Critical exposure windows for wildfire PM_2.5 during pregnancy and preterm birth in Brazil

Wildfire smoke is an increasingly important air-pollution exposure, yet evidence on critical windows of susceptibility for pregnancy outcomes remains limited in Brazil. We evaluated the association between wildfire-related fine particulate matter (PM₂.₅) and preterm birth, focusing on lag-specific effects across pregnancy weeks and heterogeneity by sociodemographic groups. We conducted a nationwide ecological time-series study using weekly municipality-level panels in Brazil from 2002 to 2022. The outcome was weekly counts of preterm births, modeled with a quasi-Poisson regression and the log of total births as an offset. Wildfire-related PM₂.₅ and ambient (non-wildfire) PM₂.₅ were represented using distributed lag non-linear model (DLNM) cross-basis functions across lags 0-35 weeks. Temporal confounding was controlled using harmonic terms and a smooth function of time. Models were fit separately for each federative unit (26 states plus the Federal District) and pooled using random-effects multivariate meta-analysis. Effect modification was assessed via stratified outcome-offset models by infant sex, maternal age, maternal education, prenatal visits, and maternal race. Across 54, 531, 555 births and 5,264,325 preterm births (overall proportion 9.65%), pooled national estimates showed a distinct mid-gestation susceptibility window. A 10 μg/m³ increase in wildfire PM₂.₅ was associated with elevated preterm birth risk around lags 16-20 weeks before delivery, peaking near lag 18 (RR ≈ 1.011; 95% CI 1.003-1.020). Stratified analyses generally reproduced the mid-gestation positive window, with stronger and more precise associations among mothers aged 20-34 and ≥ 35 years, among those with 4-6 prenatal visits, and among White and Brown (mixed-race) births, while estimates for Black and Indigenous births were imprecise. These findings highlight the potential for prevention through wildfire-risk reduction, smoke exposure mitigation, and integration of smoke advisories into prenatal care and public-health surveillance.