Personal versus fixed-site monitoring for assessing PM2.5 exposure in an industrial city, Saudi Arabia

摘要

Background: Air pollution is a known risk factor for adverse health effects. Many epidemiological studies use outdoor air pollution levels based on fixed-site monitoring data as a surrogate for human exposure, however individuals spend, on average, 85% of their time indoors, where exposure sources differ from outdoors. Personal monitoring allows more appropriate exposure estimation, but, to date, little has been done to compare exposure based on fixed-site versus personal monitoring, and no such comparisons have been carried out in the Middle East. Aims: To investigate the validity of fixed-site versus personal monitoring of PM2.5 in an industrial city in Saudi Arabia. Methods: We collected 24-hour personal monitoring of PM2.5 exposure from 28 students aged 16-18 years, using a SidePak AM510 set to record PM2.5 levels every minute. Students completed a time-activity diary to identify time spent in key microenvironments, especially outdoors and indoors. Students also carried a GPS device to log their geographic position. Hourly ambient PM2.5 levels for the same 24 hour periods were extracted for the nearest fixed-site monitoring station. The GPS data and home locations are being used to plot the geographic distribution of residuals in GIS to explore the spatial variation in the fixed versus personal monitor data. Results: Hourly mean PM2.5 exposures were higher when assessed via personal monitoring (Mean=39.3 μg/m3, Median=30.5, SD=52.2, Range=816.3) than fixed-site monitoring (Mean=20.4, Median=12.2, μg/m3, SD=27.1, Range=257.6). There was a non-significant correlation between log transformed personal and fixed-site monitor PM2.5 levels when subjects were predominantly outdoors (Correlation=0.31, p=0.07, n=39hrs), and a weak, but significant correlation when subjects were indoors (Correlation=0.14, p＜0.00, n=605hrs). Conclusion: The validity of fixed site air monitoring data as a proxy for personal exposure to PM needs to be characterised so that the exposure error associated with this proxy measure is better understood.