Does residential mobility during pregnancy induce exposure misclassification for air pollution?

BACKGROUND: Prenatal exposure to outdoor air pollution has been shown to have health effects in many studies; low birth weight, preterm delivery, small for gestational age, and stillbirth are the most often cited. However, exposure of pregnant women is difficult to quantify, especially with regard to their mobility, which is rarely taken into account in epidemiological studies. This study aimed to assess the impact of mobility of pregnant women living in Paris, France, on their exposure estimates to nitrogen dioxide (NO(2)). METHODS: A total of 486 pregnant women were recruited in 5 maternity hospitals in Paris between January and April 2016. A questionnaire was used to collect mothers’ characteristics (demography, education, etc.) and to assess their daily mobility during pregnancy (time spent at work, commuting time and mode used to move from residential to occupational places). Daily NO(2) concentrations were estimated based on the combination of annual average concentrations modeled at the census block scale and daily concentrations measured from fixed monitoring stations. Different models were used to compare the exposure of pregnant women in residential and occupational places, also taking into account travel time and travel mode. The socioeconomic profile of the census blocks was characterized using a multi-component index. RESULTS: During the first trimester of pregnancy, women living in the least deprived census blocks were exposed to higher concentrations of NO(2) than those living in the most deprived ones. Occupational mobility had a small impact on exposure levels (average increase after taking account of mobility: + 0.52 μg/m(3)) which was not related to the socioeconomic profile of the women. The commuting mode made a greater difference (+ 1.46 μg/m(3) on average), in particular among women living in the most deprived census blocks. CONCLUSIONS: Our study illustrates that air pollution exposure can be underestimated when ignoring occupational mobility and commuting mode of pregnant women. This effect might be differential according to the neighborhood deprivation profile. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1186/s12940-018-0416-8) contains supplementary material, which is available to authorized users.