Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Aug 27.
Published in final edited form as: Environ Int. 2025 Jul 5;202:109666. doi: 10.1016/j.envint.2025.109666

Preconception, prenatal and early childhood exposure to green space and risk of neurodevelopmental delays: a national cohort study among Medicaid enrollees

Hayon Michelle Choi a, Krista F Huybrechts b, Sonia Hernandez-Diaz c, Xinye Qiu a, Michael Leung a, Peter James a,d,j, Matthew Shupler c, Wanyu Huang a, Yaguang Wei a,e, Antonella Zanobetti a, Christopher J McDougle f,g, Joel Schwartz a, Brent Coull h, Marc Weisskopf a,c, Stefania Papatheodorou c,i,*
PMCID: PMC12380237  NIHMSID: NIHMS2105994  PMID: 40639188

Abstract

Background:

Exposure to green space is associated with children’s mental health, but its impact on neurodevelopment has been underexplored, especially in socioeconomically disadvantaged populations. This study examined the link between exposure to green space before, during, and after pregnancy and neurodevelopmental delays in children enrolled in Medicaid.

Methods:

This cohort study of 1,841,915 mother–child pairs used data from the Medicaid Analytic Extract (MAX) from 2001 to 2014, with up to 14 years of follow-up. The population of pregnant women enrolled in Medicaid is characterized by younger age, racial and ethnic diversity, lower income levels, and includes individuals with disabilities. Green space exposure was measured using the Normalized Difference Vegetation Index (NDVI) at the maternal residential zip code level. We examined exposure to green space during the preconception, prenatal, and postnatal periods to capture critical developmental windows both separately and with mutual adjustment. Neurodevelopmental outcomes were identified using validated algorithms and included autism spectrum disorder (ASD), attention-deficit/hyperactivity disorder (ADHD), learning disabilities, speech and language disorders, coordination disorders, intellectual disabilities, and behavioral disorders. We applied a stratified Cox model accounting for individual and area-level confounders and examined effect measure modification by urbanicity, child’s race/ethnicity, and sex.

Findings:

The study found protective associations between green space exposure and most neurodevelopmental disorders. The strongest associations were seen for preconception exposure and intellectual disability (HR 0.66 [95 % CI: 0.48–0.95]), pregnancy exposure and ASD (HR 0.83 [95 % CI: 0.73–0.95]), and postnatal exposure for learning difficulties (HR 0.81 [95 % CI: 0.68–0.97]) per interquartile range (IQR = 0.12) increase in NDVI. The protective effects were stronger for Black/Hispanic children and for those living in urban areas.

Interpretation:

Green space exposure could benefit the children’s neurodevelopment, with more significant benefits for the Black and Hispanic populations.

Funding:

National Institute of Environmental Health Sciences R01-ES034038.

Keywords: Children’s neurodevelopment, Green space, Medicaid

1. Introduction

Neurodevelopmental disorders, such as autism spectrum disorder (ASD) or attention deficit hyperactivity disorder (ADHD), are a significant public health concern due to their increasing prevalence. (Centers for Disease Control and Prevention) They are characterized by impairments in social communication and interaction and delayed development of skills, such as speech, social, emotional, behavioral, and cognitive skills. (American Psychiatric Association, 2013) In the United States, the prevalence of neurodevelopmental disorders has increased dramatically over the past two decades, placing a growing burden on healthcare, education, and social support systems. (Danielson, et al., 2016) Recent CDC estimates indicate that approximately 1 in 36 children are diagnosed with ASD and about 1 in 10 with ADHD. (Maenner, 2023; QuickStats, 2024) Among Medicaid enrollees, the cumulative incidence of ASD and ADHD is estimated to be 1.6 % and 14.5 %, respectively, by age eight. (Straub, 2022) Neurodevelopmental disorders can be caused by a complex interplay of genetic, social, clinical, and environmental risk factors (e.g., air pollution, pesticides, and heavy metals). (Raz, 2015) However, the role of preconception, pregnancy, and postnatal exposure to green space in the risks of various neurodevelopmental disorders remains relatively unexplored.

Previous studies indicate that natural environments, such as green space, positively impact human health and children’s development, operating through various pathways, including stress relief, social interactions, and improved physical activity levels. (Thygesen, 2020; Luque-García, 2025; Luque-García, 2022) In addition, exposure to green space during pregnancy may improve physical and mental health in pregnant people, thereby mitigating the risks of neurodevelopmental disorders in their children. (Pagalan, 2022) Most studies on greenspace and neurodevelopmental disorders focus on prenatal or early childhood exposure. However, preconception exposure to greenspace may influence maternal mental and physiological health, which could in turn play a critical role in fetal development or potential epigenetic processes. (Kuiper, 2020; Willis, 2023).

Emerging research also highlights the contribution of environmental stressors—such as air pollution, pesticides, and toxic metals—to the etiology of neurodevelopmental disorders. These exposures, particularly during sensitive developmental windows, have been shown to negatively affect brain development. In this context, green space may serve as a protective factor by reducing exposure to harmful environmental conditions, improving air quality, and buffering physiological stress responses. However, due to limited sample sizes, most of the previous studies examined ASD or ADHD only, while few investigated other domains of mental health (e.g., emotional symptoms and conduct problems). (Jimenez, 2021) Moreover, a more protective effect from green space has been observed in the lower-SES population, while no difference was detected between racial/ethnic groups. (Choi et al., 2024) Given that environmental variables have been linked to neurodevelopment (Granés, 2024) and that green space may help counteract these effects, (Palmsten, 2013) identifying modifiable protective factors is critically important.

This study aimed to fill these gaps and explore the association between preconception, pregnancy, and early childhood exposure to green space and the risks of neurodevelopmental disorders in the United States, utilizing a population-based cohort of mother–child pair Medicaid enrollees. Furthermore, this study aims to investigate factors that may modify the association between perinatal green space and the risks of neurodevelopmental disorders.

2. Methods

2.1. Data sources and study population

We constructed a population-based cohort study utilizing nationwide Medicaid data from the 2001–2014 Medicaid Analytic Extract (MAX) database. (Palmsten, 2013) States provide Medicaid claims to the Centers for Medicare and Medicaid Services (CMS) through the Medicaid Statistical Information System (MSIS). (Medicaid Analytic eXtract (MAX) General Information) Medicaid covers the medical expenses of over 40 % of births in the US. (Markus, 2013) The population of pregnant women enrolled in Medicaid is characterized by younger age, racial and ethnic diversity, lower income levels, and includes individuals with disabilities. The study population included women aged 12 to 55 years and their liveborn children. We defined the cohort based on the last menstrual period date to approximate the conception date, where only children with birth dates between January 1st, 2001, and December 31st, 2014, were included. By accounting for the date of conception, we addressed the “fixed cohort bias” (Strand et al., 2011); by allowing for births of all gestational lengths to be included in the cohort. Exposure to green space was assigned at the zip code level. Children were followed from birth until their first continuous enrollment in Medicaid ended, developed a neurodevelopmental delay diagnosis, the study period ended, or death, whichever came first. Children with a known chromosomal or genetic abnormality were excluded. The study protocol was approved by Harvard TH Chan School of Public Health (Protocol Approval IRB23-1090), Mass General Brigham (Protocol Approval 2022P002615), and Rutgers School of Public Health (Protocol Approval 2024000897).

2.2. Green space

Green space exposure was quantified as the average annual (December-to-December) mean Normalized Difference Vegetation Index (NDVI) from Landsat 7 and Landsat 8. Greenspace was atmospherically corrected using the Landsat Surface Reflectance Code (LaSRC) for Landsat 8 and the LEDAPS algorithm for Landsat 7. (Vermote, 2016) The images were generated every 16 days at 30 m resolution, aggregating to each ZIP code and each year, and then linked to each individual’s ZIP code. NDVI is an indicator of vegetation health quantified by the ratio of near-infrared to visible light, which ranges between values of −1 to 1 (1 signals high green space, values close to 0 represent no green space, and −1 are generally water bodies). (DeFries and Townshend, 1994) Negative NDVI values were set to zero; based on the assumption that these areas lacked vegetation. NDVI values were then spatially averaged at the ZIP code level and summarized as annual means. We used annual instead of the seasonal mean NDVI values because we hypothesized that annual green space in an area might be more influential than the immediately available seasonal values. We calculated preconception (9 months), pregnancy, and postnatal (9 months) exposure using the annual NDVI values. The preconception and postnatal periods were restricted to 9 months for consistency with the pregnancy period. In case they spanned different years, we created a weighted average of the annual values for each year depending on the date of birth. For instance, if the delivery happened in March 2010, then the green space exposure during pregnancy was computed by giving weights of 6/10 (six months among ten months of pregnancy) for 2009 annual green space exposure and weights of 3/10 (three months among ten months of pregnancy) for 2010 annual green space exposure. Residential ZIP code information was available at two time points: the estimated date of the last menstrual period (LMP) and at delivery. To account for potential changes in residence during pregnancy, we assigned green space exposure assuming equal time spent at each location. Specifically, if the ZIP codes at LMP and delivery differed, we attributed 50 % of the pregnancy exposure duration to each ZIP code. This approach provided a balanced estimate of exposure across the prenatal period while acknowledging limitations in the absence of detailed residential history.

2.3. Neurodevelopmental disorders

The included neurodevelopmental disorders in the US cover ASD, ADHD, learning disability, developmental speech or language disorders, developmental coordination disorder, intellectual disability, and behavioral disorder. Validated claims-based algorithms with high positive predictive values (ranging from 82–98 %) were used to define each outcome (Table 1). (Straub, 2021) The algorithms incorporate data from the ICD-9 claims codes, age at diagnosis (with several sensitivity analyses using alternate cut-offs), and pharmacy dispensing records for relevant medications and have been used previously. (Suarez, 2024; Hernandez-Diaz, 2024) The event rates in this cohort are in accordance with the national US statistics. (Straub, 2022).

Table 1.

Definition of neurodevelopmental disorders of interest.

Outcome Algorithm
Autism Spectrum Disorder At ≥ 1 year of age: ≥ 2 dates with ICD-9 Dx 299.xx (except 299.1x)
Attention Deficit Hyperactivity Disorder/ Hyperkinetic Syndrome of Childhood At ≥ 2 years of age, any of the following:• ≥ 2 dates with ICD-9 Dx 314.xx• ≥ 2 dispensings of atomoxetine, clonidine, guanfacine, (dextro/lisdex) amphetamine, (dex) methylphenidate• ≥ 1 Dx 314.xx & ≥ 1 dispensing
Learning Difficulty At ≥ 2 years of age: ≥ 1 ICD-9 Dx 315.0x, 315.1, 315.2
Developmental Speech or Language Disorder At ≥ 1.5 years of age: ≥ 2 dates with ICD-9 Dx 315.3x (except 315.34)
Intellectual disability At ≥ 2 years of age: ≥ 2 dates with ICD-9 Dx 317, 318.x, 319
Developmental Coordination Disorder ≥ 2 dates with ICD-9 Dx 315.4 (any age)
Behavioral Disorder At ≥ 2 years of age: ≥ 2 dates with ICD-9 Dx 312.xx, 313.xx
Open in a new tab

2.4. Meteorological variables

We have acquired the daily 4 km × 4 km gridded data set of surface temperature and relative humidity variables across the contiguous United States (gridMET). (Abatzoglou, 2013) Daily maximum temperature averages were calculated for each ZIP code using an area-weighted approach, where each grid cell’s contribution was weighted by the proportion of the ZIP code’s land area it covered. Temperature and humidity were included in the main analysis as potential confounders, given that greener areas are typically cooler and more humid, (Li, 2024) which may independently influence neurodevelopmental outcomes.

2.5. Covariates

A broad range of covariates was considered based on a directed acyclic graph and prior literature. At the individual level, we included maternal age, maternal race/ethnicity, substance abuse, smoking, alcohol abuse, poor nutrition, birth year, and the season of the delivery (spring (March to May), summer (June to August), fall (September to November), and winter (December to February)) (e-Methods). Individual covariates for substance abuse, smoking, alcohol abuse, and poor nutrition were coded as a binary variable (Yes or No). Substance abuse was coded as yes if they had used opioids, cannabis, sedatives, cocaine, other stimulants, hallucinogens, nicotine, inhalant abuse, and other psychoactive drugs. Smoking was coded as yes if they did tobacco smoking, chewing, or nicotine addiction. Alcohol abuse was coded as yes if they had alcohol abuse or alcohol-related diseases (e.g., alcoholic fatty liver, alcohol-induced mental disorders). Poor nutrition was coded as yes if they were deficient in nutritional elements (e.g., vitamins, zinc, calcium, selenium).

We used ZIP code level measures of SES (median household income and percent of people who did not complete high school) and population density from the 2000 to 2014 American Community Survey (ACS) and the 2000 and 2010 US Census. We also included the county average Body Mass Index (BMI) from the Behavioral Risk Factor Surveillance System. (Centers for Disease Control and Prevention) We calculated the distance to the nearest hospital by matching the distance from the centroid of each ZIP Code Tabulation Area (ZCTA) to the nearest hospital every year for each individual’s residential ZIP code level from the Dartmouth Atlas of Health Care, serving as a proxy for access to healthcare. (Kind and Buckingham, 2018) We also adjusted for the 5 US regions, divided into 5 categories: Midwest, Northeast, Southeast, Southwest, and West. Even though individual-level variables on maternal comorbidities (such as gestational diabetes, hypertensive disorders in pregnancy, and others) were available, we decided not to adjust for them because they may be causal intermediates. Fig. S1 shows the directed acyclic graphs (DAGs) for the selected covariates.

2.6. Statistical analysis

Hazard ratios (HRs) with 95 % CIs were estimated through a stratified Cox proportional hazard model. We allowed the baseline hazard to vary by birth year and county. We accounted for all covariates described in the previous section, considering the ZIP code as the cluster index. Green space exposures were analyzed for different exposure windows—preconception, pregnancy, and postnatal—initially modeled separately to assess potential sensitive periods. To further explore the independent contribution of each exposure window, we additionally fitted models that included all exposure windows simultaneously. This allowed us to assess the robustness of associations while accounting for potential correlation between windows. To assess potential multicollinearity among green space exposure variables across time windows, we also calculated variance inflation factor (VIF) values (O’brien, 2007) for NDVI at preconception, pregnancy, and postnatal periods. We examined the associations in the full cohort and the subset of urban ZIP codes, defined as ZIP codes with population density ≥ 1000 people/square mile.

The shape of the exposure–response curves for greenness was examined using natural splines with 2 to 3 degrees of freedom. Hazard ratios (HRs) were expressed per IQR increase in exposure. Effect modification by children’s sex, race/ethnicity, urbanicity and median household income as a proxy for SES was examined by stratification. Urbanicity was defined using Rural–Urban Commuting Area (RUCA) codes, which classify ZIP codes into 10 levels based on population density, urbanization, and commuting patterns. We grouped these into three categories: Urban Areas (≥50,000 people), Urban Clusters (2,500–49,999 people), and Rural Areas (<2,500 people) based on U.S. Census definitions. (U.S. Census Bureau) Median household income was divided into tertiles based on its distribution in the study population. We also repeated the effect modification by sex and race, restricting to the urban zip codes. We selected these variables because previous research has shown different effects of environmental exposures on the risks of neurodevelopmental disorders within their strata level. (Durkin and Yeargin-Allsopp, 2018; Kerin, 2018).

2.7. Sensitivity analyses

We implemented a series of sensitivity analyses to examine the robustness of our findings. We examined tree canopy cover as an alternative measure of green space instead of NDVI. The annual mean of tree canopy cover with a 30 m resolution was obtained from the National Land Cover Database. (Homer, 2004) Tree canopy was defined as the percentage of the land covered by tree canopy for each ZIP code level derived from multi-spectral satellite imagery. The tree canopy captures only trees; focusing on the height and coverage of the trees, whereas NDVI captures all green vegetation, including trees, grass, and shrubs. (Stojanova, 2010) We also restricted the analysis to children diagnosed at 3 or older and 8 or older to address potential outcome misclassification due to preliminary diagnosis. We repeated the analysis using inverse probability of censoring weights to account for the potential of selection bias due to informative censoring. (Hernán, 2006).

We used SAS version 9.3 (SAS Institute Inc., Cary, NC) for data extraction, and R version 4.2.0 (https://www.r-project.org/foundation/) for Linux-gnu for analyses. Our interpretation of the results focused on the consistency of estimates across the main and sensitivity analyses, the strength of the adjusted HR, and its precision as reflected in the width of the 95 % CI, instead of statistical significance at the p < 0.05 level.

3. Results

3.1. Population characteristics

Fig. 1 provides an overview of the flow chart, illustrating the selection process for the total study population. After excluding pregnancies with missing covariates, our analytic sample included 1,841,915 mother–child pairs. Table 2 shows the study population characteristics in total and stratified by NDVI levels dichotomized using the median value (0.27). All green space exposure metrics were highly correlated across time windows, with Spearman correlation coefficients for NDVI between different exposure periods ranging from 0.95 to 0.98 (Table S1). To assess potential collinearity, we calculated pairwise Spearman correlations for all key exposure and covariate variables, presented in Supplementary Table S2. Participants with higher green space exposure were generally more likely to be White, smoke, have more maternal mental health-associated diagnoses, and reside in less populated and more deprived areas. The residual plots demonstrating linearity of the association between NDVI and ASD/ADHD are shown in Fig. S2. The characteristics of the study population stratified by ASD, ADHD status (as the most commonly diagnosed), are shown in Table S3. The crude cumulative incidence curves of any neurodevelopmental disorder stratified by NDVI exposure dichotomized at the median (0.27) show higher incidence in higher greenness areas in the full cohort (Fig. 2). This finding is in accordance with the US national statistics. (Zablotsky and Black, 2020).

Fig. 1.

Fig. 1.

Open in a new tab

Flowchart of the study population (Medicaid Analytic eXtract).

Table 2.

Cohort characteristics by the median value of annual mean normalized difference vegetation (NDVI) index. Median average annual mean NDVI was 0.27 (IQR: 0.12). Abbreviations: NDVI, normalized difference vegetation index; IQR, interquartile range; follow-up years after birth.

Median annual NDVI levels (Median = 0.27)
Total N (%) Below median Above or equal to the median
Population
Person [n] 1,841,915 897,066 944,849
Follow-up years (years) [median (IQR)] 4.6 (3.7) 4.6 (3.7) 4.3 (3.6)
Autism Spectrum Disorder [n (%)] 16,517 (0.9) 9,029 (1.0) 7,488 (0.8)
Attention Deficit Hyperactivity Disorder/ Hyperkinetic Syndrome of Childhood [n (%)] 81,183 (4.4) 28,608 (3.2) 52,575 (5.6)
Learning Difficulty [n (%)] 9,717 (0.5) 4,883 (0.5) 4,834 (0.5)
Developmental Speech or Language Disorder [n (%)] 90,598 (4.9) 41,589 (4.6) 49,009 (5.2)
Intellectual disability [n (%)] 14,033 (0.8) 8,656 (1.0) 5,377 (0.6)
Developmental Coordination Disorder [n (%)] 4,871 (0.3) 2,285 (0.3) 2,586 (0.3)
Behavioral Disorder [n (%)] 56,620 (3.1) 23,827 (2.7) 32,793 (3.5)
Any Specific Neurodevelopmental Disorder [n (%)] 184,027 (10.0) 80,750 (0.9) 103,277 (10.9)
Individual covariates
Sex of children (Male) [n (%)] 930,600 (50.5) 462,366 (50.2) 468,234 (50.6)
Race or ethnic group [n (%)]
White 757,633 (41.1) 239,914 (27) 517,719 (54.8)
Black/African-American 603,314 (32.8) 274,230 (30.6) 329,084 (34.8)
Asian/Other Pacific 61,153 (3.3) 50,641 (5.6) 10,512 (1.1)
Hispanic/Latino 358,435 (19.5) 287,788 (32) 70,647 (7.5)
Unknown 61,380 (3.3) 44,493 (4.9) 16,887 (1.8)
Maternal age at delivery (years) [mean (sd)] 24.7 (5.9) 25.4 (6.1) 24.0 (5.6)
Substance abuse [n (%)] 27,197 (1.5) 10,888 (1.2) 16,309 (1.7)
Smoking [n (%)] 39,545 (2.1) 11,510 (1.3) 28,035 (2.9)
Alcohol abuse [n (%)] 9,660 (0.5) 4,397 (0.5) 5,263 (0.6)
Poor nutrition [n (%)] 9,084 (0.5) 4,203 (0.5) 4,881 (0.5)
Maternal mental health [n (%)] 220,765 (12) 83,201 (9.3) 137,564 (14.7)
BMI [mean (sd)] 27.8 (1.4) 27.5 (1.2) 28 (1.5)
Ecological variables
Median household income ($1,000) [mean (sd)] 42.5 (15) 42.3 (15.5) 42.7 (15.3)
Did not complete high school [% (sd)] 35 (16) 38.5 (16.6) 31.8 (13.8)
ADI [% (sd)] 56 (25) 46.8 (26.4) 65.5 (20.2)
Population Density [mean, (sd)] 7,473 (15,503) 13,740 (20,293) 1,520 (2,186)
Nearest hospital (km) [mean (sd)] 5.5 (6.8) 4.2 (6.1) 6.7 (7.1)
Meteorologic variables
Average temperature (°C) [mean (IQR)] 13.6 (5.8) 13.8 (4.0) 13.3 (4.0)
Open in a new tab

Fig. 2.

Fig. 2.

Open in a new tab

Cumulative Incidence of Any neurodevelopmental disorder in Children, by NDVI dichotomized at the median value (0.27).

3.2. Hazard ratios

VIF values for NDVI exposure variables were 2.48 (preconception), 1.85 (pregnancy), and 2.22 (postnatal), suggesting limited collinearity among exposure windows. Fig. 3 shows the results without (A) and with (B) mutual adjustment for multiple windows of exposure. In the adjusted stratified Cox models for the full cohort, exposure to green space was generally protective for the development of neurodevelopmental disorders. Preconception greenness exposure was associated with a reduced risks of ASD (HR 0.93, [0.85–1.02]), ADHD (HR 0.94, [0.90–0.99]), learning difficulty (HR 0.89, [0.81–0.98]), intellectual disability (HR 0.84, [0.74–0.96]) and behavioral disorder (HR 0.89, [0.85–0.94]) per IQR increase in NDVI. Pregnancy exposure was associated with a reduced risks of ASD (HR 0.90, [0.83–0.98]), learning difficulties (HR 0.93, [0.85–1.02]), intellectual disability (HR 0.91, [0.80–1.04]), behavioral disorders (HR 0.92, [0.88–0.96]) and slightly protective for ADHD (HR 0.97, [0.92–1.01] per IQR increase in NDVI. Postnatal exposure is associated with a reduced risks of ASD (HR 0.94, [0.85–1.03]) ADHD (HR 0.94, [0.89–0.99]), learning difficulty (HR 0.85, [0.77–0.95]), intellectual disability (HR 0.91, [0.80–1.04]) and behavioral disorder (HR 0.91, [0.88–0.95]) per IQR increase in NDVI. Mutual adjustment for NDVI across preconception, pregnancy, and postnatal periods introduced notable shifts in some associations. For instance, the inverse association between postnatal NDVI and behavioral disorders observed in the unadjusted model (HR = 0.89, 95 % CI: 0.85–0.94) was attenuated after mutual adjustment (HR = 1.01, 95 % CI: 0.92–1.11). Some associations, such as ASD during pregnancy and intellectual disability in the preconception period, became more pronounced. When we performed the analysis in the urban ZIP codes, the effects were similar to the full cohort (Fig. S3).

Fig. 3.

Fig. 3.

Open in a new tab

The hazard ratio of neurodevelopmental diseases per IQR increases in green space in the full cohort. NDVI IQR: 0.12. (A) Adjusted for covariates (B) Mutually adjusted for the other two exposure periods, as well as all other covariates listed. Note: The analysis was adjusted for covariates (Race, age, temperature, population density, season of delivery, median household income, percent of people who did not complete high school, strata(FIPS), region, distance to the nearest hospital, substance abuse, smoking, alcohol abuse, BMI, and poor nutrition, cluster = ZIP_index). Green space was defined as exposure to NDVI; autism spectrum disorder (ASD), attention deficit hyperactivity disorder (ADHD), learning disability (LD), developmental speech or language disorders (DSLD), developmental coordination disorder (DCD), intellectual disability (ID), and behavioral disorder (BD).

3.3. Effect modification

The results for the stratified analysis by sex, race, and urbanicity, examining effect modification, are shown in Table 3. Regarding pregnancy exposures, NDVI was slightly protective in males but not in females for learning disability and developmental speech or language disorders, which is consistent with prior literature showing higher baseline vulnerability among boys for these disorders. (Adani and Cepanec, 2019) NDVI also had a stronger protective effect for ASD and learning disability for Black/Hispanic children compared to Whites. This effect was also protective in Black/Asian children with intellectual disability compared to Whites/Hispanics. In urbanized areas (Urban Area and Urban Cluster), NDVI had a protective effect for most neurodevelopmental delays, while this was not shown in rural regions. There was no clear pattern of effect modification by median income. These effects were similar in the other exposure windows (Table S4 and Table S5). When we restricted the race and sex effect modification analysis to urban ZIP codes, the results were the same (Table S6).

Table 3.

Analysis of effect modification between pregnancy green space-neurodevelopmental disorders by urbanicity, sex, race/ethnicity and median household income as a proxy of SES.

Urbanicity Sex Race/Ethnicity Median annual household income
Urban Area (N = 1,487,098) Urban Cluster (N = 292,866) Rural (N = 61,950) Male (N = 938,298) Female (N = 903,618) White (N = 759,065) Black/African American (N = 605,291) Asian/ Other Pacific (N = 61,893) Hispanic/ Latino (N = 352,504) Others (N = 63,164) Low Income (≤ $34,700) (N=613,970) Medium Income ($34,701-$45,562) (N=613,933) High income (≥$45,562) (N=614,012)
ASD 0.90 (0.81, 0.99)* 0.89 (0.70, 1.13) 1.05 (0.60, 1.82) 0.90 (0.81,0.99)* 0.90 (0.77, 1.04) 0.95 (0.84, 1.08) 0.79 (0.70, 0.89)** 1.19 (0.89, 1.58) 0.89 (0.80, 1.00) 0.76 (0.56, 1.02) 0.84 (0.73, 0.97)** 1.01 (0.81, 1.26) 0.90 (0.80, 1.01)
ADHD 1.04 (0.98, 1.09) 0.86 (0.76, 0.97)** 1.04 (0.80, 1.35) 1.04 (0.98, 1.09) 1.00 (0.93, 1.06) 1.02 (0.96, 1.08) 1.00 (0.95, 1.06) 0.83 (0.62, 1.11) 1.03 (0.95, 1.11) 0.96 (0.78, 1.19) 1.10 (1.03, 1.17) 1.02 (0.94, 1.11) 0.90 (0.84, 0.97)*
LD 0.96 (0.87, 1.07) 0.99 (0.75, 1.30) 1.07 (0.61, 1.88) 0.93 (0.83, 1.04) 1.02 (0.89, 1.18) 1.04 (0.88, 1.23) 0.93 (0.81, 1.07) 1.30 (0.80, 2.09) 0.85 (0.73, 1.00) 0.88 (0.55, 1.40) 0.99 (0.86, 1.31) 1.14 (0.95, 1.35) 0.90 (0.78, 1.05)
DSLD 0.99 (0.95, 1.02) 0.90 (0.80, 1.01) 1.15 (0.90, 1.48) 0.98 (0.94, 1.02) 1.01 (0.96, 1.07) 0.99 (0.94, 1.05) 0.98 (0.93, 1.03) 0.99 (0.83, 1.16) 1.00 (0.94, 1.07) 0.86 (0.74, 0.99)* 0.97 (0.90, 1.03) 1.01 (0.94, 1.09) 0.97 (0.92, 1.04)
DCD 1.12 (0.98, 1.29) 0.99 (0.73, 1.33) 0.91 (0.48, 1.72) 1.11 (0.96, 1.30) 1.05 (0.90, 1.23) 1.13 (0.89, 1.43) 1.10 (0.94, 1.28) 1.19 (0.85, 1.66) 0.97 (0.86, 1.09) 0.96 (0.72, 1.28) 0.86 (0.73, 1.01) 1.38 (0.97, 1.97) 1.09 (0.91, 1.30)
ID 0.92 (0.80, 1.05) 1.47 (0.98, 2.21) 1.32 (0.58, 2.98) 0.98 (0.85, 1.14) 0.96 (0.76, 1.21) 1.17 (0.95, 1.43) 0.79 (0.65, 0.98)* 0.60 (0.29, 1.27) 1.03 (0.80, 1.32) 0.65 (0.27, 1.60) 1.11 (0.92, 1.35) 0.94 (0.74, 1.18) 0.73 (0.57, 0.92)*
BD 0.97 (0.92, 1.02) 0.90 (0.79, 1.02) 0.71 (0.53, 0.95)* 0.95 (0.90, 1.01) 0.97 (0.91, 1.04) 0.96 (0.90, 1.03) 0.97 (0.91, 1.04) 0.87 (0.61, 1.23) 0.93 (0.85, 1.01) 0.88 (0.70, 1.10) 1.04 (0.97, 1.11) 0.94 (0.85, 1.03) 0.88 (0.81, 0.96)*
Open in a new tab

Note: green space defined as pregnancy exposure of NDVI; autism spectrum disorder (ASD), attention deficit hyperactivity disorder (ADHD), learning disability (LD), developmental speech or language disorders (DSLD), developmental coordination disorder (DCD), intellectual disability (ID), and behavioral disorder (BD). Association as hazard ratio per 0.12 NDVI IQR increase.

***

indicating less than 0.001p-value,

**

indicating less than 0.01 p-value, and

*

indicating less than 0.05 p-value.

3.4. Sensitivity analyses

When using tree canopy, the results did not meaningfully change (Table S7). Restricting children diagnosed after the age of 3 years and 8 years, the protective associations of pregnancy NDVI exposure were similar (Table S8). When using inverse probability weighting to account for informative censoring, the associations were similar to the main results (Table S9).

4. Discussion

To our knowledge, this is the first study to examine the association of preconception, pregnancy, and postnatal green space exposure and the risks of an array of neurodevelopmental delays in a national US population-based cohort of Medicaid enrollees with more than 1.8 million mother–child pairs. We found protective associations between preconception, pregnancy, and postnatal green space exposure and the risks of several neurodevelopmental delays. These protective effects were stronger in Black/African American and Hispanic children compared to Whites for ASD and learning disabilities. The ASD association’s specificity to the prenatal period aligns with multiple lines of evidence suggesting a prenatal origin of ASD, including findings of altered brain cytoarchitecture in children with ASD. (McFadden and Minshew, 2013; Stoner, 2014) Our results were consistent across several sensitivity analyses using tree canopy as a green space indicator. In the model mutually adjusted for all three exposure periods, the results remained robust for some associations, while others were attenuated. The small changes on the HRs with mutual adjustment did not appear to be an artifact of collinearity because the precision of the mutually adjusted model was not substantially lower than the single-exposure model. These changes may reflect some shared variance across NDVI windows, where mutually adjusting for temporally correlated exposures helps isolate window-specific effects but can also reduce power. Our VIF analysis suggested low multicollinearity (all VIFs < 2.5), indicating that these shifts are unlikely to result from unstable model estimates. Instead, they may reflect biologically or behaviorally relevant differences in timing, which warrant further exploration in studies with finer temporal resolution and replication datasets. The 95 % CIs were not notably larger in this analysis, suggesting that collinearity did not diminish the stability of the models.

Our results are in alignment with the results from previous studies. A study in Canada found protective effects between pregnancy NDVI and autism with an odds ratio (OR) of 0.96 (95 % CI: 0.90, 1.02) per 0.12 NDVI IQR increase (Wu and Jackson, 2017); which were consistent with our study results. Research in China has found that higher NDVI exposure before birth is associated with lower risks of ASD (24), and high levels of residential NDVI in early childhood were associated with a protective effect for ADHD in Denmark. (Thygesen, 2020) Other studies mainly focused on how childhood green space (tree canopy, grass, and forest) exposure affects children’s neurodevelopmental delays. (Wu and Jackson, 2017).

The stronger protective effects of green space observed among Black and Hispanic children may reflect greater relative benefits in communities with lower baseline access to greenness, higher exposure to environmental and social stressors, and limited availability of other health-promoting resources. In such contexts, green space may act as a critical buffer against adversity, enhancing resilience and promoting child development. These study findings highlight the importance of green space access in more disadvantaged populations. Interestingly, the protective association between NDVI and certain neurodevelopmental outcomes—specifically learning disabilities and developmental speech or language disorders—was more evident among males than females. This may reflect heightened biological sensitivity to environmental exposures during early brain development in males, a phenomenon supported by prior studies on sex-specific neurodevelopmental vulnerability.

We observed the protective effects of exposure to residential green space in various windows of exposure for different neurodevelopmental delays and were robust using different analytical strategies. The mechanisms through which exposure to green space is associated with neurodevelopmental delays may not be uniform and may include gamete epigenetic changes (for the preconception period) or effects in the in-utero and early life brain development for exposures after conception. (Jacob and Moley, 2005) Recent research has shown that epigenetic changes in sperm are associated with childhood quantitative autistic traits, highlighting both the paternal role and the preconception exposures’ contribution to risks. (Feinberg, 2024) Also, green space is associated with higher semen quality, and epigenetic inheritance is associated with environmental factors. Green space exposure also has general health benefits, such as lowering stress and depression, increasing social engagement, and exposure to noise and extreme temperatures. (Jarvis, 2021) Because in our dataset, areas with higher NDVI were more deprived, we also repeated the analysis in urban areas to disentangle the complex interplay of NDVI, area deprivation, and risks of neurodevelopmental delays. These results confirmed the protective associations of greenness in neurodevelopmental disorder in urban areas. Also, exposure to green space early in childhood affects the children’s behavioral and mental health, resulting in improved cognitive skills, early development, enhanced interactions with other children, and reduced attention-related problems. (Jarvis, 2021).

Our study has certain strengths. The Medicaid population includes young, racially diverse, low-income, and people with disabilities; this type of population is typically underrepresented in birth cohort studies examining environmental risk factors for adverse pregnancy and childhood outcomes. Also, this study consists of a large sample size that enables the evaluation of the association between NDVI and various neurodevelopmental delays, for some of which data is lacking. Our analysis adjusted for a broad range of potential confounding variables at the individual and area levels, and also our findings were robust to our sensitivity analyses. Finally, since our results were robust when we applied inverse probability weights, our results are unlikely to be prone to informative censoring.

There are a few limitations to this study. First, the green space exposure in the form of NDVI was measured using the satellite-derived dataset, which might not fully capture the true green space values and does not capture the specificity of the different vegetation types. While these metrics do not explicitly distinguish between specific land use types (e.g., public parks vs. croplands), prior studies have demonstrated that NDVI is a good proxy of urban greenspace exposure and strong correlations with more detailed green space typologies, (Ju, 2024) while tree canopy provides a complementary metric of vertical green structure relevant for shading and cooling. Also, the green space was at an annual ZIP code level instead of a more granular geolocation or temporal resolution. Prior research on fertility outcomes has demonstrated that the annual and seasonal metrics yield similar estimates, suggesting a more chronic effect. (Willis, 2023) While ZIP code–level exposures are more distal than personal-level measurements and inevitably introduce non-differential misclassification, they reduce susceptibility to biases from individual-level factors that are difficult to measure or control in large-scale studies. (Weisskopf and Webster, 2017) This approach helps mitigate risks of residual confounding and reverse causation, thereby strengthening the validity of our findings. Also, recent evidence suggests that individual-level factors are not strong confounders in the health effects of area-level ambient exposures, with personal and area exposure levels providing similar results. (Klompmaker, 2024) We also did not have data on time-activity patterns to capture the individual-level green space exposure, so unmeasured confounding cannot be entirely ruled out. In this study, NDVI values were assigned based on the mean greenness within ZIP code boundaries using 30-meter resolution Landsat imagery. While we did not apply an additional buffer beyond the ZIP code boundary, the use of high-resolution satellite data partially mitigates misclassification by capturing adjacent vegetated pixels near boundary areas. We acknowledge, however, that individuals living near ZIP code edges may experience greenness beyond the immediate administrative area, particularly in smaller ZIP codes. As a result, potential exposure misclassification may occur for individuals residing near boundary areas, particularly in smaller ZIP codes. Future studies employing individual-level geocoded addresses or circular buffers, as well as evaluating various green space indices (e.g., parks or grass) and higher-resolution measures like street imagery views, may better account for fine-scale spatial variation in green space exposure. Since we included only live births in our analysis, there is a potential for live birth bias. However, residential green space has not been associated with miscarriages, while its effect on fecundability is modestly beneficial. (Willis, 2023) Although mutual adjustment for exposure windows was included in our models and variance inflation factor (VIF) values remained below commonly accepted thresholds (generally < 2.5), indicating limited multicollinearity, we acknowledge that distinguishing the timing-specific effects of green space remains challenging. Finally, our study population includes women and offspring covered by Medicaid, so our results may not be generalizable to people covered by private insurance.

In this geographically diverse mother–child pair cohort study, we observed that higher exposure to residential green space during the preconception, pregnancy, and postnatal period was associated with a lower risk of neurodevelopmental delays in children. If the relationships we observe are causal, area-level investments in increasing accessible green spaces may reduce the population burden of neurodevelopmental delays.

Supplementary Material

1

Appendix A. Supplementary material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.envint.2025.109666.

Footnotes

CRediT authorship contribution statement

Hayon Michelle Choi: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Resources, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Krista F. Huybrechts: Writing – review & editing, Writing – original draft, Resources, Data curation. Sonia Hernandez-Diaz: Writing – review & editing, Writing – original draft, Formal analysis, Data curation, Conceptualization. Xinye Qiu: Writing – review & editing, Writing – original draft, Data curation, Conceptualization. Michael Leung: Writing – review & editing, Writing – original draft, Data curation, Conceptualization. Peter James: Writing – review & editing, Writing – original draft, Data curation. Matthew Shupler: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization. Wanyu Huang: Writing – review & editing, Writing – original draft, Formal analysis, Data curation, Conceptualization. Yaguang Wei: Writing – review & editing, Writing – original draft, Data curation. Antonella Zanobetti: Writing – review & editing, Writing – original draft, Methodology, Funding acquisition, Data curation, Conceptualization. Christopher J McDougle: Writing – review & editing, Writing – original draft. Joel Schwartz: Writing – review & editing, Writing – original draft, Methodology, Data curation. Brent Coull: Writing – review & editing, Writing – original draft, Methodology. Marc Weisskopf: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization. Stefania Papatheodorou: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

The authors do not have permission to share data.

References

  1. Abatzoglou JT, 2013. Development of gridded surface meteorological data for ecological applications and modelling. Int. J. Climatol 33 (1), 121–131. [Google Scholar]
  2. Adani S, Cepanec M, 2019. Sex differences in early communication development: behavioral and neurobiological indicators of more vulnerable communication system development in boys. Croat. Med. J 60 (2), 141–149. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. American Psychiatric Association, D. and A.P. Association, Diagnostic and statistical manual of mental disorders: DSM-5. Vol. 5. 2013: American psychiatric association Washington, DC. [Google Scholar]
  4. Centers for Disease Control and Prevention, no date. Behavioral Risk Factor Surveillance System: BRFSS Prevalence & Trends Data. U.S. Department of Health and Human Services. Available at: https://www.cdc.gov/brfss/brfssprevalence/ (Accessed: 5 July 2025). [Google Scholar]
  5. Centers for Disease Control and Prevention, n.d. Autism Spectrum Disorder (ASD): Data & Statistics. Available at: https://www.cdc.gov/ncbddd/autism/data.html (Accessed: 3/6/2025).
  6. Choi HM, Heo S, Bell ML, 2024. The effect modification of greenspace and impervious surface on the heat-mortality association: differences by the dissimilarity index. Sci. Total Environ 908, 168074. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Danielson ML, et al. , 2018. Prevalence of parent-reported ADHD diagnosis and associated treatment among U.S. children and adolescents, 2016. J. Clin. Child Adolescent Psychol 47(2), 199–212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. DeFries RS, Townshend J, 1994. NDVI-derived land cover classifications at a global scale. Int. J. Remote Sens 15 (17), 3567–3586. [Google Scholar]
  9. Durkin MS, Yeargin-Allsopp M, 2018. Socioeconomic status and pediatric neurologic disorders: current evidence. Semin. Pediatr. Neurol 27, 16–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Feinberg JI, et al. , 2024. Epigenetic changes in sperm are associated with paternal and child quantitative autistic traits in an autism-enriched cohort. Mol. Psychiatry 29 (1), 43–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Granés L, et al. , 2024. Early life cold and heat exposure impacts white matter development in children. Nat. Clim. Chang 14 (7), 760–766. [Google Scholar]
  12. Hernán MA, et al. , 2006. Comparison of dynamic treatment regimes via inverse probability weighting. Basic Clin. Pharmacol. Toxicol 98 (3), 237–242. [DOI] [PubMed] [Google Scholar]
  13. Hernandez-Diaz S, et al. , 2024. Risk of autism after prenatal topiramate, valproate, or lamotrigine exposure. N. Engl. J. Med 390 (12), 1069–1079. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Homer C, et al. , 2004. Development of a 2001 national land-cover database for the United States. Photogramm. Eng. Remote Sens 70 (7), 829–840. [Google Scholar]
  15. Jacob S, Moley KH, 2005. Gametes and embryo epigenetic reprogramming affect developmental outcome: implication for assisted reproductive technologies. Pediatr. Res 58 (3), 437–446. [DOI] [PubMed] [Google Scholar]
  16. Jarvis I, et al. , 2021. Assessing the association between lifetime exposure to greenspace and early childhood development and the mediation effects of air pollution and noise in Canada: a population-based birth cohort study. Lancet Planet. Health 5 (10), e709–e717. [DOI] [PubMed] [Google Scholar]
  17. Jimenez MP, et al. , 2021. Early life exposure to greenness and executive function and behavior: an application of inverse probability weighting of marginal structural models. Environ. Pollut 291, 118208. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Ju Y, et al. , 2024. Assessing normalized difference vegetation index as a proxy of urban greenspace exposure. Urban For. Urban Green 99, 128454. [Google Scholar]
  19. Kerin T, et al. , 2018. Association between air pollution exposure, cognitive and adaptive function, and ASD severity among children with autism spectrum disorder. J. Autism Dev. Disord 48 (1), 137–150. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kind AJH, Buckingham WR, 2018. Making neighborhood-disadvantage metrics accessible — the neighborhood atlas. N. Engl. J. Med 378 (26), 2456–2458. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Klompmaker JO, et al. , 2024. Associations of fine particulate matter with incident cardiovascular disease; comparing models using ZIP code-level and individual-level fine particulate matter and confounders. Sci. Total Environ 926, 171866. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kuiper IN, et al. , 2020. Associations of preconception exposure to air pollution and greenness with offspring asthma and hay fever. Int. J. Environ. Res. Public Health 17 (16). [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Li Y, et al. , 2024. Green spaces provide substantial but unequal urban cooling globally. Nat. Commun 15 (1), 7108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Luque-García L, et al. , 2022. Does exposure to greenness improve children’s neuropsychological development and mental health? a navigation guide systematic review of observational evidence for associations. Environ. Res 206, 112599. [DOI] [PubMed] [Google Scholar]
  25. Luque-García L, et al. , 2025. Exploring the pathways linking prenatal and early childhood greenness exposure to attention-deficit/hyperactivity disorder symptoms during childhood: an approach based on robust causal inference. Int. J. Hyg. Environ. Health 263, 114475. [DOI] [PubMed] [Google Scholar]
  26. Maenner MJ, et al. , 2023. Prevalence and characteristics of autism spectrum disorder among children aged 8 years - autism and developmental disabilities monitoring network, 11 sites, United States, 2020. MMWR Surveill. Summ 72 (2), 1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Markus AR, et al. , 2013. Medicaid covered births, 2008 through 2010, in the context of the implementation of health reform. Womens Health Issues 23 (5), e273–e280. [DOI] [PubMed] [Google Scholar]
  28. McFadden K, Minshew NJ, 2013. Evidence for dysregulation of axonal growth and guidance in the etiology of ASD. Front. Hum. Neurosci 7, 671. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Medicaid Analytic eXtract (MAX) General Information. Available at: https://www.medicaid.gov/medicaid/data-systems/macbis/medicaid-chip-research-files/medicaid-analytic-extract-max-general-information (Accessed: 2/15/2025).
  30. O’brien RM, 2007. A caution regarding rules of thumb for variance inflation factors. Qual. Quant 41 (5), 673–690. [Google Scholar]
  31. Pagalan L, et al. , 2022. The association between prenatal greenspace exposure and Autism spectrum disorder, and the potentially mediating role of air pollution reduction: a population-based birth cohort study. Environ. Int 167, 107445. [DOI] [PubMed] [Google Scholar]
  32. Palmsten K, et al. , 2013. Harnessing the medicaid analytic eXtract (MAX) to evaluate medications in pregnancy: design considerations. PLoS One 8 (6), e67405. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. QuickStats: Percentage* of Children and Adolescents Aged 5–17 Years Who Had Ever Received a Diagnosis of Attention-Deficit/Hyperactivity Disorder,(†) by Urbanization Level(§) and Age Group - National Health Interview Survey, United States, 2020–2022(¶). MMWR Morb Mortal Wkly Rep., 2024. 73(5), 116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Raz R, et al. , 2015. Autism spectrum disorder and particulate matter air pollution before, during, and after pregnancy: a nested case–control analysis within the Nurses’ Health Study II cohort. Environ. Health Perspect 123 (3), 264–270. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Stojanova D, et al. , 2010. Estimating vegetation height and canopy cover from remotely sensed data with machine learning. Eco. Inform 5 (4), 256–266. [Google Scholar]
  36. Stoner R, et al. , 2014. Patches of disorganization in the neocortex of children with autism. N. Engl. J. Med 370 (13), 1209–1219. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Strand LB, Barnett AG, Tong S, 2011. Methodological challenges when estimating the effects of season and seasonal exposures on birth outcomes. BMC Med. Res. Methodol 11, 49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Straub L, et al. , 2021. Validity of claims-based algorithms to identify neurodevelopmental disorders in children. Pharmacoepidemiol. Drug Saf 30 (12), 1635–1642. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Straub L, et al. , 2022. Neurodevelopmental Disorders among publicly or privately Insured Children in the United States. JAMA Psychiat. 79 (3), 232–242. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Suarez EA, et al. , 2024. Prescription stimulant use during pregnancy and risk of neurodevelopmental disorders in children. JAMA Psychiat. 81 (5), 477–488. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Thygesen M, et al. , 2020. The association between residential green space in childhood and development of attention deficit hyperactivity disorder: a population-based cohort study. Environ. Health Perspect 128 (12), 127011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. U.S. Census Bureau, no date. Urban and rural. U.S. Department of Commerce. Available at: https://www.census.gov/programs-surveys/geography/guidance/geo-areas/urban-rural.html (Accessed: 5 July 2025). Bureau, U.S.C., Urban and Rural Definition. [Google Scholar]
  43. Vermote E, et al. , 2016. Preliminary analysis of the performance of the Landsat 8/OLI land surface reflectance product. Remote Sens. Environ 185, 46–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Weisskopf MG, Webster TF, 2017. Trade-offs of Personal Versus more Proxy Exposure measures in Environmental Epidemiology. Epidemiology 28 (5), 635–643. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Willis MD, et al. , 2023. Associations between residential greenspace and fecundability in a North American preconception cohort study. Environ. Health Perspect 131 (4), 047012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Wu J, Jackson L, 2017. Inverse relationship between urban green space and childhood autism in California elementary school districts. Environ. Int 107, 140–146. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Zablotsky B, Black LI, 2020. Prevalence of Children Aged 3–17 Years with Developmental Disabilities, by Urbanicity: United States, 2015–2018. Natl Health Stat Report 139, 1–7. [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2105994-supplement-1.docx (293.1KB, docx)

Data Availability Statement

The authors do not have permission to share data.

RESOURCES