Surrounding greenness is associated with lower risk and burden of low birth weight in Iran

The study addresses whether higher surrounding greenness during pregnancy reduces the risk of low birth weight (LBW) and term low birth weight (TLBW), particularly in low- and middle-income countries (LMICs) where evidence is scarce. Although greenness has been linked to multiple health benefits and favorable birth outcomes, prior LBW evidence is inconsistent and predominantly from high-income countries (HICs). Critical gaps include a lack of large, nationally representative data in LMICs and absence of quantitative estimates of LBW burden reductions achievable through specified greenness targets. The purpose of this nationwide Iranian study (2013–2018) was to derive a nationally representative exposure–response relationship between prenatal greenness and LBW risk, and to estimate the avoidable LBW/TLBW burden under counterfactual scenarios of improved greenness. The findings have implications for meeting the Global Nutrition Targets 2025 to reduce LBW and for informing greening policies in LMICs.

Prior population-based studies associate residential greenness with improved birth outcomes, but LBW findings are mixed. A meta-analysis of 10 observational studies (mostly HICs) suggested ~10% reduced LBW odds per 0.1-unit NDVI increase, but with heterogeneity in effect sizes. Evidence from LMICs remains limited and inconsistent, with studies in China showing reduced odds in some regions and null findings in others. Few studies provide national-scale evidence with millions of births, and none have quantified the LBW burden potentially avoidable in LMICs by achieving predefined greenness targets. The study situates itself to fill these gaps by providing national-level evidence from Iran and burden estimates under counterfactual greenness improvements.

Design and population: Nationwide retrospective birth cohort including 4,068,843 birth records from 749 hospitals across 31 Iranian provinces (2013–2018). After exclusions (stillbirths, neonatal deaths, undetermined sex, unmatched exposure), 4,021,741 live births were analyzed. Outcomes: Primary outcome LBW (birth weight <2500 g, irrespective of gestational age). Secondary outcome TLBW (birth weight <2500 g among gestations ≥37 weeks). Gestational age was calculated from last menstrual period to birth date. Exposure assessment: Greenness exposure during the entire pregnancy quantified using MODIS-derived NDVI and EVI (Terra satellite), 16-day, 250 m resolution. Both indices range −1 to 1; negative/water-body values treated as zero. Maximum value composite across pregnancy was used to mitigate cloud/season effects. Greenness was averaged within circular buffers of 500, 1000, 2000, and 3000 m around each maternal delivery hospital, representing approximately 10–30 min walking distances, to capture local and larger-scale green space. Environmental covariates: Monthly mean PM2.5 at 0.1° × 0.1° from ACAG global datasets; temperature and dew point from ERA5 (0.1°), used to compute relative humidity via National Weather Service method. Exposures aggregated as means across months from last menstrual period to birth. Individual covariates: Maternal demographics (age, education attainment, residence, nationality, delivery type, parity, number of fetuses, delivery hospital), fetal variables (infant sex, season of birth, gestational age). Sparse missing socioeconomic data imputed to modal category to ensure model convergence. Statistical analysis: Multiple logistic regression with sequential adjustments: Model 1 (unadjusted); Model 2 (maternal age, infant sex, gestational age); Model 3 (maternal demographics and fetal variables); Model 4 (fully adjusted: maternal demographics, fetal variables, environmental factors). Temperature and relative humidity modeled with natural splines (3 df) to allow nonlinearity. Associations expressed as odds ratios (ORs) with 95% CIs per 0.1-unit increase in NDVI/EVI and across exposure quartiles (Q1 reference; linear trend tested using quartile medians as continuous terms). Nonlinear exposure–response (E–R) assessed via restricted cubic splines (three knots at 10th, 50th, 90th percentiles), with likelihood-ratio tests comparing spline vs linear models. Subgroup and sensitivity analyses: Subgroups by maternal age (≤25, 26–35, >35), education (below high school, high school, college+), and residence (city, village); heterogeneity tested via fixed-effect meta-regression. Sensitivity analyses excluded extreme birthweights (<500 or ≥5000 g), multiple gestations, mothers <13 or >50 years, and mothers with pre-existing chronic diseases or delivery complications; models additionally adjusted for provincial-level SES variables. Burden estimation (counterfactual): Assuming causality, greenness-attributable fractions and avoidable numbers of LBW/TLBW were estimated for 2015 by linking fitted E–R functions to gridded live births and maternal greenness exposures, under scenarios achieving mean NDVI/EVI within each buffer. Uncertainty propagated using lower/upper bounds of E–R estimates. Analyses conducted in R (rms, mumeta, raster, ggplot2).

- Cohort: 4,021,741 live births (2013–2018) across 31 Iranian provinces; 263,728 LBW (6.6%) and 121,852 TLBW (3.0%). Mean birthweight: LBW 2024.8 ± 479.2 g; TLBW 2196.7 ± 407.9 g. - Exposure differences: Mothers of LBW/TLBW infants had consistently lower greenness exposures across 500–3000 m buffers. Example means: LBW vs normal births NDVI ~0.287–0.292 vs 0.315–0.319; EVI ~0.207–0.210 vs 0.227–0.230. - Main associations (fully adjusted, per 0.1-unit increase, 3000 m): - LBW: NDVI OR 0.911 (95% CI 0.908–0.914); EVI OR 0.885 (0.881–0.889). - TLBW: NDVI OR 0.899 (0.895–0.902); EVI OR 0.871 (0.866–0.875). - Protective associations observed across buffers; slightly attenuated with narrower buffers and after environmental adjustment. - Quartile analyses (3000 m, fully adjusted): Monotonic trends (P-trend <0.001). Highest vs lowest quartile: - NDVI: LBW OR 0.961 (0.960–0.963); TLBW OR 0.956 (0.955–0.958). - EVI: LBW OR 0.963 (0.961–0.964); TLBW OR 0.958 (0.956–0.960). - Exposure–response: Approximately L-shaped curves for NDVI and EVI, with larger LBW/TLBW risk reductions at low greenness (e.g., NDVI <0.4 in 3000 m), and diminishing benefits at higher greenness levels (nonlinearity P<0.001). - Subgroups (3000 m): Stronger greenness-LBW associations among younger mothers (≤25 yrs; NDVI OR 0.895, 95% CI 0.890–0.900), less educated (below high school; 0.897, 0.893–0.901), and village residents (0.893, 0.887–0.898). Similar patterns for TLBW and with EVI. - Burden under counterfactual mean greenness targets (2015): - Avoidable LBW: 3,931–5,099 cases (4.4%–5.6% of total LBW), highest for 3000 m buffers. - Avoidable TLBW: 2,173–2,697 cases (5.6%–6.9% of total TLBW). - Attributable fractions and avoidable numbers increase with larger buffers (500 m to 3000 m).

Findings demonstrate that higher prenatal exposure to surrounding greenness is associated with lower odds of LBW and TLBW in a large, nationally representative LMIC setting. The approximately L-shaped exposure–response indicates that increasing vegetation in low-greenness areas yields the greatest marginal benefits, consistent with evidence from both HICs and LMICs that risk reductions plateau at higher greenness. Subgroup analyses suggest greenness may confer disproportionate benefits to socioeconomically disadvantaged groups—younger, less educated, and village-dwelling mothers—supporting the equigenic environments hypothesis that natural environments can mitigate health inequalities. Comparisons with prior literature show slightly smaller effect sizes than meta-analytic estimates mostly from HICs, highlighting contextual differences in demographics, baseline vegetation, and environmental co-exposures. Counterfactual burden estimates suggest that achievable gains in greenness could prevent a meaningful fraction of LBW/TLBW cases nationally, and co-benefits with air quality improvements may further advance progress toward Global Nutrition Targets for LBW. Overall, the study strengthens causal plausibility through robustness and dose–response patterns, and underscores policy relevance of urban and rural greening as a public health measure in LMICs.

This nationwide Iranian cohort provides robust evidence that higher surrounding greenness during pregnancy is associated with reduced risks of LBW and TLBW, with strongest marginal benefits at low greenness levels and greater benefits among socially disadvantaged mothers. Under plausible counterfactual scenarios achieving mean NDVI/EVI, thousands of LBW/TLBW cases could be avoided annually. These results support greening strategies as potential public health interventions to reduce LBW burden and health inequities in LMICs. Future research should employ multi-country designs, finer exposure measures capturing green space type and quality, residential address-based assessments, and causal inference frameworks to refine burden estimates and guide targeted greening interventions.

- Exposure misclassification: Maternal greenness exposure proxied by delivery hospital location rather than residential address; buffer analyses (500–3000 m) partly mitigate this concern. - Greenness metrics: NDVI/EVI capture vegetation density but not type, accessibility, or quality of green spaces, which may influence health effects. - Unmeasured confounding: Individual-level factors such as BMI and smoking were unavailable; despite SES adjustments and robustness checks, residual confounding cannot be excluded. - Attributable burden assumptions: Counterfactual avoidable LBW/TLBW estimates are based on associations (not established causality) and should be interpreted cautiously. - Data processing: Negative/water-body index values were set to zero, potentially introducing minor exposure measurement error. - Generalizability: Findings pertain to Iran (2013–2018) and may differ in settings with distinct greenness profiles, healthcare access, or environmental co-exposures.