UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

From measures to models : predicting exposure to air pollution among pregnant women Nethery, Elizabeth Michel Kennedy


Introduction: Exposure assessment is a key challenge in environmental epidemiology. When modeling exposures for populations, one should consider (1) the applicability of the exposure model to the health effect of interest (i.e. chronic, acute), (2) the applicability of the model to the population of interest, (3) the extent to which modeled exposures account for individual factors and (4) the sources of variability within the model. Epidemiological studies of traffic-related air pollution and birth outcomes have used a variety of exposure models to estimate exposures for pregnant women. These models are rarely evaluated, let alone specifically for pregnant women. Methods: Measured and modeled personal exposures to air pollutants (nitric oxide: NO, nitrogen dioxide: N0₂ , filter absorbance and fine particles: PM₂․₅) were obtained for 62 pregnant women from 2005-2006 in Vancouver, Canada. Exposures were measured for 48-hours, 1-3 times over the pregnancy. Mobility was assessed using Global Positioning System monitoring and self-reported activity logs; individual factors (dwelling characteristics, socio-economic factors) were assessed using questionnaires. Results: Modeled home concentrations using a traffic-based land-use regression model were moderately predictive of personal samples for NO only (Pearson's r=0.49). Models for NO including home and work locations explained more between subject variance than using home only (4% home only, 2 0 % with home and work). Modeled exposures using ambient monitoring stations were predictive of personal samples for NO (Pearson's r=0.54), absorbance (r=0.29) and PM₂․₅ (r=0.12) mainly due to temporal correlations (within subject variance: NO = 37% , absorbance = 11%, PM, 5 = 9%). Home gas stove was an important determinant of personal exposure for all pollutants. There was a significant (1 hour/day/trimester) increase in time spent at home with increased trimester of pregnancy. Conclusions: In this evaluation, based upon repeated 48-hour exposure measurements, models currently used in air pollution studies were moderately reflective of personal exposures, depending on the specific pollutant and model. Land-use regression shows promise for capturing spatial variability, especially when including mobility (work or school locations) in exposures, whereas monitor-based models are better for capturing temporal variability. Future models should include mobility, where possible, and consider the implications of increasing time at home over pregnancy in assessing exposures for pregnant women.

Item Media

Item Citations and Data


For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.