A spatial-temporal regression model to predict daily outdoor residential PAH concentrations in an epidemiologic study in Fresno, CA
Highlights
► Daily ambient PAHs were measured in Fresno, CA both centrally and residentially. ► The data were modeled in a land use regression model using mixed effects. ► The model accounted for 80% of between-home variability in residential PAHs. ► The model accounted for 18% of within-home variability in residential PAHs. ► Daily outdoor residential exposure to PAHs is variable within Fresno, CA.
Introduction
The majority of health effect research focused on daily exposure to ambient pollutants has used a single central monitor to assign exposure to participants (Peel et al., 2005, Penttinen et al., 2001), relying on the assumption that temporal variability overshadows spatial variability such that exposure misclassification is negligible. When incorrect, this assumption could lead to exposure misclassification of spatially heterogeneous pollutants and result in significant differences in the correlation to the health outcome (Sarnat et al., 2010, Wilson et al., 2005, Wilson et al., 2007). The degree of heterogeneity of a pollutant’s spatial distribution can be tested by examining the absolute concentrations, correlation coefficients, and the coefficient of divergence between measured data at different sites (Wilson et al., 2005). Using these techniques, researchers have found many air toxics, including PAHs, with enough heterogeneity that using a single monitor would cause some degree of exposure misclassification (Lehndorff and Schwark, 2004, Levy et al., 2001). If the distribution of PAHs is heterogeneous within the study area and some spatially resolved measurement data are available, modeling the spatial distribution may be preferable to using a central monitor value directly (Ito et al., 2004). However, in a moderate to large urban area for a cohort epidemiology study collecting personal PAH samples or even a large number of cross-sectional samples is not feasible. The intent of this paper is to model the daily individual exposures to outdoor residential PAHs, over 8 years of follow-up, through land use regression (LUR) modeling.
PAHs are a class of compounds characterized by fused aromatic rings that form when organic matter undergoes incomplete combustion. PAHs generally exist in complex mixtures of combustion products such as diesel exhaust, soot, and wood and tobacco smoke. They exist in ambient air as gases (gas-phase) and adsorbed to particulate matter (particle-phase or particle-bound). PAHs are produced by both diesel and gasoline fuel combustion (Cadle et al., 1999, Marr et al., 1999, Riddle et al., 2007, Rogge et al., 1993), as well as biomass burning (Jenkins et al., 1996, Schauer and Cass, 2000). However, PAHs are not merely another proxy for traffic exhaust, they are well-known as carcinogens (International Agency for Research on Cancer, 1989) and toxic air contaminants (Office of Environmental Health Hazard Assessment, 2001). PAHs have most recently been implicated in short-term respiratory health outcomes (Delfino, 2002, Diaz-Sanchez et al., 1994) and immunological functioning related to mechanisms involved with asthma and atopy (Finkelman et al., 2004, Nadeau et al., 2010, Perera et al., 2009). Despite this increasing interest in health effects related to exposure to ambient concentrations of PAHs, to our knowledge, no data exist on daily intra-urban spatial distributions or individual exposure estimates in the context of an epidemiologic cohort study.
LUR models are spatial regression models that relate location-specific data on pollutant concentrations to location-specific source and environment data using regression (Briggs et al., 1997). Unlike interpolation methods, such as kriging, LUR models are able to exploit measurement data to build a smooth pollutant surface even when there are significant local sources and intra-urban variability (Jerrett et al., 2005). The majority of LUR models and spatial models for air pollution exposures related to health effects has focused on modeling annual average exposure to NO2, CO, or particulate matter (Hoek et al., 2008). While PAHs share some emission sources with these pollutants, the spatial distributions of these three pollutants are not identical (Fischer et al., 2000, Levy et al., 2001, Sarnat et al., 2010).
Section snippets
Study background and population
The combination of Fresno’s geographic location and meteorology contributes to very poor air quality in both the summer and the winter (Blanchard et al., 1999). Because of the Southern Sierra Nevada in the West, the Tehachapi Mountains on the south, and the Coastal Range Mountains in the East, the San Joaquin air basin (in which Fresno is located) does not have an outlet for air pollution. Additionally, during the winter months, inversion layers from lowered mixing heights cause stagnation in
Study background and population
During the time from 1/1/2001 to 9/30/2008, there were 315 FACES participants who lived at a total of 465 residences, a consequence of the fact that some participants moved during the study (Fig. 1). These residences were geocoded using the TAMN roadway database. More than 97% of the residences matched an exact street address in the database, and the remaining were geocoded with online mapping services or GPS coordinates from home visits.
Field sampling
The 24-h average concentrations of particle-bound PAHs at
Discussion
The goal of this research was to specify and implement a model for daily, outdoor, residential PAH concentrations for each FACES participant for use in further epidemiological investigations of acute and longitudinal effects of air pollution on asthmatic children (Mann et al., 2010, Margolis et al., 2009, Nadeau et al., 2010). This goal was achieved with a mixed-effects model based on measurement data combined with longitudinal data collected over more than seven years. The single most
Conclusion
In conclusion, we have estimated individual daily outdoor PAH exposure for the 315 participants in the FACES study for over seven years using LUR modeling with mixed-effects regression. We found that traffic characteristics, home heating, season, and meteorology each play an important role in characterizing PAH exposure in Fresno, CA. While temporal variables accounted for more of the total variability within the model, the estimates were significantly improved by the addition of spatial
Acknowledgments
We wish to thank Fred Lurmann, Paul Roberts, Charles Perrino, Masahiko Sugihara, Li Ding, and Wei Hu. We also wish to thank the California Air Resources Board (Contract Nos. 99-322, 99-323 and -01-346). The statements and conclusions in this article are those of the author and not necessarily those of the California Air Resources Board. The mention of commercial products, their source or their use in connection with material reported herein is not to be construed as actual or implied
References (43)
- et al.
Single-station integral measures of atmospheric stagnation, recirculation and ventilation
Atmospheric Environment
(1994) - et al.
Predicting residential indoor concentrations of nitrogen dioxide, fine particulate matter, and elemental carbon using questionnaire and geographic information system based data
Atmospheric Environment
(2007) - et al.
Spatial representativeness and scales of transport during the 1995 Integrated Monitoring Study in California’s San Joaquin Valley
Atmospheric Environment
(1999) - et al.
Traffic-related differences in outdoor and indoor concentrations of particles and volatile organic compounds in Amsterdam
Atmospheric Environment
(2000) - et al.
Particle-associated polycyclic aromatic hydrocarbons in urban air of Hong Kong
Atmospheric Environment
(2003) - et al.
A review of land-use regression models to assess spatial variation of outdoor air pollution
Atmospheric Environment
(2008) - et al.
Spatial variation of PM2.5 chemical species and source-apportioned mass concentrations in New York City
Atmospheric Environment
(2004) - et al.
Particle concentrations, gas-particle partitioning, and species intercorrelations for polycyclic aromatic hydrocarbons (PAH) emitted during biomass burning
Atmospheric Environment
(1996) - et al.
PCDD/F- and PAH-emission from house heating systems
Chemosphere
(1998) - et al.
Biomonitoring of air quality in the Cologne conurbation using pine needles as a passive sampler – Part II: polycyclic aromatic hydrocarbons (PAH)
Atmospheric Environment
(2004)