Agreement of central site measurements and land use regression modeled oxidative potential of PM2.5 with personal exposure

Abstract

Oxidative potential (OP) of ambient particulate matter (PM) has been suggested as a health-relevant exposure metric. In order to use OP for exposure assessment, information is needed about how well central site OP measurements and modeled average OP at the home address reflect temporal and spatial variation of personal OP. We collected 96-hour personal, home outdoor and indoor PM2.5 samples from 15 volunteers living either at traffic, urban or regional background locations in Utrecht, the Netherlands. OP was also measured at one central reference site to account for temporal variations. OP was assessed using electron spin resonance (OP(ESR)) and dithiothreitol (OP(DTT)). Spatial variation of average OP at the home address was modeled using land use regression (LUR) models. For both OP(ESR) and OP(DTT), temporal correlations of central site measurements with home outdoor measurements were high (R>0.75), and moderate to high (R=0.49-0.70) with personal measurements. The LUR model predictions for OP correlated significantly with the home outdoor concentrations for OP(DTT) and OP(ESR) (R=0.65 and 0.62, respectively). LUR model predictions were moderately correlated with personal OP(DTT) measurements (R=0.50). Adjustment for indoor sources, such as vacuum cleaning and absence of fume-hood, improved the temporal and spatial agreement with measured personal exposure for OP(ESR). OP(DTT) was not associated with any indoor sources. Our study results support the use of central site OP for exposure assessment of epidemiological studies focusing on short-term health effects.