Abstract
BACKGROUND: Exposure to PM(2·5) from wildfire smoke during pregnancy has been implicated as a risk factor for preterm birth. We investigated this association in the prospective nationwide US Environmental Influences on Child Health Outcomes (ECHO) Cohort, focusing on prenatal wildfire PM(2·5) exposure intensity, duration, and timing. METHODS: In this cohort analysis, we included live singleton births recorded in the ECHO Cohort with available data on gestational age at birth and birthweight and dates of conception between Jan 1, 2006, and March 20, 2020. Census tract-level estimates of daily mean wildfire-derived PM(2·5) for the years 2006-20 from a previous machine learning model were linked to residential address history. We calculated the mean concentration of daily wildfire PM(2·5), days with wildfire PM(2·5) (>0, ≥2·5, ≥5·0, and ≥10·0 μg/m(3); termed smoke days) and consecutive smoke days (2, 3, or ≥4 days; termed smoke waves) above the prespecified concentration thresholds across pregnancy. Associations of cumlative pregnancy wildfire PM(2·5) exposure with preterm birth (delivery before 37 weeks of gestation) were analysed by adjusted pooled logistic regression in the nationwide ECHO sample and in the US West census region. Associations between smoke days in gestational weeks 0-35 and preterm birth were evaluated by logistic regression in the national sample. FINDINGS: We included 20 034 births from 30 ECHO Cohort study sites, with residences during pregnancy in all 48 contiguous US states and the District of Columbia. 1687 (8·4%) of the 20 034 infants were preterm. The mean daily wildfire PM(2·5) concentration during pregnancy was 0·36 μg/m(3) (SD 0·46), with exposure to a mean of 22·2 smoke days (SD 16·6) of any wildfire PM(2·5) concentration (>0 μg/m(3)). Estimates of association between wildfire PM(2·5) exposure metrics and preterm birth included the null in nationwide analyses; whereas, in the US West sample (N=5807), we estimated increased odds of preterm birth associated with mean daily wildfire PM(2·5) (odds ratio [OR] 1·139 per 1-μg/m(3) increase [95% CI 1·001-1·296]), exposure to smoke days with a wildfire PM(2·5) concentration of 5·0 μg/m(3) or greater (OR 1·018 per additional smoke day [1·003-1·032]) and 10·0 μg/m(3) or greater (OR 1·030 [1·006-1·054]), and exposure to ≥4-day smoke waves of 5·0 μg/m(3) or greater (OR 1·185 per additional smoke wave [1·044-1·347]) and 10·0 μg/m(3) or greater (OR 1·232 [1·029-1·475]). At the national level, by week of gestation, associations with preterm birth were observed in mid-pregnancy for smoke days with wildfire PM(2·5) concentrations above 0 μg/m(3), of 2·5 μg/m(3) or greater, and of 5·0 μg/m(3) or greater, and in late pregnancy for smoke days of 10·0 μg/m(3) or greater. INTERPRETATION: In a prospective cohort, we observed increased odds of preterm birth associated with wildfire PM(2·5) exposure in the western USA, with findings suggesting an exposure-response relationship for increasing exposure intensity and duration. Preterm birth was also associated with exposure to smoke days in mid-to-late pregnancy at the national level. For practice and policy, these findings support the need for public health interventions aimed at reducing exposure to wildfire smoke during pregnancy. FUNDING: ECHO Program, US National Institutes of Health Office of the Director.