We did not only consider exposure during pregnancy but also periconceptional exposure because air pollution may affect sperm cells and ova through genetic or epigenetic mechanisms (Somers et al., 2004), and because placental methylation status at birth has been found to depend on exposures around implantation (Janssen et al., 2013). These early life exposure windows were significantly associated with both outcomes. The high correlation between entire pregnancy and trimester exposures hampers the identifi- cation of the time window in which the fetus is most vulnerable to air pollution. However, the observed effects tend to be higher for exposure in the first trimester than for exposure late in pregnancy (third trimester and last month), especially for SGA and for mod- erately preterm births. The stronger association observed for en- tire pregnancy exposure compared with other time windows might indicate the importance of chronic exposure. However, be- cause longer time windows typically have a smaller range and lower variability in exposure values than shorter windows, a
10 mg/m3 increase in average PM10 for the entire pregnancy is re-
latively larger than the same increase for a shorter exposure