Validity of using annual mean particulate matter concentrations as measured at fixed site in assessing personal exposure: An exposure assessment study in Japan

•This was the first large-scale assessment of exposure to PM anywhere in Asia.•For PM and NO2, the annual correlations between fixed site and outdoor were good.•Correlations of fixed site with personal exposure were not estimated to be high.•Our findings are useful to estimate the effects of resulting measurement errors.

From 2003 through 2005, we compared annual mean particulate matter (PM) and nitrogen dioxide (NO2) concentrations as measured at fixed-site monitoring stations in 6 Japanese cities with those measured inside and outside subject residences and during personal monitoring. A total of 65 households participated in indoor and outdoor residential exposure monitoring. In summer and autumn, we also performed personal monitoring of one resident of each household. On each day, personal samplers were used to collect 24-h samples of PM and NO2 simultaneously from the fixed sites, indoor and outdoor, and from those undergoing personal monitoring. We found good correlations between the fixed-site and outdoor measurements for annual mean (average of 7-day × 4-season) concentrations of PM2.5, PM10-2.5, PM10 and NO2 (Spearman's rank correlation coefficients (ρ) ≥ 0.75). However, the correlations between the fixed-site and indoor measurements were moderate to low. In summer and autumn, the correlations between the fixed-site and personal mean concentrations of PM2.5 (ρ = 0.62), PM10 (ρ = 0.58), and NO2 (ρ = 0.70) were acceptable. However, because people spend most of their time indoors, these correlations for annual mean concentrations were not estimated to be high. Our results are important in allowing researchers to estimate the effects of resulting measurement errors of PM and NO2.