 0.75 to 0.97). Important predictors were central site measurement, distance to highway and wind variables. However, only PM2.5 model estimates correlated well with daily personal measurements (R2 = 0.65 to 0.69).
In the MfM-U study, traffic-related pollutants NO2, EC and PN showed high concentrations at the highway site decaying some 30-40% to background levels within 150-200m. Weekday patterns of traffic pollutants followed the heavy-duty truck traffic counts on the highway. All pollutants showed higher levels in winter than in summer. The NO2 model explained a large part of variance (adj R2 = 0.91) and estimates matched very well the validation measurements (R2 = 0.74).
We identified nine sources contributing to PM10. Traffic (29%) was the main source, including traffic exhaust (18%), road dust (8%), tire & brake wear (1%), and road salt (2%). Other contributions came from secondary particles (27%), biomass burning (18%), railway traffic (11%) and mineral sources from mineral dust (7%) and a tunnel construction site (6%). There were higher contributions from secondary particles (37%) in summer and from biomass burning (26%) and traffic (30%) in winter. Traffic, railway and mineral contributions to PM10 were higher at sites close to the specific source. Biomass burning estimates correlated well (R2 = 0.81) with levoglucosan (wood burning marker), while traffic exhaust estimates were weakly associated (R2=0.13) with 1-nitropyrine (diesel exhaust marker) due to the mixture of diesel and gasoline in the traffic fleet.
Mean levels of FeNO, eBC nitrite, and eBC pH measured in the thirteen children were 17.04ppb, 0.82µM, and 7.06, respectively, indicative for mild asthma. For days without report of any cold symptoms, FeNO levels increased by 15%, 13% and 6% if NO2, EC and total PM10 on the prior day of the health measurement were increased by one inter quartile range, respectively. eBC pH levels decreased significantly with increasing PM10, NO2, and EC concentrations measured one, two or three days prior the health monitoring. However, no significant associations were observed between source-specific PM10 concentrations and FeNO, and between eBC nitrite and any of the pollutants.
Conclusions
We were able to build models to estimate residential outdoor air pollution exposure using only a limited number of spatially distributed monitoring sites.
We could identify traffic as the major source contributing to PM10 in Erstfeld and observed a distinct relationship between highway traffic and concentration levels of NO2, EC and PN. Despite relatively low air pollution levels in Switzerland, we still detected associations between traffic-related air pollution and pulmonary inflammation markers in children with asthma.