Taking into Account Space-Time Activity and Indoor Air Levels in the Estimation of Exposure to Atmospheric Pollutants during Pregnancy: Does It Make a Difference?

Abstract

Background: Epidemiological studies have suggested adverse affects of atmospheric pollutants on birth outcomes. In most studies, exposure assessment did not consider space-time activity or indoor air pollution levels. Objectives: Our objective was to compare different approaches allowing to characterize atmospheric pollution exposure during pregnancy. A secondary aim was to quantify the impact of the cleaning of the GPS data used to assess space-time activity. Methods: We recruited 40 pregnant women living in an urban area and had them carry a GPS during one week at each trimester of pregnancy; indoor levels of fine particulate matter (PM2.5) were measured in the homes of 10 women. Outdoor air pollutants concentrations were estimated from a dispersion model with a high spatial resolution (10x10 meters grid) and were corrected for seasonal variations. We compared trimester- specific exposure levels of the women assessed only from outdoor air pollutants levels at the home address (Model 1) with approaches incorporating space-time activity (raw GPS data: Model 2, clean GPS data: Model 3) and with a model incorporating indoor PM2.5 levels when the woman was home (Model 4). Results: Considering only outdoor air pollutants levels, the exposure model taking into account space-time activity with clean GPS data (Model 3) was more strongly correlated with the model estimating the levels of air pollutants at the home address (Model 1, r=0.85) than with the model taking into account space-time activity with raw GPS data (Model 2, r=0.73). Each of these models based on outdoor air pollutants levels (Models 1, 2 and 3) were moderately correlated (r=0.4-0.5) with the clean GPS data model incorporating PM2.5 indoor measures when the woman was home (Model 4). Conclusion: GPS data require cleaning as classical GPS often lose their signal indoors. In this urban area, a purely outdoor model had a moderate correlation with an exposure estimate incorporating indoor levels.