Performance of a microenviromental model for estimating personal NO2 exposure in children

Abstract

Abstract A common problem in epidemiological studies on air pollution is exposure misclassification, because investigators often assume exposure is equivalent to outdoor concentrations at participants’ homes or at the nearest urban monitor. The aims of this study were: (1) to develop a new microenvironmental exposure model (MEEM), combining time-activity data with modelled outdoor and indoor NO 2 concentrations; (2) to evaluate MEEM against data collected with Ogawa™ personal samplers (OPS); (3) to compare its performance against datasets typically used in epidemiological studies. Schoolchildren wore a personal NO 2 sampler, kept a time-activity diary and completed a questionnaire. This information was used by MEEM to estimate individuals’ exposures. These were then compared against concentrations measured by OPS, modelled outdoor concentrations at the children’s home (HOME) and concentrations measured at the nearest urban monitoring station (NUM). The mean exposure predicted by MEEM (mean = 19.6 μg m − ³) was slightly lower than the mean exposure measured by OPS (mean = 20.4 μg m − ³). The normalised mean bias factor (0.01) and normalised mean absolute error factor (0.25) suggested good agreement. In contrast, the HOME (mean = 31.2 μg m − ³) and NUM (mean = 28.6 μg m − ³) methods overpredicted exposure and showed systematic errors. The results indicate that personal exposure can be modelled by MEEM with an acceptable level of agreement, while methods such as HOME and NUM show a poorer performance.