Prenatal Exposure to PM2·5 and Pregnancy Loss in Spain, 2003-2017: A Nationwide Time Series Analysis

Background: Air pollution has been associated with adverse birth outcomes, but few studies have evaluated pregnancy loss, and none at a nationwide level. We evaluated prenatal PM2·5 exposure and pregnancy loss using live birth-identified conceptions (LBIC) as a proxy to identify critical windows of susceptibility. Methods: We conducted a nationwide time series study using a novel approach based on LBICs derived from administrative birth data from Spain between 2003 and 2017. We applied a two-stage modeling approach. Location-specific quassi-Poisson regression models combined with distributed lag nonlinear models were fitted, adjusting for temperature and temporal trends. Location-specific estimates were pooled using a multivariate mixed-effects meta-analysis. We further applied g-computation models to estimate the number of lost pregnancies attributable to PM2·5 exposure. Findings: Prenatal exposure to PM2·5 during gestational weeks 6 to 21 was associated with a decrease in LBIC counts (RR per 10 µg/m3: 0·92, 95% CI: 0·87, 0·96). In g-computation analyses, the largest week-specific estimate was observed around gestational week 12, where a 5 µg/m3 increase in PM2·5 was associated with 49·1 (95% CI: 12·2, 129·6) additional pregnancy losses per conception week. The cumulative estimate across the window of susceptibility was 561·8 (95% CI 158·7, 1,277) additional pregnancy losses per conception week. Interpretation: Our findings showed that prenatal PM2·5 exposure was associated with reduced LBIC counts, consistent with increased pregnancy loss risk during weeks 6 to 21. These findings have important implications for both pregnant individuals and fetal health, highlight the importance of reducing air pollution exposure during early-to-mid pregnancy.