Abstract

Background/Aim: Methods for estimating air pollutant exposures for epidemiological studies are rapidly evolving with more complex methods emerging. Our aim was to determine the agreement between two methods for estimating NO2 exposures for a children’s cohort (CAPS), based in Sydney, Australia. The two methods were a traditional city-scale land use regression (LUR) model and a national satellite-based LUR (Sat-LUR) model. Methods: We used methods consistent with ESCAPE methodology to develop a NO2 LUR model, using Ogawa passive sampler data collected at 46 locations across western Sydney (two week periods over three seasons in 2013 and 2014). We collected data on land use, population density, and traffic volumes as potential predictor variables. These LUR estimates were compared to those from a national Sat-LUR model, which used data from the OMI sensor on the Aura satellite and land use data to predict NO2 concentrations at 68 fixed-site monitors located across Australia. Each model was used to estimate annual average NO2 concentrations for 2013 at each of 964 cohort addresses. A Bland-Altman assessment was used to compare the estimates. Results: The traditional LUR model predicted 84% of variability in NO2 (adjusted R2=0.84; RMSE 1.2ppb; 82% cross validation) with predictors being major roads, population and dwelling density, heavy traffic and commercial land use. The Sat-LUR adjusted R2 was 0.81 (RMSE 1.4 ppb) and predictor variables included major roads, industrial emissions, industrial land use and open space. The annual average means were 7.3 (SD 1.95) and 6.9 (SD 1.96) respectively. Comparing the two sets of estimates, the mean difference was 0.31 ppb (CI0.209 to0.404), limits of agreement ranged from -2.78 to 3.40, and the ICC was 0.69. Conclusions: The results indicate very good agreement between traditional LUR and satellite-based methods for estimating NO2 in Sydney, providing confidence in their use for exposure-response analyses.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call