Prenatal phthalate exposure and neurodevelopmental delay in early childhood (1 to 3 years): An Environmental influences on Child Health Outcomes (ECHO) study

Abstract

Phthalates are widely used in consumer products and are recognized as endocrine disruptors. Prenatal exposure to phthalates has been associated with various adverse health outcomes, including preterm birth and impaired fetal growth, and growing attention is being paid to their potential impact on child neurodevelopment. However, previous epidemiological studies examining prenatal phthalate exposure and child neurodevelopment have produced inconsistent or inconclusive findings, and evidence on phthalate mixtures remains limited. In this study, we utilized data from the Environmental influences on Child Health Outcomes (ECHO) Cohort to investigate associations between urinary biomarkers of prenatal phthalate exposure, both individually and as a mixture, and likelihood of neurodevelopmental delay (NDD) in offspring at ages 1 to 3 years. This analysis included 2378 pregnant person–child dyads from 10 ECHO cohorts who had measurements of NDD odds assessed using the Ages and Stages Questionnaire, Third Edition (ASQ-3). Our single-pollutant analyses revealed mixed findings. Higher prenatal exposure to certain phthalates was associated with higher odds of NDD across multiple domains, including motor and problem-solving skills, with evidence of effect modification by fetal sex. Conversely, we observed negative associations between specific prenatal phthalate concentrations and lower odds of NDD, particularly in communication domain. From mixture analyses, however, no significant associations were observed between the overall phthalate mixture and NDD odds in most domains, except for negative association for the personal-social domain. Further investigation into the biological mechanisms underlying these relationships, as well as more detailed evaluations of phthalate mixtures, will help advance our understanding of how prenatal phthalate exposure may influence early childhood neurodevelopment.

1. Introduction

Phthalates are a class of chemicals widely used in consumer products, including plastics, food packaging, vinyl floor coverings, and building materials, as well as solvents in personal care products, including lotions, nail polish, and perfumes (Wang & Qian, 2021). Human exposure to phthalates is widespread, occurring through the ingestion of contaminated food, as well as through dermal contact, inhalation, and parenteral administration of medical products (e.g., intravenous delivery of medications) (Koch et al., 2013; Wittassek et al., 2011).

Phthalates are known to disrupt the endocrine system and have also been linked to increased oxidative stress and epigenetic modification in pregnant people (Dutta et al., 2020; Holland et al., 2016; Qian et al., 2020). The prenatal period is a particularly vulnerable period for phthalate exposure due to significant physiological changes in the body, including hormone-mediated adaptations and substantial immune system modifications to accommodate the developing fetus (Varshavsky et al., 2020). Phthalates can freely cross the placenta (Latini et al., 2003), thus exposing the fetus to these chemicals. Additionally, phthalates may disrupt placenta development and function (Warner et al., 2021), which may lead to pregnancy complications and adversely affect fetal growth (Burton & Jauniaux, 2018) and neurodevelopment (Shallie & Naicker, 2019).

Neurodevelopmental delay (NDD) refers to slower progress in cognitive, speech and language, motor, and personal–social domains compared to that of others at the same age (Gupta & Kabra, 2014; Tervo, 2006). Significant development in these domains occurs within the first three years of life, making early diagnosis and intervention crucial. Persistent NDD during early childhood can indicate or evolve into more severe neurodevelopmental differences, such as learning disabilities, autism, and attention-deficit/hyperactivity disorder (ADHD) (Johnson et al., 2015; Megari & Miliadi, 2024; Mitchell et al., 2011). The developing brain is highly sensitive to environmental influences, and various neurotoxic exposures can increase the likelihood of neurobehavioral differences (Grandjean & Landrigan, 2014; Rice & Barone Jr, 2000). Previous animal studies have shown that phthalate exposure during pregnancy may disrupt neuronal differentiation and maturation (You et al., 2018), cause DNA damage in neurons (Xu et al., 2020), induce apoptosis in hippocampal neurons (Li et al., 2013), and alter hippocampal network plasticity (Holahan & Smith, 2015) in the developing fetal brain. Furthermore, in rodent models, offspring prenatally exposed to phthalates have exhibited neurobehavioral disturbances, including autistic traits (Li et al., 2022; Quinnies et al., 2017).

Despite growing evidence from experimental studies and the importance of NDD in early childhood, epidemiological studies investigating the association between prenatal phthalate exposure and child neurodevelopment during early childhood have reported inconsistent results (Antoniou & Otter, 2024; Ejaredar et al., 2015; Radke et al., 2020). For example, some studies found that urinary biomarkers of phthalate exposure during gestation were adversely associated with child neurodevelopment across multiple domains (e.g., cognitive, motor, and social behaviors) (Kim et al., 2011; Polanska et al., 2014; Whyatt et al., 2012), while other studies reported no significant associations with NDD. Additionally, a recent systematic review found significant, but inconsistent, sex-specific effects of prenatal phthalate exposure on NDD, with some studies reporting stronger associations among female offspring and others reporting associations only among males (Jankowska et al., 2021). Heterogeneity across studies may be attributed to (1) differences in phthalate exposure assessment, including the use of single phthalate measurements to represent exposure during pregnancy despite phthalates’ short half-lives, and differences in the phthalate metabolites under investigation; (2) differences in exposure levels across study sites; (3) differences in the cognitive and behavioral domains assessed and in the neurodevelopmental assessments utilized; (4) differences in the ages at which children were assessed; and (5) limited sample sizes. Furthermore, among the few studies that have examined the impact of phthalate mixtures on NDD, the findings have been inconclusive. One recent study (Loftus et al., 2021) found no associations between a mixture of 22 phthalate metabolites and measures of child language and cognitive abilities. In contrast, a study in a different population (Day et al., 2021) found that a mixture of nine phthalate metabolites was associated with differences in social communication and patterns of restricted behaviors. These limitations highlight the need for studies incorporating multiple study sites with varied exposure levels, assessing NDD probability at consistent developmental stages using similar assessment tools, and utilizing larger sample sizes powered to detect small associations. To fill this knowledge gap, our study leveraged data from the Environmental influences on Child Health Outcomes (ECHO) Cohort (Knapp et al., 2023) to examine associations between prenatal phthalate exposure and NDD odds in toddlerhood (ages 1 to 3 years). Specifically, we asked: In toddlers from a large U.S. child cohort, does higher prenatal exposure to phthalates, compared with lower prenatal exposure, increase the odds of NDD? We hypothesized that higher prenatal urinary phthalate metabolite concentrations, both individually and as a mixture, would be associated with higher odds of NDD and that these associations would differ by child sex.

2. Methods

2.1. Study population

The ECHO Cohort is a consortium of 69 ongoing pediatric and pregnancy cohorts across the United States established with the goal of investigating the effects of environmental exposures on children’s health (Knapp et al., 2023). The flowchart with inclusion/exclusion criteria for the current study is illustrated in Figure 1. Participants in the current study population were 2378 pregnant person–child dyads from ten ECHO cohorts where (1) the pregnant person was between 18 and 40 years of age at delivery and (2) had urinary phthalate metabolites measured during pregnancy and where (3) the child was a singleton birth with (4) a parent- or caregiver-completed Ages and Stages Questionnaire, Third Edition (ASQ-3) (Squires, Bricker, et al., 2009) when the child was eligible to receive the 12–36 month questionnaires (age 11 through 38 months). When more than one child per family (defined by the pregnant parent) was eligible, one child was randomly selected to be included in the study population. Cohorts with fewer than 25 eligible dyads were dropped (N=5 dyads from 1 cohort). Detailed cohort information is summarized in Supplementary Table 1.

Figure 1.

Flowchart describing the inclusion and exclusion criteria for the study population as well as the study sample sizes.

2.2. Prenatal phthalate exposure measurement

We included data on eleven phthalate exposure biomarkers measured in 4493 urine samples collected during pregnancy. Selected analytes included mono-benzyl phthalate (MPZP), mono-carboxy isononyl phthalate (MCINP/MCOMOP/MCNP), mono-carboxy isooctyl phthalate/mono-(carboxyoctyl) phthalate (MCIOP/MCOP), mono (3-carboxypropyl) phthalate (MCPP), mono-(2-ethyl-5-carboxypentyl) phthalate/mono (5-carboxy-2-ethylpentyl) phthalate (MECPP/MCEPP), mono (2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), mono ethyl hexyl phthalate (MEHP), mono (2-ethyl-5-oxohexyl) phthalate (MEOHP), monoethyl phthalate (MEP), and composite of mono-isobutyl phthalate and mono-n-butyl phthalate (MnBP+MiBP).

These analytes were the most commonly measured phthalate metabolites across multiple ECHO cohorts and were detected in more than 80% of our study population, except for MEHP (66.6%). Additionally, we calculated a summary measure of di-2-ethylhexyl phthalate (∑DEHP) exposure by adding the molar fractions of its metabolites – MEHP, MEHHP, MEOHP and MECPP/MCEPP – and multiplying the sum by the molecular weight of the primary metabolite, MEHP, for unit comparability (ng/mL). Samples were assayed at two laboratories: the Centers for Disease Control and Prevention (CDC; Atlanta, GA), which analyzed 1,810 samples, and the Wadsworth Human Health Exposure Analysis Resource (HHEAR) Laboratory (Menands, NY), which analyzed 2,683 samples. Analysis of samples at the CDC and HHEAR laboratories used comparable instrumentation, such as high-performance liquid chromatography-tandem mass spectrometry (Kannan et al., 2021; Kato et al., 2005; Silva et al., 2007). Both laboratories also followed strict quality assurance and control measures and participate in the German External Quality Assessment Scheme (G-EQUAS) and CDC Proficiency Testing. Additionally, quality assurance protocols followed at HHEAR have been shown to ensure reliable and reproducible data quality for phthalate metabolites (Kannan et al., 2021). We did not observe meaningful differences in phthalate metabolite concentration values across the two laboratories in the overall analysis population nor within the two cohorts that had samples assayed at both laboratories (Supplementary Table 3).

All analytes below the limit of detection (LOD) were imputed as LOD/square root of 2 regardless of the proportion of observations below the LOD, with batch-specific LODs used in the imputation. The measured concentrations were adjusted for urinary dilution using a modified Boeniger method (Kuiper et al., 2021; Kuiper et al., 2022). Briefly, urinary dilution was measured using either specific gravity or creatinine, and phthalate metabolite concentrations were multiplied by the ratio of the study population median dilution value to the individual sample dilution value. Kuiper et al. (2022) reported high agreement between urinary phthalate biomarker concentrations corrected using creatinine and specific gravity when applying this method to samples from the same trimester in pregnancy, and this approach has been used in previous ECHO studies (Bloom et al., 2025; Boissiere-O’Neill et al., 2025; Meeker et al., 2024; Oh et al., 2024). Specific gravity correction was preferred where available as sampling occurred at variable times over pregnancy. Most urine samples with phthalate exposure measures were collected during the second trimester (63%), followed by the third (18%) and first trimesters (17%) (Supplementary Table 4). Generally, urine samples were collected at each study visit which occurred several weeks apart, though not all visits aligned precisely with each trimester. Exposure measures were derived either from the single available pregnancy urine sample for approximately half of participants or from the average of repeated pregnancy urine samples for the remaining participants (median number of samples = 3, range = 2–6). Exposure measures were log2-transformed to account for the log-normal distribution.

2.3. ASQ-3 Neurodevelopmental Assessment

The ASQ-3 is a 30-item screening tool designed to assess development across five domains, namely, communication, gross motor, fine motor, problem solving, and personal–social skills, in children ages 1 month through 5 ½ years (Squires, Twombly, et al., 2009). The ASQ-3 takes 10–15 minutes to complete, and it was administered in either English or Spanish in our study. A parent or caregiver answered six developmentally appropriate questions about their child’s behavior in each domain based on the child’s age; >95% of respondents in the sample were the child’s biological mother. All cohorts collected ASQ-3 data from caregivers in a clinic/research setting. Responses correspond to point values (Yes=10 points; Sometimes=5 points; Not yet=0 points); points were summed to give a total score within each domain, which determines a child’s score compared to a US reference population age 1 and 66 months. Scores are considered above the cutoff (> 1 standard deviation (SD) below the mean), close to the cutoff (≧ 1 and < 2 SD below the mean), or below the cutoff (≧ 2 SD below the mean) for on-schedule age-specific development in that domain (Nozadi et al., 2023; Squires, Twombly, et al., 2009). The appropriate questionnaire was selected based on a child’s age at assessment and gestational age at delivery; collection windows for children who were born at or before 36 completed weeks gestation and who were under two years old at ASQ-3 administration were adjusted to account for expected delays due to prematurity (Squires, Twombly, et al., 2009).

The ASQ-3 demonstrates high test–retest and interobserver reliability (>90%), indicating (1) consistent parental evaluations over time and (2) strong agreement between professional examiners and parental reports. Correlations across developmental domains and the overall ASQ-3 score further support its moderate to strong internal consistency (Squires, Twombly, et al., 2009). The ASQ-3 also shows moderate to high concordance with standardized assessments, such as the Battelle Developmental Inventory (BDI) for children aged 1 to 66 months (Squires, Twombly, et al., 2009), and the Bayley Scales of Infant and Toddler Development, 3rd Edition, with moderate sensitivity and specificity for identifying significant neurodevelopmental differences in children aged 12 to 60 months (Muthusamy et al., 2022). The ASQ-3 also has demonstrated high validity and reliability in diverse populations (Singh et al., 2017).

For the current analysis, we included children with at least one ASQ-3 assessment at 12–36 months. We selected this age range to reflect “toddlerhood”, a critical period of rapid cognitive, emotional, and social development, as well as a key period for interventions in the case of potential developmental delays. If a child had more than one ASQ-3 assessment during that timeframe, the most recent questionnaire was selected for analysis. We collapsed the close to the cutoff and below the cutoff categories to create a binary outcome of likely for developmental delay (close to or below the cutoff) vs. typical development (above the cutoff).

2.3. Statistical analysis

We calculated descriptive statistics for both participant-level (e.g., pregnant person’s age and education) and specimen-level (e.g., season of collection) characteristics, as well as descriptive statistics for pregnancy-average phthalate concentrations (unadjusted and dilution-adjusted) in the study population and by each ECHO cohort. We also calculated Spearman correlations between analytes and intraclass correlation coefficients (ICCs) to assess between- and within-individual variation of analyte concentrations in repeated measurements. We used the multiple imputation by chained equations (MICE) method to impute missing data for race/ethnicity (<1%), maternal educational attainment (5%), and pre-pregnancy body mass index (BMI) (9%). We generated 25 complete datasets using MICE and pooled the results from each dataset using Rubin’s rules (Rubin, 2004).

We selected potential confounders a priori based on existing literature and data availability. Adjusted models accounted for the pregnant individual’s age at delivery (years; continuous), race/ethnicity (Hispanic, non-Hispanic Black/African-American, non-Hispanic White, non-Hispanic other [Asian, Native Hawaiian or other Pacific Islander, American Indian or Alaska Native, or multiple races]) as a proxy for exposure to structural racism, educational attainment (high school or less, some college or Associate’s degree or trade/vocational school, Bachelor’s degree or higher), pre-pregnancy BMI (categorical), child’s age at ASQ-3 assessment (months; continuous), and (except in sex-stratified models) child’s sex. A directed acyclic graph (DAG) illustrating exposure, outcomes, and potential confounders is presented in Supplementary Figure 1. Pregnancy complications and preterm birth were not included as potential confounders, as they may lie on the causal pathway between prenatal exposure and child outcome. Additionally, breastfeeding was not adjusted for in our models because it is not associated with prenatal exposure (i.e., not a confounder), though it may serve as an indicator of postnatal phthalate exposure, suggesting a distinct exposure window that could be explored in future work. In our primary analyses, we used generalized estimating equations (GEEs) (Chen et al., 2002) fit with the binomial family and logit link, with exchangeable correlation structure and a robust standard error estimator, to estimate the association between each individual analyte and NDD odds in each ASQ-3 domain. We clustered GEEs on ECHO cohort to account for non-independence of observations among children enrolled to the same extant cohort protocol. Two models did not converge using GEEs, so for those models only, we used logistic regression with clustered standard errors to conduct the analysis. To reduce any influence of extreme exposure values we conducted sensitivity analyses utilizing quartiles of urinary phthalate metabolite concentrations.

We estimated the association between exposure to the overall mixture of all included analytes for each ASQ-3 domain using quantile-based g-computation, the details of which have been previously published elsewhere (Keil et al., 2020). Briefly, quantile-based q-computation uses a marginal structural model to evaluate the expected change in the outcome per quantile increase in all mixture analytes, thus providing an overall mixture effect. In addition to providing the overall mixture effect, quantile-based q-computation is advantageous over other mixtures methods as it does not have a directional homogeneity assumption and can be used with both multiple imputation for missing covariates and effect measure modification analyses. We ran unadjusted and adjusted iterations of each analyte (or mixture)–outcome pairing in the overall sample and in sex-stratified groups. A threshold of P<0.05 was used for statistical significance. Finally, we investigated the potential for effect measure modification by child sex by including a term for the interaction of each analyte and child sex in single-pollutant analyses and an interaction term of the phthalate mixture and child sex in mixture analyses using an augmented product term approach (Buckley et al. 2017); a threshold of P<0.10 was used to determine the significance of an interaction. In sensitivity analyses, we conducted leave-one-cohort-out analyses to assess the robustness of our primary findings to differences in the included cohorts. Moreover, we repeated our primary analyses with additional adjustment for prenatal smoking, alongside all covariates in the main model.

Descriptive statistical analysis, multiple imputation for missing data, and individual analyte analyses were conducted using Stata 18.0 (StataCorp LLC; College Station, TX). Spearman correlations and mixture analyses were conducted using R Studio Version 2024.04.0 (R Core Team, 2021).

3. Results

The study sample comprised 2378 pregnant person–child dyads from 10 ECHO cohorts. The sociodemographic and pregnancy characteristics of all ECHO cohorts, cohorts included in this study, and the study participants are summarized in Table 1. In the current study, pregnant participants were demographically diverse, with 44% being non-Hispanic White; 40% Hispanic; 11% non-Hispanic, other races, or multiple races; and 5% non-Hispanic Black/African American. Most pregnant participants gave birth at age 30–34 years (40%), had a normal pre-pregnancy BMI between 18.5 and 24.5 kg/m² (45%), had a bachelor’s degree or higher (60%), were married or living with a partner (75%), and did not smoke during pregnancy (90%). Most children (94%) were born at term (≥ 37 completed weeks), with similar numbers of males (51%) and females (49%) assigned at birth. The mean age at ASQ-3 assessment was 26.3 (SD=8.1) months. These characteristics are largely consistent with the larger population (N=13641) from ten ECHO cohorts included in this study. Compared to the full ECHO cohort (N = 65,218), some differences were observed in race/ethnicity, pre-pregnancy BMI, maternal age at delivery, and other characteristics, but these differences may reflect the substantial proportion of missing data (~20%) in the overall cohort. Additional sociodemographic characteristics by each ECHO cohort are presented in Supplementary Table 2. Distributions of urinary phthalate metabolite concentrations from the study population are summarized in Table 2, and corresponding distributions by laboratory of analysis are provided in Supplementary Table 3. Additional characteristics for urine sample collection from pregnant participants, including time of day, trimester, season, and calendar year of collection, collection type, number of freeze-thaw cycles, and urinary dilution measure, from study population, laboratory of analysis, and by each ECHO cohort are described in Supplementary Table 4 and Supplementary Table 5. The detection rate for all phthalate metabolites exceeded 80%, except for MEHP, which was detected in 66.6% of samples. Among the metabolites, MEP had the highest average concentration (median = 24.1 ng/mL), while MCPP had the lowest concentration (median = 0.7 ng/mL) in the study population. Nine of the ten cohorts included in this study have specific gravity measurements, while one cohort (Healthy Start) only has creatinine measurements available (Supplementary Table 5). Correlations between phthalate metabolites ranged from moderately weakly positive to strongly positive, with metabolites from the same parent compound showing the strongest correlations (Supplementary Figure 2). For example, the four DEHP metabolites were moderately to strongly correlated, with correlation coefficients ranging from 0.6 to 0.9. In contrast, MEP, the primary metabolite of DEP, showed low correlations (ρ: 0.13–0.28) with all other metabolites. The ICCs for repeat measures of a given analyte ranged from 0.08 to 0.29. Among participants with multiple samples, these samples were not always provided during different trimesters. Most samples in the study population were collected during the second trimester (63%), and this pattern was consistent across nearly all study cohorts. A few exceptions included Healthy Start, where most samples were collected in the third trimester (63%), and MADRES (68%) and PETALs (52%), where most samples were collected in the first trimester. Sample collection seasons were evenly distributed both overall and within each study cohort. Additional descriptive statistics for prenatal urinary phthalate measurements by study cohort are presented in Supplementary Table 6. The distribution of ASQ-3 neurodevelopmental outcomes is summarized in Table 3. Briefly, 13–19% of study participants were flagged for potential delay in each domain as assessed by the ASQ-3: communication (19%), gross motor (13%), fine motor (18%), problem solving (16%), and personal–social (18%).

Table 1.

Sociodemographic characteristics of 2378 pregnant person-child dyads in the study population

	Entire ECHO Cohort (N=65218)	Ten Cohorts Included (N=13641)	Study Sample (N=2378)
Pregnant participant characteristics
Maternal age at delivery (years)
<25 years	12092 (19%)	2142 (16%)	315 (13%)
25 to <30 years	14304 (22%)	3253 (24%)	562 (24%)
30 to <35 years	16276 (25%)	4673 (34%)	958 (40%)
>=35 years	10383 (16%)	3326 (24%)	543 (23%)
Missing	12163 (19%)	247 (2%)
Maternal race
White	31545 (48%)	8692 (64%)	1445 (61%)
Black	8769 (13%)	1170 (9%)	141 (6%)
Asian	2631 (4%)	937 (7%)	193 (8%)
Native Hawaiian Pacific Islander	204 (<1%)	122 (1%)	22 (1%)
American Indian or Alaska Native	1194 (2%)	85 (1%)	13 (1%)
Multiple race	1968 (3%)	570 (4%)	112 (5%)
Other race	2167 (3%)	942 (7%)	220 (9%)
Missing	16740 (26%)	1123 (8%)	232 (10%)
Maternal ethnicity
Non-Hispanic	39497 (61%)	8875 (65%)	1415 (60%)
Hispanic	12420 (19%)	4570 (34%)	958 (40%)
Missing	13301 (20%)	196 (1%)	5 (<1%)
Race/ethnicity
Non-Hispanic white	25527 (39%)	6515 (48%)	1045 (44%)
Non-Hispanic Black/African-American	7790 (12%)	1016 (7%)	113 (5%)
Non-Hispanic, other or multiple race	5038 (8%)	1318 (10%)	252 (11%)
Hispanic	12420 (19%)	4570 (34%)	958 (40%)
Missing	14443 (22%)	222 (2%)	10 (<1%)
Pre-pregnancy BMI (kg/m2)
Underweight	1294 (2%)	338 (2%)	71 (3%)
Normal	18494 (28%)	5382 (39%)	1061 (45%)
Overweight	10889 (17%)	3115 (23%)	561 (24%)
Obese	11045 (17%)	2928 (21%)	466 (20%)
Missing	23496 (36%)	1878 (14%)	219 (9%)
Highest educational attainment
High school, GED, equivalent, or less than high school	15006 (23%)	2703 (20%)	386 (16%)
Some college, Associate's degree, or trade/vocational school	16335 (25%)	2867 (21%)	459 (19%)
Bachelor's degree or higher	25426 (39%)	6471 (47%)	1424 (60%)
Missing	8451 (13%)	1600 (12%)	109 (5%)
Marital status
Married or living with a partner	29936 (46%)	9684 (71%)	1793 (75%)
Not married (widowed, separated, divorced, or single never married) or partnered not living together	8729 (13%)	1829 (13%)	218 (9%)
Missing	26553 (41%)	2128 (16%)	367 (15%)
Parity
Nulliparous	16462 (25%)	5679 (42%)	1107 (47%)
Primiparous	13746 (21%)	4408 (32%)	774 (33%)
Multiparous	10809 (17%)	1914 (14%)	242 (10%)
Missing	24201 (37%)	1640 (12%)	255 (11%)
Child year of birth
Before 2012	27722 (43%)	976 (7%)	0 (0.0%)
2012–2014	12193 (19%)	2012 (15%)	313 (13%)
2015–2017	10006 (15%)	3264 (24%)	806 (34%)
2018–2021	11561 (18%)	5751 (42%)	1259 (53%)
After 2021	3736 (6%)	1638 (12%)	0 (0.0%)
Tobacco use during pregnancy
No	48054 (74%)	10888 (80%)	2133 (90%)
Yes	5521 (8%)	521 (4%)	72 (3%)
Missing	11643 (18%)	2232 (16%)	173 (7%)
Gestational age at birth, median (IQR)	39.0 (38.0, 40.0) (n=52961)	39.0 (38.0, 40.0) (n=13533)	39.0 (38.0, 40.0) (n=2378)
Preterm
Term >=37 completed weeks	53043 (81%)	12425 (91%)	2240 (94%)
Preterm <37 weeks	8565 (13%)	1108 (8%)	138 (6%)
Missing	3610 (6%)	108 (1%)	0 (0.0%)
Infant sex
Male	33610 (52%)	6865 (50%)	1204 (51%)
Female	31419 (48%)	6682 (49%)	1174 (49%)
Intersex	8 (<1%)	0 (0%)	0 (0.0%)
Missing	181 (<1%)	94 (1%)	0 (0.0%)

Abbreviations: ASQ-3, Ages and Stages Questionnaire, Third Edition; BMI, body mass index; CIOB, Chemicals in Our Bodies; GA, gestational age; GED, General Educational Development; IKIDS, Illinois Kids Development Study; IQR, interquartile range; kg, kilograms; m², meters squared; MADRES, Maternal And Developmental Risks from Environmental and Social Stressors; MARCH, Michigan Archive for Research with Mothers on Child Health; NHBCS, New Hampshire Birth Cohort Study; NYU CHES, New York University Children’s Health and Environment Study; PETALS, Pregnancy Environment and Lifestyle Study; PROTECT, Puerto Rico Testsite for Exploring Contamination Threats; SD, standard deviation.

All statistics are frequency (%) unless noted otherwise.

Table 2.

Descriptive statistics of prenatal urinary phthalate metabolite concentrations (ng/mL) from 2378 pregnant study participants.

Full analyte name	Abbreviation	Detection	Min	10th percentile	25th percentile	Median	75th percentile	90th percentile	Max
Raw concentrations
mono-benzyl phthalate	MBZP	89.7%	<LOD	0.2	0.9	2.6	6.7	17.9	351.0
mono-carboxy isononyl phthalate	MCINP/MCOMOP/MCNP	96.3%	<LOD	0.2	0.5	1.0	2.1	4.1	139.0
mono-carboxy isooctyl phthalate/mono-(carboxyoctyl) phthalate	MCIOP/MCOP	98.7%	<LOD	0.7	1.5	3.7	8.7	23.5	1570.0
mono (3-carboxypropyl) phthalate	MCPP	82.0%	<LOD	0.1	0.3	0.7	1.3	2.7	120.0
mono-(2-ethyl-5-carboxypentyl) phthalate/mono (5-carboxy-2-ethylpentyl) phthalate	MECPP/MCEPP	99.4%	<LOD	1.4	2.9	5.8	10.7	19.6	1305.9
mono (2-ethyl-5-hydroxyhexyl) phthalate	MEHHP	99.4%	<LOD	1.3	2.4	4.7	8.7	15.0	1522.8
mono ethyl hexyl phthalate	MEHP	66.6%	<LOD	0.0	0.2	0.8	2.0	4.2	584.1
mono (2-ethyl-5-oxohexyl) phthalate	MEOHP	99.6%	<LOD	0.7	1.6	3.1	5.8	10.0	855.2
monoethyl phthalate	MEP	99.6%	<LOD	5.2	10.4	24.1	59.3	141.3	11564.4
Composite of mono-isobutyl phthalate and mono-n-butyl phthalate	MnBP+MiBP	(see footnote)	<LOD	4.0	8.1	17.0	33.0	61.2	4471.1
∑ di-2-ethylhexyl phthalate	DEHP	NA	<LOD	3.7	7.3	14.3	26.2	44.2	3535.9
Dilution-adjusted concentrations
mono-benzyl phthalate	MBZP	--	0.0	0.4	1.2	2.8	6.7	16.3	246.0
mono-carboxy isononyl phthalate	MCINP/MCOMOP/MCNP	--	0.0	0.2	0.6	1.2	2.2	4.2	231.8
mono-carboxy isooctyl phthalate/mono-(carboxyoctyl) phthalate	MCIOP/MCOP	--	0.0	1.0	1.8	3.9	9.2	23.9	1388.8
mono (3-carboxypropyl) phthalate	MCPP	--	0.0	0.1	0.4	0.8	1.5	2.8	129.8
mono-(2-ethyl-5-carboxypentyl) phthalate/mono (5-carboxy-2-ethylpentyl) phthalate	MECPP/MCEPP	--	0.0	2.1	3.7	6.4	10.8	18.5	768.9
mono (2-ethyl-5-hydroxyhexyl) phthalate	MEHHP	--	0.0	1.8	3.1	5.1	8.3	13.8	864.5
mono ethyl hexyl phthalate	MEHP	--	0.0	0.0	0.2	1.0	2.1	3.8	331.6
mono (2-ethyl-5-oxohexyl) phthalate	MEOHP	--	0.0	1.2	2.0	3.4	5.5	9.5	503.5
monoethyl phthalate	MEP	--	0.0	6.9	12.3	25.9	62.9	143.6	6564.9
Composite of mono-isobutyl phthalate and mono-n-butyl phthalate	MnBP+MiBP	--	0.0	6.3	10.4	18.8	31.9	53.7	2361.6
∑ di-2-ethylhexyl phthalate	DEHP	--	0.0	5.7	9.4	15.5	25.2	42.1	2007.3

Abbreviations: Min, minimum; Max, maximum; LOD, limit of detection. All 2,378 participants have all analytes measured -- these are average concentrations. MnBP/MiBP was measured either as the composite or as its component parts and summed. Detection frequencies are: MiBP: 98.3%, MnBP: 98.8%, Composite: 99.5%. LODs vary across labs analyzing the included specimens. Cells labeled "<LOD" indicate that the value was below the lowest LOD and therefore represent imputed values. Values below the LOD were imputed as LOD/√2. DEHP was calculated based on the molar sum of its parts and therefore does not have an LOD.

Table 3.

ASQ-3 neurodevelopmental assessment outcome measures in the study population (N=2,378) overall and stratified by child sex.

	Communication
	Overall (N=2378)	Male (N=1204)	Female (N=1174)
Typical development	1923 (81%)	892 (74%)	1031 (88%)
Potential delay	455 (19%)	312 (26%)	143 (12%)
	Gross Motor
	Overall (N=2378)	Male (N=1204)	Female (N=1174)
Typical development	2071 (87%)	1047 (87%)	1024 (87%)
Potential delay	307 (13%)	157 (13%)	150 (13%)
	Fine Motor
	Overall (N=2378)	Male (N=1204)	Female (N=1174)
Typical development	1956 (82%)	954 (79%)	1002 (85%)
Potential delay	422 (18%)	250 (21%)	172 (15%)
	Problem Solving
	Overall (N=2378)	Male (N=1204)	Female (N=1174)
Typical development	2004 (84%)	980 (81%)	1024 (87%)
Potential delay	374 (16%)	224 (19%)	150 (13%)
	Personal-Social
	Overall (N=2378)	Male (N=1204)	Female (N=1174)
Typical development	1949 (82%)	919 (76%)	1030 (88%)
Potential delay	429 (18%)	285 (24%)	144 (12%)

In our single-pollutant analyses, we identified significant associations between individual phthalate metabolites and the odds of potential neurodevelopmental delays assessed by the ASQ-3 in both unadjusted and adjusted models (Supplementary Table 7). For example, the unadjusted models revealed several univariate associations: monobenzyl phthalate (MBZP) was associated with the communication (odds ratio [OR] = 1.05, 95% confidence interval [CI]: 1.01, 1.10), fine motor (OR = 1.07, 95% CI: 1.03, 1.12), and problem-solving (OR = 1.04, 95% CI: 1.00, 1.07) domains. Additionally, MEHHP and MEHP were associated with the fine motor domain, and MEP with the problem-solving domain. After adjusting for potential confounders, several of these associations remained significant. A doubling of prenatal urinary MBZP concentrations was associated with higher odds of potential developmental delay in fine motor (OR=1.05, 95% CI: 1.00, 1.09) skill (Figure 2, Supplementary Table 7). Additionally, a doubling of prenatal urinary MEP concentrations was associated with 8% higher odds of potential delay in problem-solving skills (OR: 1.08, 95% CI: 1.02, 1.13). Unexpectedly, we observed a few inverse associations. For instance, a doubling of prenatal urinary MCINP/MCOMOP/MCNP concentration was associated with lower odds of potential delay in communication skills (OR=0.94, 95% CI: 0.89, 0.99), and a doubling of prenatal urinary MCPP concentrations was associated with 3% lower odds of potential delay in problem-solving skills (OR=0.97, 95% CI: 0.94, 1.00). The overall phthalate mixture was not associated with the odds of experiencing potential delays in most ASQ-3 domains, except for an association with lower odds of delay in the personal-social domain (OR=0.83, 95%CI: 0.7, 0.99) (Supplementary Table 9). To reduce any influence of extreme values in our population, we conducted additional analyses examining associations between quartiles of phthalate concentrations with outcomes and results were consistent findings from continuous exposure models (data not shown).

Figure 2.

Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) for associations of doubling prenatal urinary phthalate metabolite concentrations, or their mixture, with ASQ-3-assessed neurodevelopmental delay in the a) communication, b) gross motor, c) fine motor, d) problem-solving, and e) personal–social domains. Models were adjusted for pregnant parent’s age, race/ethnicity, education, pre-pregnancy BMI, child age at ASQ-3 assessment, and child sex. ‘*’ indicates that GEE models did not converge; logistic regression was used with clustered SE for study cohorts.

In our exploration of effect modification by child sex, we identified several sex-specific associations between biomarkers of prenatal phthalate exposure and child ASQ-3 outcomes and significant pollutant-sex interactions (Figure 3, Supplementary Table 8). In both unadjusted and adjusted sex-stratified models, associations were predominantly observed among male children. For instance, in sex-stratified adjusted models, a doubling of prenatal urinary DEHP metabolite concentrations was associated with 14% higher odds of delay in the fine motor domain specifically among male children (OR=1.14, 95% CI: 1.02, 1.28), but not among female children (OR=0.92, CI: 0.82, 1.04), with a significant sex-by-exposure interaction in the overall model (p-interaction<0.01). A similar trend was observed for DEHP metabolites MEHHP and MEHP. Similarly, the positive association between prenatal urinary MBZP concentrations and potential fine motor delay seen in the overall model was driven by the association among males (males: OR=1.08, 95% CI: 1.03, 1.13; females: OR=1.00, 95% CI: 0.94, 1.07; p-interaction from overall model=0.01). In addition, the composite of MnBP and MiBP was significantly associated with higher odds of fine motor delay among male children (OR = 1.16; 95% CI: 1.05, 1.28), but not among female children (OR = 1.01; 95% CI: 0.88, 1.06). The interaction term in the overall model indicated sex-specific differences (p-interaction = 0.05). In contrast, the association of prenatal urinary MEP concentrations with delays in problem solving was driven by the association among females, though the trend remained in the same direction for males (males: OR=1.05, 95% CI: 0.99, 1.12; females: OR=1.10, 95% CI: 1.02, 1.20; p-interaction from overall model=0.34). In mixture models, we examined effect modification by child sex by including interaction terms and through stratification (Supplementary Table 9). In sex-stratified mixture analyses (Figure 3), although small differences in effect estimates were observed, we did not identify significant mixture effects within either sex. Consistently, the sex-by-mixture interaction was not statistically significant (Supplementary Table 9).

Figure 3.

Sex-stratified adjusted odds ratios (aORs) and 95% confidence intervals (CIs) for associations of doubling prenatal urinary phthalate metabolite concentrations, or their mixture, with ASQ-3-assessed neurodevelopmental delay in the a) communication, b) gross motor, c) fine motor, d) problem-solving, and e) personal–social domains. Models were adjusted for pregnant parent’s age, race/ethnicity, education, pre-pregnancy BMI, and child age at ASQ-3 assessment. ‘*’ indicates that GEE models did not converge; logistic regression was used with clustered SE for study cohorts.

In sensitivity analyses, leave-one-out testing of associations between individual phthalate metabolites and ASQ-3 domain scores confirmed the robustness of our results to specific cohort exclusions (results not shown). Similarly, leave-one-out testing of associations between phthalate mixture and ASQ-3 only made minimal changes in our results (Supplementary Figure 3). Furthermore, we conducted additional sensitivity analyses including prenatal smoking and found that effect estimates did not meaningfully change (Supplementary Table 10).

4. Discussion

This study investigated associations between prenatal phthalate exposures and oddds of child developmental delays at ages 12 to 36 months assessed via a commonly used developmental screening tool. We utilized data from the nationwide ECHO Cohort, which is a large, racially and ethnically diverse cohort, to examine these associations. Within this study population, 13–19% were flagged for potential delay, which is consistent with previous studies that have examined neurodevelopmental delays among U.S. children aged 1 to 5 years (Johnson et al., 2024; Leeb et al., 2024).

We identified several significant associations between individual urinary phthalate metabolites and higher odds of potential NDD across multiple domains – communication, gross motor, fine motor, problem solving, and personal–social – assessed by the ASQ-3. At the same time, we also observed some negative associations, where higher prenatal concentrations of specific phthalates were linked to lower odds of NDD. These associations differed by child sex, in general showing stronger associations among male children. However, mixture analyses did not reveal associations between overall phthalate exposure and most ASQ-3 domains, except for a negative association with the personal-social domain.

Our findings align with previous studies suggesting adverse effects of prenatal phthalate exposure on neurodevelopment among infants and toddlers (Minatoya & Kishi, 2021). For example, urinary MBZP concentrations during pregnancy were associated with poorer psychomotor development at ages 1, 4, and 7 years in a birth cohort in Spain (Gascon et al., 2015) and with poorer communication skills at age 6 months and gross motor skills at age 1 year among children in Puerto Rico (Park et al. 2023), both of which are consistent with our present findings in this larger cohort. In addition, Kim et al. (2018) showed that prenatal exposure to MEP was associated with mental, psychomotor, and social developmental problems at ages 1–2 years in a birth cohort in Korea, while we reported similar findings in relation to problem solving.

In the current work, there are also some discrepancies compared to previous studies. For example, while we did not observe any association between DEHP and/or its metabolites with ASQ-3 scores in the overall sample, previous studies from various countries including the United States, Norway, Spain, Taiwan, Korea, and Mexico have all reported associations between DEHP metabolites and potential delay in multiple domains (Radke et al., 2020). Specifically, DEHP metabolites were associated with lower cognition among children in Korea at 6 months (Kim et al., 2011), in Poland at 2 years (Polanska et al., 2014), in Mexico at 2–3 years (Téllez-Rojo et al., 2013), and in the United States at 3 years of age (Li et al., 2019). The observed discrepancies in the study findings may be attributed to differences in the developmental measures utilized or differences in prenatal phthalate exposure across populations. Urinary phthalate concentrations reported in this analysis are indeed lower than those reported in many previous studies of phthalate exposure and neurodevelopmental outcomes, possibly due to variations in both geographical and temporal patterns of phthalate exposure. For example, a recent study reported that urinary phthalate concentrations differ significantly by region, with levels in the United States being comparable to or higher than those in Korea and Norway, yet lower than those observed in Puerto Rico, Spain, and Mexico (Zhang et al., 2021). In addition, the years during which phthalate exposure was measured could influence findings, as trends in phthalate usage have shifted over time. For example, the recruitment periods of the above studies ranged from 2001 to 2011, potentially resulting in varying exposure over time. Additionally, changes in relevant lifestyle factors may alter the associations between phthalate exposure and neurodevelopmental outcomes.

Sex-specific differences between prenatal phthalate exposure and child neurodevelopment across multiple child ages have been reported (Jankowska et al., 2021). For example, the associations of DEHP metabolites with motor skills were stronger among male infants than female infants at 6 months (Kim et al., 2011) and at 1 year (Park et al., 2023). In our analysis, we observed consistent trends of stronger association of DEHP with higher odds of delay in fine motor skills among male children compared to female children. In addition, Doherty et al. (2017) showed that MBZP and MCPP concentrations were associated with worse psychomotor ability at 2 years specifically among female children. However, in our analysis, MCPP associations were comparable by fetal sex, while MBZP was only associated with odds of worse fine motor skills among males. Moreover, although we observed a child-sex–specific association between the MnBP+MiBP composite and higher odds of fine motor delay among male children, such sex differences were not evident in a recent study conducted in Puerto Rico (Park et al., 2023). In contrast, a study from New York City reported stronger associations between MnBP and psychomotor delay among female children compared with males, whereas MiBP showed similar adverse associations across both sexes (Whyatt et al., 2012). Limited epidemiological studies have examined the impact of overall phthalate mixtures in relation to child neurodevelopment, and results are largely inconsistent. Some reported null associations (Dewey et al., 2023; Loftus et al., 2021), while others suggested adverse cumulative effects (Day et al., 2021; Kearns et al., 2024).

Prenatal exposure to phthalates may affect child neurodevelopment through multiple mechanisms, including disruption of the thyroid hormone system (Qian et al., 2020; Salazar et al., 2021). Phthalates may bind to thyroid hormone receptors (TRα and TRβ) to disturb the signaling of thyroid hormone and affect thyroid hormone production (Zoeller, 2005). Previous epidemiological studies have reported altered levels of thyroid hormones, such as thyroid-stimulating hormone (TSH) and free T4 (FT4), in pregnant people (Al-Saleh et al., 2024; Johns et al., 2015; Nakiwala et al.; Wu et al., 2022) as well as disruption in neonatal thyroid function (Coiffier et al., 2023) associated with gestational phthalate exposure. Maternal thyroid levels during pregnancy are critical for neuronal migration, synaptogenesis, and fetal brain development (de Escobar et al., 2008; Moog et al., 2017; Thompson et al., 2018). A recent study in mice showed that emotional and cognitive impairment of prenatally phthalate exposed offspring was induced through interrupted thyroid signaling system (Lv et al., 2022). However, further epidemiological research is needed to examine the role of maternal thyroid function as a mediator linking prenatal phthalate exposure and neurodevelopmental outcomes.

In addition, prenatal phthalate exposure may affect child neurodevelopment by elevating maternal inflammation and oxidative stress. Mounting evidence suggests that phthalate exposure during pregnancy is associated with elevated maternal levels of inflammation and immune activation biomarkers such as C-reactive protein, IL-6, and IL-10 (Liu et al., 2022; Sweeney et al., 2019), while proinflammatory cytokines, such as IL-6 and IL-1β, can cross the placenta and enter the fetal bloodstream (Dahlgren et al., 2006; Girard et al., 2010; Zaretsky et al., 2004). Previous studies also have shown that phthalate exposure is associated with higher maternal oxidative stress during pregnancy (Ferguson et al., 2014; Holland et al., 2016; Waits et al., 2020). The fetal brain is susceptible to oxidative stress because the expression of antioxidant enzymes is relatively low, and oxidative stress may alter neural pathways and development of the nervous system (Ikonomidou & Kaindl, 2011). In addition, prenatal phthalate exposure may activate maternal inflammation and oxidative stress in a fetal-sex-specific manner (Lee et al., 2023; Puttabyatappa et al., 2020; Wang et al., 2020). A recent mouse study also found that male offspring were more vulnerable to prenatal inflammation, showing long-term neurologic effects from chronic brain inflammation (Dada et al., 2014).

Phthalates’ ability to disrupt the sex steroid hormone system may also help understand the observed sex-specific associations. For example, in vitro estrogenic (Harris et al., 1997), anti-estrogenic, and anti-androgenic activities (Christen et al., 2012; Czernych et al., 2017) of various phthalates have been reported. Moreover, epidemiological studies have consistently shown that phthalate exposure is associated with altered reproductive hormone concentrations during pregnancy (Cathey et al., 2019; Johns et al., 2015; Pacyga et al., 2021; Sathyanarayana et al., 2014), with implications for neurodevelopment. The brain is a sexually dimorphic organ, with variations in maternal sex steroid levels during gestation contributing to sex-specific brain structures (Romano et al., 2016). As a result, altered reproductive hormone levels during pregnancy, potentially linked to phthalate exposure, could mediate the association between prenatal phthalate concentrations and child neurodevelopment.

Our findings should be interpreted with acknowledgement of the current study’s limitations. We included a limited number of phthalate metabolites in our analyses, excluding increasingly common phthalate replacement chemicals such as di-2-ethylhexyl terephthalate (DEHTP) and diisononyl-1,2-cyclohexanedicarboxylate (DINCH), which may have restricted our ability to fully capture the effects of phthalate mixtures on child neurodevelopment. Additionally, exposure levels may vary across study cohorts in ways that are not fully explained by individual-level characteristics. However, we did not examine such differences, as this was beyond the scope of the present study. In addition, for participants with repeated phthalate measurements, we calculated the gestational average of urinary phthalate metabolite concentrations to better reflect overall exposure during pregnancy. However, approximately half of the study population provided only a single urine sample. Given the short biological half-lives of phthalates and the observed low ICCs for urinary phthalate measurements, this lack of repeated measurements among a large subset of participants differentially impacts our ability to characterize exposure across pregnancy. Future studies should incorporate more comprehensive repeated phthalate measurements to better characterize exposure patterns across pregnancy, as well as potentially identify sensitive windows of susceptibility to phthalate exposure. In addition, we used both creatinine-adjusted and specific-gravity–adjusted urinary concentrations when pooling exposure data across study cohorts due to limited availability of specific gravity measurements in samples collected from one cohort, which may introduce some variability. However, nine of the ten cohorts in our study measured specific gravity, and prior research has shown relatively consistent results across adjustment methods for samples collected from the same trimester of pregnancy. Although the one cohort where specific gravity was not available did collect samples across two trimesters, we do not expect this to substantially affect our findings. Additionally, we did not account for the time of day of phthalate sample collection, as a high proportion (44%) of urine samples were missing sampling time information. Sampling time may still contribute to exposure variability, and future studies should consider diurnal patterns to further improve exposure precision. We used batch-specific phthalate LODs, which varied over time for some analytes in three of the ten cohorts, to impute values for samples with non-detectable concentrations. This may have also contributed to some imprecision in the exposure measure, but given the high detection rate for most analytes and the small number of samples involved, we do not expect this to affect our findings. Moreover, the ASQ-3 is a screening tool that depends on parental report, which can lead to potential misclassification of outcomes and requires diagnostic follow-up for confirmation. Future studies with more sensitive assessment should be conducted to verify our results. Additionally, we administered age-specific ASQ-3 questionnaires, and children could vary by up to 3 months in age at the time of questionnaire administration. To reduce age-related outcome variability, we adjusted for child age as a continuous covariate in all models. However, residual age effects may remain. In addition, we included children across a broad age range (12 to 36 months) in this study. Given the rapid and varied trajectories of neurodevelopment during this period, this may have introduced additional heterogeneity to the developmental profiles. Furthermore, the limited number of repeated outcome measures constrained our ability to examine longitudinal patterns of neurodevelopment in relation to prenatal phthalate exposure and limited our ability to examine developmental differences that may emerge at other points in development. Similarly, parental abilities—such as cognitive, language, or motor skills—may influence child neurodevelopment, but we were not able to adjust for these factors as parental measures were not available for the majority of participants. Moreover, we did not adjust for multiple comparisons given the somewhat arbitrary nature and independence assumptions of such tests, and have instead evaluated trends and the magnitude of effects (Amrhein et al., 2019; Althouse, 2016; Rothman, 1990; Savitz & Olshan, 1988). Furthermore, although there is potential for variation due to heterogeneity in study design, population and protocol across cohorts, our models account for cohort-specific clustering. In addition, we used maternal education as a proxy for socioeconomic status and conducted sensitivity analyses adjusting for prenatal smoking. However, unmeasured confounding by variables such as lifestyle factors, dietary patterns, residential factors, and parental factors (e.g., alcohol or drug use, smoking, and parental mental health) cannot be completely ruled out and may still contribute to residual bias. Future research should investigate these factors alongside chemical exposures to better understand their combined impact on child neurodevelopmental outcomes. Lastly, although we included participants from multiple study sites across the United States, our results may not be fully generalizable to the overall U.S. population. Our study has several strengths. First, it utilized a large cohort with diverse demographics and locations, and the phthalate metabolite levels in the study population were comparable to or lower than those found in pregnant people of reproductive age in recent national studies in the United States (Beckingham et al., 2022). In addition, we utilized the widely recognized and validated ASQ-3, enabling comparisons across studies and providing an easily interpretable outcome with established clinical predictive value. We also identified sex-specific associations, which highlight more vulnerable populations and provide insight into biological differences in susceptibility to prenatal phthalate exposures. Additionally, we conducted sensitivity analyses, demonstrating the robustness of our results.

5. Conclusion

In conclusion, this study presents evidence that prenatal phthalate exposure may increase the likelihood of NDD in children aged 12 to 36 months, with potential differences based on child sex. Our findings are robust, supported by the inclusion of multiple cohorts with diverse exposure levels, geographic locations, and demographic backgrounds. Future studies examining a wider range of phthalates and considering postnatal exposures could provide a more thorough understanding of these effects on neurodevelopment. Additionally, incorporating repeated neurodevelopmental assessments would help identify long-term trends related to prenatal phthalate exposure.

ECHO Acknowledgements

The authors wish to thank our ECHO Colleagues; the medical, nursing, and program staff; and the children and families participating in the ECHO cohort. They also thank Diana Steele Jones, PhD, of the Duke Clinical Research Institute (DCRI), Durham, NC, who provided editorial assistance in preparing this manuscript. Dr. Steele Jones did not receive compensation for her contributions, apart from her employment at the institution in which this study was conducted.

Data Availability

Select de-identified data from the ECHO Program are available through NICHD’s Data and Specimen Hub (DASH). Information on study data not available on DASH, such as some Indigenous datasets, can be found on the ECHO study DASH webpage.