From measures to models: an evaluation of air pollution exposure assessment for epidemiological studies of pregnant women
- 1School of Environmental Health, The University of British Columbia, Vancouver, BC, Canada
- 2Department of Health Care and Epidemiology, The University of British Columbia, Vancouver, BC, Canada
- Elizabeth Nethery, School of Environmental Health, The University of British Columbia, 3rd Floor, 2206 East Mall, Vancouver, BC, Canada V6T 1Z3;
- Accepted 22 November 2007
- Published Online First 10 December 2007
Objectives: To evaluate exposure estimation methods such as spatially resolved land-use regression models and ambient monitoring data in the context of epidemiological studies of the impact of air pollution on pregnancy outcomes.
Methods: The study measured personal 48 h exposures (NO, NO2, PM2.5 mass and absorbance) and mobility (time activity and GPS) for 62 pregnant women during 2005–2006 in Vancouver, Canada, one to three times during pregnancy. Measurements were compared to modelled (using land-use regression and interpolation of ambient monitors) outdoor concentrations at subjects’ home and work locations.
Results: Personal NO and absorbance (ABS) measurements were moderately correlated (NO: r = 0.54, ABS: r = 0.29) with monitor interpolations and explained primarily within-subject (temporal) variability. Land-use regression estimates including work location improved correlations for NO over those based on home postal code (for NO: r = 0.49 changed to NO: r = 0.55) and explained more between-subject variance (4–20%); limiting to a subset of samples (n = 61) when subjects spent >65% time at home also improved correlations (NO: r = 0.72). Limitations of the GPS equipment precluded assessment of including complete GPS-based mobility information.
Conclusions: The study found moderate agreement between short-term personal measurements and estimates of ambient air pollution at home based on interpolation of ambient monitors and land-use regression. These results support the use of land-use regression models in epidemiological studies, as the ability of such models to characterise high resolution spatial variability is “reflected” in personal exposure measurements, especially when mobility is characterised.
▸ Additional appendices are published online only at http://oem.bmj.com/content/vol65/issue9
Funding: This research was funded by the British Columbia Centre for Disease Control, via an agreement with Health Canada as part of the U.S.-Canada Border Air Quality Strategy. Additional support was provided by the Center for Health and Environment Research at The University of British Columbia, funded by the Michael Smith Foundation for Health Research. Elizabeth Nethery was supported by a trainee award from the Michael Smith Foundation for Health Research.
Competing interests: None.