Child marriage, maternal serum metal exposure, and risk of preterm birth in rural Bangladesh: evidence from mediation analysis

Preterm birth, defined as delivery before 37 completed weeks of gestation, is associated with substantial short- and long-term morbidity and accounts for 35% of neonatal mortality and approximately 16% of deaths in children under 5 years. Globally, about 15 million premature babies are born each year, with 81.1% in Asia and Sub-Saharan Africa. Bangladesh has the highest estimated national prevalence of preterm birth at 19.1%, with even higher rates in rural areas, imposing significant burdens. Multiple socioeconomic and demographic factors contribute to preterm birth risk, including low socioeconomic status, multiple pregnancies, higher maternal age, history of abortion, pregnancy complications, and child marriage (marriage before 18 years). Women with a history of child marriage may have poorer nutrition and greater opportunities for harmful metal exposures due to living and working environments and income status. Bangladesh experiences high environmental and industrial pollution, leading to elevated exposures to several metals via multiple routes, often exceeding levels in developed countries and World Health Organization safety thresholds. Prior cohort work in Bangladesh has examined effects of metal exposures on birth outcomes, with specific metals such as arsenic, lead, cadmium, and copper associated with preterm birth in hypothesis-driven studies, typically assessing metals individually and at a single timepoint during pregnancy. Few studies have explored causal mechanisms linking upstream socioeconomic factors, environmental metal exposures, and preterm birth. Therefore, leveraging a rural Bangladesh birth cohort, this study measured 19 serum metals in the first and second trimesters to identify novel metal markers, critical exposure windows, and to use mediation analysis to identify upstream socioeconomic factors influencing metal exposure and preterm birth risk.

The paper situates its research within evidence that environmental metal exposures are elevated in Bangladesh and have been linked to adverse birth outcomes. Previous studies, often hypothesis-driven, reported associations between metals such as arsenic, lead, cadmium, and copper and preterm birth or other adverse outcomes. However, prior work largely evaluated metals individually and at one pregnancy timepoint, and few studies examined how socioeconomic determinants (e.g., child marriage) may influence metal exposures that, in turn, affect preterm birth risk. This gap motivates assessing multiple metals across trimesters and applying mediation analysis to elucidate causal pathways from socioeconomic factors to preterm birth via metal exposures.

Design and population: Prospective birth cohort in rural Bangladesh (2008–2011). Inclusion criteria: maternal age ≥18 years, ultrasound-confirmed singleton pregnancy ≤16 weeks gestation, use of tube wells supplying groundwater as primary drinking source, and plan to remain at current residence during pregnancy. A total of 780 pregnant women had first-trimester serum samples; 610 had second-trimester samples. Ethical approvals were obtained from Harvard T.H. Chan School of Public Health, Nanjing Medical University, and Dhaka Community Hospital Trust; informed consent was obtained. Exposure assessment: Peripheral venous blood was collected after overnight fast into EDTA tubes, stored at 4 °C, transported to the Trace Metals Laboratory at Harvard Chan School, and centrifuged to obtain serum. Labware was acid-cleaned. For analysis, 60 µL serum was diluted to 1.8 mL with 1% HNO3, 0.1% Triton X-100, and internal standards. Nineteen metals were quantified via ICP-MS (iCAP Qc, Thermo Scientific) at Nanjing Medical University: Na, Mg, K, Ca, Mn, Fe, Co, Cu, Zn, As, Rb, Sr, Mo, Cd, Sb, Ba, Hg, Tl, U. Limits of detection (LOD) were set at 3× the average of 10 blank measurements; concentrations below LOD were imputed as LOD/2. All concentrations were log10-transformed. For metals with >50% below LOD (Mn, Cd, U), values were dichotomized (detected vs non-detected) for analyses. Quality control employed Seronorm Trace Elements Serum L-2 with standards run every 20 samples. Outcomes and covariates: Preterm birth defined as live birth <37 completed weeks. Gestational age was determined by ultrasound (gestational sac mean diameter at 4–7 weeks or crown-rump length at 7–16 weeks). Covariates collected via structured questionnaire at enrollment included maternal age, BMI, marriage age, maternal and spouse education, number of past pregnancies, family income, and secondhand smoking. Statistical analysis: Baseline characteristics compared via t-test or Mann-Whitney U for continuous variables and chi-square for categorical variables. Correlations among metals within and across trimesters were assessed using Pearson or Spearman correlations. Mixed-effects logistic models with repeated measures (first and second trimester log-transformed metal concentrations) estimated odds ratios (ORs) for preterm birth, adjusting for maternal age, marriage age, secondhand smoking, BMI categories, parental education levels, and family income; false discovery rate (FDR) controlled multiple comparisons (q<0.05 significant). Trimester-specific logistic regression models assessed critical exposure windows for each metal with the same covariate adjustments. Partial correlation analyses examined associations between socioeconomic factors (notably marriage age) and metal concentrations in each trimester. Mediation analysis evaluated whether marriage age influenced preterm birth risk via metal concentrations using two models: an outcome model for preterm birth including marriage age, ln(metal), and covariates, and a mediator model for ln(metal) including marriage age and covariates, allowing decomposition into direct and indirect (mediated by metals) effects. Analyses were conducted in R v3.6.2 with two-sided P<0.05 unless otherwise specified.

- Cohort characteristics: Among 780 singleton live births, 175 (22.4%) were preterm. Mean marriage age was 17.46 ± 2.31 years; 44.4% married before 18. Women with preterm births were more likely to have married before 18, had secondhand smoking exposure, lower baseline weight, lower parental education, and lower household income. - Metals associated with preterm birth (mixed-effects model across both trimesters): Zinc (Zn), arsenic (As), strontium (Sr), and barium (Ba) were significantly associated with preterm birth risk after FDR correction (q<0.05). - Critical exposure windows (trimester-specific models): First trimester: Zn (OR=0.28; 95% CI: 0.13–0.58; P=6.0×10^-4; FDR q=5.7×10^-3), As (OR=1.49; 95% CI: 1.20–1.84; P=3.0×10^-4; FDR q=5.7×10^-3), and Sr (OR=0.39; 95% CI: 0.20–0.74; P=4.3×10^-3; FDR q=2.7×10^-2) were significantly associated with preterm birth. Second trimester: Ba remained significantly associated (OR=1.25; 95% CI: 1.10–1.41; P=7.0×10^-4). - Marriage age and preterm birth: Restricted cubic spline analysis showed an exponential dose–response relationship between younger marriage age and higher preterm birth risk, particularly among women married before 18 years; women who married at ≥18 years had substantially lower risk after adjustment for confounders. - Socioeconomic determinants and metals: Marriage age was independently and significantly correlated with all four preterm-associated metals after adjusting for sociodemographic factors. - Mediation: Zn, As, Sr, and Ba collectively mediated 30.2% of the effect of marriage age on preterm birth risk.

The study addresses the research question by demonstrating that specific maternal serum metal exposures are associated with preterm birth and that these exposures partly mediate the relationship between child marriage (younger marriage age) and preterm birth. Early pregnancy appears to be a critical window for Zn, As, and Sr, while mid-pregnancy is critical for Ba. The exponential increase in preterm risk with younger marriage age, coupled with significant associations between marriage age and metal concentrations, supports a pathway in which socioeconomic factors influence environmental exposures that impact birth outcomes. These findings highlight the importance of upstream interventions, such as delaying marriage age and reducing environmental metal exposures, to mitigate preterm birth risk in high-exposure settings like rural Bangladesh.

Maternal serum metal exposures, specifically Zn, As, Sr (first trimester) and Ba (second trimester), are significantly associated with preterm birth risk in rural Bangladesh. Younger marriage age confers a substantially higher risk of preterm birth, and approximately 30% of this effect is mediated through these metal exposures. The study identifies critical exposure windows and an upstream socioeconomic determinant (child marriage), providing insights that can inform interventional strategies aimed at delaying marriage age and reducing maternal metal exposures. Future research should further validate these findings, elucidate biological mechanisms, and evaluate targeted public health interventions in similar high-exposure contexts.

- Measurement limitations: For Mn, Cd, and U, more than 50% of samples were below the limit of detection in one or both trimesters, necessitating dichotomization (detected vs non-detected), which may reduce sensitivity to dose–response relationships. - Sample availability: Only 610 of 780 participants had second-trimester serum samples, potentially affecting power for trimester-specific analyses. - Observational design: Although mediation analysis was conducted, causal inference is limited by the observational cohort design and potential residual confounding. - Generalizability: Findings are from rural Bangladesh with high environmental metal exposures and may not generalize to regions with different exposure profiles.