Comparison of Remote Sensing, Land-use Regression, and Fixed-site Monitoring Approaches for Estimating Exposure to Ambient Air Pollution Within a Canadian Population-based Study of Respiratory and Cardiovascular Health

O-31A5-5 Background/Aims: Remote sensing (RS) has emerged as a cutting edge approach for estimating ground-level concentrations of ambient air pollution. While the validity of RS has been demonstrated through comparisons to values obtained from fixed-site monitoring, no previous epidemiological studies have investigated the implications of using RS to characterize health risks. We examined respiratory and cardiovascular health outcomes associated with longer -term exposure measures of air pollution in a national population-based survey (N = 125,574), using estimates of annual average based on RS, land-use regression (LUR) models, and measured concentrations at the nearest fixed site monitor station. Methods: RS estimates of NO2 and PM2.5 were derived using satellite measurements from OMI, MODIS, and MISR. Multi-city LUR estimates were based on spatial models incorporating land-use characteristics such as traffic and industrial sources. Measured concentrations at the nearest regulatory continuous monitoring site were obtained from the National Air Pollution Surveillance Network. Self-reported health outcomes including diagnosis, age of onset, symptoms, and medication use for: asthma, bronchitis, COPD, heart disease, hypertension, congestive heart failure, angina, heart attack, and diabetes were collected through the Canadian Community Health Survey, a representative sample of Canadians 12 years of age and older. Results: RS estimates of PM2.5 and NO2 were highly correlated with ground-based measurements in North America (R = 0.9 and 0.8, respectively). Long-term exposures to ambient NO2 and PM2.5 were significantly associated with respiratory and cardiovascular health outcomes (OR = 1.1–1.4, P < 0.05) adjusting for age, sex, socioeconomic status, smoking status, and second-hand smoke. Effect estimates for RS were similar to those obtained using LUR and nearest fixed site monitor. Conclusion: These results suggest that RS can provide useful estimates of individual long-term exposure to ambient air pollution in epidemiologic studies, particularly in remote and rural areas for which monitoring and modeled air quality data are unavailable.