The relationship between air pollutants and maternal socioeconomic factors on preterm birth in California urban counties

The study addresses whether maternal exposure to ambient air pollutants (PM2.5 and O3) interacts with neighborhood socioeconomic factors to influence the risk of preterm birth (<37 weeks gestation) in California urban counties. Preterm birth affects about 10% of infants in California and is a leading cause of perinatal morbidity and mortality, with long-term developmental and chronic disease consequences. Preterm birth is multifactorial, involving genetic, social, and environmental determinants. Prior work identifies risk factors such as maternal age, parity, comorbidities, prior preterm birth, and social determinants including systemic racism and chronic stress (“weathering”). Neighborhood socioeconomic deprivation has been linked to preterm birth independent of individual-level SES. Environmental exposures, particularly air pollution (PM2.5 and O3), have been associated with preterm birth through pathways involving inflammation, oxidative stress, endocrine disruption, and impaired placental oxygen transport. Given environmental justice concerns—higher pollution exposures in communities of color and low SES—and the reproductive justice framework, the study tests the hypothesis that exposures to PM2.5 or O3 interact with neighborhood SES to increase preterm birth risk.

Prior studies demonstrate associations between traffic-related air pollution and preterm birth, including increased odds of early preterm birth with higher TRAP during the second trimester. Multiple U.S. and international studies have linked ozone exposure to preterm birth. Communities of color and those with lower SES bear disproportionate air pollution burdens, intersecting with existing health inequities in preterm delivery among Black births. Environmental justice emphasizes equal rights to clean environments, and air monitoring under the U.S. EPA’s Clean Air Act provides data to assess compliance and potential health impacts. Studies incorporating neighborhood-level sociodemographic covariates often find stronger associations between pollution and preterm birth, suggesting effect modification by socioeconomic context. The reproductive justice framework underscores the right to safe and healthy environments for parenting, motivating investigation of pollutant exposures alongside social factors during pregnancy.

Design and setting: Retrospective cohort study of births from six California urban counties (Alameda, Contra Costa, Fresno, Los Angeles, San Diego, San Francisco) from 2007–2011. Data sources: Birth Statistical Master files (California Department of Public Health) provided geocoded maternal residential addresses at delivery and individual covariates. Hospital discharge records (Office of Statewide Health Planning and Development) were probabilistically linked for additional maternal clinical variables. Neighborhood socioeconomic variables were obtained from the American Community Survey (2007–2011 5-year estimates). Ambient air pollutant data (daily 8-h max O3 and 24-h average PM2.5) were derived from the EPA Bayesian space-time fusion model integrating monitoring network data and air quality model outputs to census tract centroids (12×12 km grid). Study population: Initial N=1,132,953 births. Exclusions: multiples (n=36,160), implausible/missing gestational age (<20 or >42 weeks or missing; n=11,578), missing key covariates (maternal age, education, race/ethnicity; n=49,169), unavailable air pollutant data (n=68,394), incomplete exposure coverage (required 3 months pre-pregnancy and at least 5 months post-conception; n=13,701). Final analytic sample: 953,951 singleton births. Exposure assessment: For each birth, pollutant estimates were summarized into monthly averages backward from live birth up to 12 months and then aggregated into 3-month periods (e.g., trimesters and 3 months pre-pregnancy). Pollutants were analyzed as continuous (per 1-unit increase: µg/m³ for PM2.5, ppb for O3) and dichotomized at the median for some analyses (PM2.5: 12.9 µg/m³; O3: 39 ppb). Outcomes: Total preterm birth (<37 weeks) and early preterm birth (<32 weeks). Covariates: Maternal age (<20, 20–24, 25–29, 30–34, ≥35), race, ethnicity, education, prenatal care in first trimester, cigarette use in first trimester, year of birth, insurance/payment source at delivery; additional clinical variables from hospital discharge included smoking status, prenatal care details, pre-pregnancy and gestational diabetes and hypertension, maternal height and pre-pregnancy weight. Statistical analysis: Logistic regression estimated odds ratios (ORs) and 95% CIs for associations between pollutant exposures and outcomes by exposure windows. Models were adjusted per an a priori plan for the listed covariates. Effect modification by neighborhood SES (poverty, GINI index, education, public assistance, unemployment, median household income) was evaluated using multiplicative interaction terms; SES variables were dichotomized (high/low). Mixed-effects logistic models with random intercepts for census tract accounted for spatial autocorrelation. Sensitivity analysis addressing fixed cohort bias excluded births with conception dates ≥20 weeks before cohort start or >42 weeks before cohort end (excluded ~8% of births). Ethical approvals were obtained from California’s Committee for the Protection of Human Subjects and IRBs at UCSF and Stanford. Geocoding used ArcGIS and address correction with ZP4 software.

• Study population: 953,951 singleton births; preterm birth prevalence 9.2% (87,495). Highest preterm prevalence among Black women (12.9%). Preterm births were more common at maternal ages <20 and ≥35 years. About 26.5% had a college degree or higher; 48.6% were covered by Medi-Cal. - Neighborhood SES differences: Census tracts of preterm births had lower median household income ($55,716 vs $59,497 among term births) and higher poverty, unemployment, and public assistance participation; GINI index did not differ significantly. - Primary pollutant associations: Positive associations were observed between pollutant exposure and preterm birth. For total preterm birth, the highest risk for PM2.5 was for whole pregnancy exposure per 1 µg/m³ increase (AOR 1.05, 95% CI: 1.03, 1.07). The strongest O3 association was for the 3-month pre-pregnancy window per 1 ppb increase (AOR 1.08, 95% CI: 1.06, 1.10). For early preterm birth (<32 weeks), the strongest association was with O3 in the third trimester (AOR 1.05, 95% CI: 0.93, 1.10), while PM2.5 showed no significant associations across periods. - Mixed-effects models: Associations persisted but were smaller: PM2.5 averaged across pregnancy (AOR 1.02, 95% CI: 1.01, 1.02) and O3 during the 3 months pre-pregnancy (AOR 1.03, 95% CI: 1.02, 1.04). - Effect modification: Interactions suggested increased risk of total preterm birth among mothers in higher socioeconomic areas with higher PM2.5 (e.g., higher GINI index areas had AOR 1.04, 95% CI: 1.01, 1.07 per 1 µg/m³; lower GINI areas AOR 0.97, 95% CI: 0.94, 0.99). For early preterm birth, most interactions were not significant; a notable finding was increased risk in areas with high unemployment and high PM2.5 in specific exposure periods. However, after sensitivity analysis for fixed cohort bias, these interaction associations were not statistically significant. - Seasonal patterns: Associations varied by season of conception. Spring conceptions showed larger O3–preterm associations (AORs ≈1.03 to 4.48 across exposure periods), whereas summer conceptions showed stronger PM2.5–preterm associations (AORs ≈1.06 to 1.41). - Temporal trend: High PM2.5 exposure category declined from 31.5% in 2007 to 2.3% in 2011, yet associations with preterm birth remained detectable.

Findings support the hypothesis that maternal exposure to ambient PM2.5 and O3 is associated with increased risk of preterm birth, with critical windows differing by pollutant (whole pregnancy for PM2.5; pre-pregnancy and certain gestational periods for O3). Despite declining PM2.5 levels over the study period, associations persisted, reinforcing concerns that even relatively low contemporary concentrations may affect adverse birth outcomes. The observed effect modification by neighborhood socioeconomic factors suggests contextual influences on vulnerability; however, these interactions were sensitive to fixed cohort bias adjustments and were not robustly significant afterward. Seasonal heterogeneity implies that timing of conception and seasonal pollutant patterns or co-exposures may influence risk. The results align with prior literature linking air pollution to preterm birth and underscore environmental and reproductive justice considerations, given disproportionate pollution burdens and existing disparities in preterm birth. These findings can inform policies targeting air quality improvements and maternal health equity.

In a large retrospective cohort from six California urban counties (2007–2011), maternal exposure to PM2.5 and O3 showed small but significant associations with preterm birth risk, with pollutant- and window-specific effects. Evidence for interaction with neighborhood socioeconomic context was suggestive but not consistent after accounting for fixed cohort bias. Results corroborate prior studies and indicate that adverse associations persist at contemporary pollutant levels, supporting continued air quality improvements and targeted maternal health interventions. Further research should elucidate biological mechanisms, refine critical exposure windows, explore intersectional social-environmental interactions, and assess interventions and policies to reduce disparities in preterm birth.

• Neighborhood exposure context was assigned at the census tract of maternal residence at delivery, which may not reflect where individuals spend their time; census tract served as a proxy for neighborhood. - Potential fixed cohort bias was identified and addressed via sensitivity analysis; interaction findings were not robust after this adjustment. - Strong correlations among census tract-level SES variables may complicate disentangling specific socioeconomic effects. - Some births lacked pollutant data or complete exposure windows and were excluded, which may affect generalizability.