Home outdoor models for traffic-related air pollutants do not represent personal exposure measurements in Southern California

Abstract

Recent studies have used measurements or estimates of traffic-related air pollutants at home or school locations to link associations between exposure and health. However, little is known about the validity of these outdoor concentrations as an estimate for personal exposure to traffic. This paper compares modelled outdoor concentrations at home with personal exposure to traffic air pollution of 63 children in two areas in Los Angeles in 2003/2004. Exposure monitoring consisted of sixteen 10-day monitoring runs, with each run monitoring 4 subjects concurrently with the active personal DataRAM for particulate matter <2.5 μm (PM25), elemental carbon (EC) and organic carbon (OC). One child per run had concurrent indoor/outdoor home monitoring. Measurements at central sites (24-hr PM25, EC, OC) were taken daily and concentrations of PM25, EC, and OC from traffic sources were calculated using the CALINE4 model for individual residences. We modelled outdoor concentrations of PM25, EC and OC with multilinear regression including GIS and meteorological parameters and adjusted for auto-correlation between repeated measurements. The model fit (R 2) for home outdoor estimates was 0.94, 0.74 and 0.80 for PM 25, EC and OC, respectively. Comparisons between these outdoor estimates and the personal measurements showed a good agreement for PM 25 (R2=0.65-0.70) with a mean bias of -0.711.8|ag for the smog receptor area, and 18.916.2|ag for the traffic impacted area. However the outdoor estimates were not related to personal exposure for EC (R 2=0.01-0.29) and OC (R2=0.03- 0.14). Conclusions: Predictions of outdoor concentrations can be used as approximations of personal exposure to PM25. However, they are not appropriate for estimating personal exposure to traffic-related air pollutants including EC and OC in studies of acute exposure-response relationships. © 2009 IOP Publishing Ltd.