Prenatal exposure to a mixture of organophosphate esters and intelligence among 8-year-old children of the HOME Study

Abstract

Many environmental chemicals are being identified as suspected neurotoxicants based on the findings of both experimental and epidemiological studies. Organophosphate esters (OPEs), which are among the chemicals that have replaced neurotoxic polybrominated diphenyl ethers (PBDEs) after 2004, have also become an important public health topic as evidence regarding their potential for early-life neurotoxicity is growing. In 233 mother child pairs from Cincinnati, OH, we measured concentrations of the OPE metabolites bis(1,3-dichloro-2-propyl) phosphate (BDCIPP), bis-2-chloroethyl phosphate (BCEP), diphenyl phosphate (DPHP), and di-n-butyl phosphate (DNBP) in the urine of pregnant women at 16 and 26 weeks gestation and at delivery. At age 8 years, we assessed children’s cognition using the Wechsler Intelligence Scale for Children-IV. In models adjusted for maternal race, income, body mass index, and IQ, maternal urinary BCEP was associated with a modest increase in child full-scale IQ (ß: 0.81 per a ln-unit BCEP increase; 95% CI: 0.00, 1.61) while other OPEs were not associated with changes in full-scale IQ or any IQ subscales. Maternal serum PBDE concentrations did not confound the relationships between urinary OPE metabolites and child IQ. Using Bayesian kernel machine regression, we did not find that concentrations of a mixture of OPE metabolites during gestation was associated with any child cognition measures. The results of this study are not consistent with other published work, and a larger sample size would be beneficial to explore potential associations more fully. Therefore, additional studies are necessary to continue studying prenatal OPE exposure and child neurodevelopment and behavior.

1. Introduction

It has long been understood that environmental chemicals can exert neurotoxic effects and that children, whose brains are still developing, are particularly susceptible to the adverse consequences of exposures. Polybrominated diphenyl ethers (PBDEs), a class of environmentally and biologically persistent flame retardant chemicals, were phased-out of use in the United States in the mid-2000s due to concern over their developmental neurotoxicity.^1,2 Other chemicals have since increased in use as flame retardants to meet the demand left by reduced production of PBDEs, one of which being organophosphate esters (OPEs).³ OPEs, which are also used as plasticizers, can leach from the consumer products on which they are applied and deposit into dust, food, and water, where humans are exposed mainly via ingestion and dermal contact.^4–8 Once in the body, OPEs are quickly metabolized and excreted in the urine, and metabolites can be detected in up to 100% of US adults, including reproductive-aged women.^9–13

Experimental studies have shown that OPEs can cause a range of adverse effects, including neurotoxicity in zebrafish,^14,15 and altered neuronal differentiation, migration, and neurotransmitter levels in zebrafish.^15–17 In rodents, at least two studies have shown that OPEs can cause hippocampal neurotoxicity and neuroinflammation with accompanying functional deficits in the animals.^18,19 Others have reported altered exploratory behaviors in rats when the animals were exposed to OPEs during gestation and early life.²⁰ However, the literature in rodents is not consistent, with others finding no biologically relevant evidence of neurotoxicity after exposure to OPEs.²¹

Epidemiologic studies exploring the potential associations between prenatal exposure to OPEs and subsequent neurodevelopmental outcomes in humans are limited. However, OPEs have been detected in human placental tissue,²² suggesting that developing fetuses might be at risk for exposure. There have been some studies of OPEs and neurobehavior in humans, although the literature is still growing. Three studies have examined prenatal exposure to OPEs and child intelligence,^23–25 while two studies have examined prenatal exposure to OPEs and other neurobehavioral outcomes.^26,27 Four of the studies reported some degree of effect modification by child sex.^24–27 To our knowledge, there has not yet been a study of prenatal exposure to OPEs and child intelligence that has formally investigated the effects of OPEs mixtures.

In response to the call for further high-quality research into the potential links between OPE exposure and child neurodevelopment,^28–30 we investigated the associations among the participants of the Health Outcomes and Measures of the Environment (HOME) Study. The HOME Study was designed to examine the impact of exposures to common environmental chemicals in pregnant women and their children, and we collected a wide variety of perinatal, early-life, and child health outcomes in a cohort of pregnant women and their children located in Cincinnati, Ohio.³¹ The present study aims to explore associations between multiple measurements of urinary OPE metabolites during pregnancy and child cognition, including the study of mixture effects.

2. Materials and methods

2.1. Study participants

All participants are from the HOME Study, a longitudinal pregnancy and birth cohort that recruited pregnant women in Cincinnati, Ohio, between March 2003 and February 2006. Women were eligible to participate if they were 1) ≥18 years of age, 2) at 16 ±3 weeks’ gestation, and 3) living in a home built before 1978. We excluded women who were taking medication for thyroid disorders or seizures; HIV positive; had a diagnosis of bipolar disorder, schizophrenia, diabetes, or cancer that required radiation or chemotherapy; not fluent in English; or planning to move outside of the Greater Cincinnati Area. To recruit the initial cohort, 5,184 women were approached about the study via letter, of which 1,263 met the eligibility criteria and 468 enrolled. More detailed enrollment criteria are described elsewhere.³¹ HOME Study participants were eligible for this study if they remained in the study to deliver a live, singleton infant, had at least one urinary OPE measurement during pregnancy, and had child intelligence measured at either 8 or 12 years of age. All women gave informed consent for themselves and their children, and the Institutional Review Board (IRB) at Cincinnati Children’s Hospital Medical Center (CCHMC) approved the study protocol. The Centers for Disease Control and Prevention (CDC) deferred to the CCHMC IRB as the IRB of record.

2.2. Quantification of urinary OPE metabolites

Pregnant study participants provided spot urine samples at approximately 16 and 26 weeks of pregnancy and within 48 hours of delivery. Samples were collected in polypropylene specimen cups and then aliquoted and frozen at −20°C until analysis at the CDC’s National Center for Environmental Health Laboratory. The samples were analyzed for four OPE metabolites: bis(1,3-dichloro-2-propyl) phosphate (BDCIPP), a metabolite of tris(1,3-dichloro-2-propyl) phosphate; bis-2-chloroethyl phosphate (BCEP), a metabolite of tris(2-chloroethyl) phosphate; diphenyl phosphate (DPHP), a metabolite of triphenyl phosphate and several other OPEs; and di-n-butyl phosphate (DNBP), a metabolite of tri-n-butyl phosphate.

Conjugates of the OPE metabolites underwent enzymatic hydrolysis followed by automated off-line solid-phase extraction, separation via reversed-phase high-performance liquid chromatography, and detection by isotope dilution-electrospray ionization tandem mass spectrometry.^32,33 The limit of detection (LOD) for all metabolites was 0.1 μg/L. Urinary specific gravity was measured using the ATAGO PAL-10S pocket refractometer (ATAGO CO., Tokyo, Japan). More details of the analysis, including quality control procedures, are described elsewhere.¹¹

2.3. Maternal and child intelligence measurement

We assessed study participants for intelligence using the Wechsler Abbreviated Scale of Intelligence (WASI, for mothers)³⁴ or the Wechsler Intelligence Scale for Children-IV (WISC-IV, for children at age 8 years)³⁵ by research assistants trained by an experienced and reliable gold standard examiner (Yolton) and blinded to all exposure variables. Training included instruction on standard administration and scoring of the instrument, extensive practice exams, and critiques of video recorded sessions until the examiners consistently administered and scored the instrument in a valid and reliable manner. Examiners were reassessed without prior notice every six months to ensure continued accuracy in assessment. Assessments were completed in the study clinic where the environment could be standardized and distractions minimized. The WASI yields scores in Verbal, Performance, and full-scale IQ. The WISC-IV measures full-scale IQ and four index scales: verbal comprehension, perceptual reasoning, working memory, and processing speed, which are standardized to a mean score of 100 with a standard deviation of 15. A higher IQ score indicates better performance on both the WASI and the WISC-IV. A few children not assessed for intelligence at age 8 years were assessed at age 12 years using the same scale. We used WISC-IV IQ scores at either age 8 (n = 221) or 12 (n = 12) years as the outcome because of intelligence’s stability during this period.³⁶

2.4. Statistical methods

We performed all statistical analyses with R version 4.0.2.³⁷ Urinary OPE metabolite concentrations were standardized by urine specific gravity to account for urinary dilution,³⁸ and then all metabolites were ln-transformed to normalize their distributions and reduce the influence of outliers.

We chose covariates in the main analysis based on a directed acyclic graph (Supplemental Figure 1), which indicated that the minimal covariate adjustment set was: income, maternal body mass index, maternal IQ (as measured by the Wechsler Abbreviated Scale of Intelligence),³⁴ maternal race, and other environmental chemicals that are associated with both maternal urinary OPE concentrations and child IQ.

The laboratory quantification methods of 16-week maternal serum PBDEs have been previously described.³⁹ Ten serum PBDE congener concentrations (BDE-17, -28, -47, -66, -85, -99, -100, -153, -154, and -183) were log₁₀-transformed and summed (ΣPBDEs). We included maternal ΣPBDEs as a potentially confounding environmental exposure in sensitivity analyses, but not in our main models.

For OPE metabolites with greater than 10% of concentrations below the LOD,⁴⁰ which were BCEP and DNBP, we utilized a multiple imputation approach to impute the left-censored data via the Markov Chain Monte Carlo algorithm and R package mice,⁴¹ which yielded 10 imputed data sets. In imputation models, we included the full set of modeling covariates plus the primary outcome of interest, child full-scale IQ.⁴² Each data set was used independently for modeling, and then the results were pooled using Rubin’s rules.⁴³ For OPE metabolites with less than 10% of concentrations below the LOD (BDCIPP and DPHP), we replaced the concentrations with LOD/√2.⁴⁴

To model the associations between prenatal concentrations of OPE metabolites and child intelligence, we used generalized estimating equations (GEE) to account for the three repeated OPE measurements during the prenatal period.⁴⁵ Each OPE metabolite was modeled separately. We tested for a chemical-by-timepoint interaction and the majority of terms were not significant (defined as p_interaction<0.1). Subsequently, we removed the interaction term from the GEE model to estimate the average association with prenatal OPE concentrations. We also tested for model linearity using generalized additive models with smoothing terms; an ANOVA test comparing models with and without smoothing terms had a p-value of 0.33, so the associations between urinary OPE metabolites and child intelligence were considered linear.

We assessed the associations between concentrations of mixtures of the four urinary OPE metabolites and child intelligence using Bayesian kernel machine regression (BKMR) with the R package bkmr.⁴⁶ BKMR creates non-parametric exposure-response functions using 10,000 iterations of component-wise or hierarchical variable selection with kernel machine regression and allows non-linear, non-additive associations to be observed. Component-wise posterior inclusion probabilities (PIPs) are generated for each OPE metabolite and represent the likelihood of that component being selected for inclusion in each of the 10,000 model iterations. Specific gravity corrected maternal urinary OPE concentrations were averaged over the three time points and then ln-transformed and examined in association with child full-scale intelligence quotient (FSIQ) in BKMR models. For BCEP and DNBP values <LOD, we averaged multiple imputation datasets within and between timepoints prior to modeling.

Exposure to both PBDE and OPE flame retardants likely occurred with the transition from PBDEs to OPEs during the early 2000s, especially considering the biological persistence of PBDEs in the human body. To assess the possibility of maternal PBDE serum concentrations confounding the association between OPE metabolites and child IQ, given our previous finding of a relationship between maternal PBDEs and child IQ,³⁹ we performed a sensitivity analysis which included maternal ΣPBDEs as a covariate in GEE models. We also tested for effect modification by child sex by including an OPE*sex interaction term in GEE models. Finally, we performed a sensitivity analysis that includes urinary specific gravity as a covariate, in addition to standardizing OPE concentrations by specific gravity, as recommended by O’Brien et al. ⁴⁷

3. Results

The final study sample includes 233 mother-child dyads, with 60.9% of mothers being non-Hispanic white, an average of 29.1 years of age at delivery, and 73.8% having greater than a high school education (Table 1). Children were 45.5% male and had a mean FSIQ of 101.8 with a standard deviation of 16.3. Specific gravity standardized median maternal urinary BCEP, BDCIPP, DNBP, and DPHP concentrations were 0.62-0.76 μg/L, 0.66-0.87 μg/L, 0.18-0.24 μg/L, and 1.37-1.76 μg/L during the evaluated three time periods in pregnancy and at delivery, respectively (Table 2). Urinary OPE metabolite concentrations were above the LOD for 85.84 – 89.7% (BCEP), 90.56 – 97.42% (BDCIPP), 66.09 – 86.27% (DNBP), and 97.85 – 99.57% (DPHP) of samples. OPEs at 26 weeks were all moderately correlated to each other (r = 0.31-0.50). DNBP at birth was only weakly correlated with the other OPEs (r = 0.14-0.23), while BCEP, BDCIPP, and DPHP were moderately correlated with each other at birth (r = 0.33-0.47). Serum ΣPBDE concentrations were generally not correlated with urinary OPEs except a few weak correlations at specific timepoints only (BDCIPP at 16 weeks: r = 0.13; BCEP at 26 weeks: r = 0.16; and DNBP at delivery: r = −0.26; Supplemental Table 1).

Table 1:

Demographic and intelligence characteristics of the 233 HOME Study mother-child dyads included in analyses.

	N (%) or Mean (SD)†
Maternal race
Non-Hispanic white	142 (60.9)
Non-Hispanic black and others	91 (29.1)
Maternal education
High school or less	61 (26.2)
Some college	61 (26.2)
Bachelor’s	70 (30.0)
Graduate or professional	40 (17.2)
Household income (USD)
<$40,000	97 (41.6)
$40,000 - $79,999	77 (33.0)
>$80,000	59 (25.3)
Infant sex
Male	106 (45.5)
Female	127 (54.5)
Child IQ
Full-Scale IQ†	101.8 (16.3)
Verbal Comprehension Index†	99.9 (15.8)
Perceptual Reasoning Index†	105.7 (16.4)
Working Memory Index†	101.2 (15.3)
Processing Speed Index†	96.4 (14.9)
Maternal IQ†	105.4 (15.2)
Maternal BMI at 16 weeks†	27.9 (7.5)
Maternal age at delivery (years)†	29.1 (5.8)

Table 2:

Maternal urinary OPE metabolite concentrations (μg/L) corrected for urinary specific gravity.

				Quantile
OPE Metabolite	# >LOD	% <LOD	Min	25	50	75	95	Max	GM	GSD
BCEP
16 weeks	238	11.16	0.13	0.45	0.76	1.34	5.42	95.70	0.88	2.73
26 weeks	229	14.16	0.08	0.34	0.65	1.31	7.71	128.00	0.75	3.30
Delivery	215	10.30	0.08	0.35	0.62	1.27	6.24	103.74	0.77	3.10
BDCIPP
16 weeks	238	2.58	0.14	0.54	0.87	1.65	4.67	10.45	0.96	2.45
26 weeks	229	9.44	0.10	0.38	0.68	1.27	4.83	39.45	0.76	2.89
Delivery	215	5.58	0.11	0.40	0.66	1.44	6.01	62.44	0.82	2.80
DNBP
16 weeks	238	13.73	0.10	0.18	0.27	0.38	0.89	1.88	0.29	1.82
26 weeks	229	23.18	0.07	0.17	0.24	0.38	1.23	8.11	0.27	2.23
Delivery	215	33.91	0.08	0.14	0.18	0.28	1.44	3.08	0.22	2.17
DPHP
16 weeks	238	0.86	0.32	1.03	1.76	3.38	8.61	76.84	1.95	2.46
26 weeks	229	2.15	0.15	0.80	1.37	2.35	6.05	103.00	1.40	2.47
Delivery	215	0.43	0.16	0.95	1.69	3.33	7.72	806.40	1.82	2.77

The LOD was 0.1 μg/L for all metabolites. Abbreviations: LOD, limit of detection; GM, geometric mean; GSD, geometric standard deviation.

Maternal urinary OPE metabolite concentrations during pregnancy were not associated with child intelligence measures at age 8 years in fully adjusted GEE models except for BCEP and FSIQ (Table 3). One natural-log increase in maternal urinary BCEP concentrations was associated with a 0.81-point increase in child FSIQ score (95% CI: 0.00, 1.61). The unadjusted association between BCEP and FSIQ was not significant (ß: −0.21; 95% CI: −1.18, 0.75; Supplemental Table 2). Other OPEs were not associated with FSIQ or any IQ sub-scale (Table 3). Child sex did not significantly modify the association between maternal urinary OPE metabolite concentrations and child intelligence for the majority of models (19 out of 20) (Supplemental Table 3). Adding urine specific gravity as a model covariate did not alter the results (Supplemental Table 4). Likewise, the results were not changed by adding ΣPBDEs to the model (Supplemental Table 5).

Table 3:

Generalized estimating equations models of specific gravity corrected and ln-transformed maternal urinary OPE metabolites and child IQ. Models are adjusted for maternal race, maternal body mass index, maternal IQ, and household income.

	FSIQ		Verbal Comprehension		Perceptual Reasoning		Working Memory		Processing Speed
	ß	95% CI	ß	95% CI	ß	95% CI	ß	95% CI	ß	95% CI
BCEP	0.81	0.00, 1.61	0.75	−0.06, 1.56	0.47	−0.37, 1.30	0.60	−0.27, 1.47	0.70	−0.17, 1.56
BDCIPP	0.45	−0.80, 1.70	0.48	−0.72, 1.68	0.52	−0.76, 1.80	0.18	−1.24, 1.60	0.15	−1.16, 1.46
DNBP	−0.18	−1.47, 1.10	0.23	−1.04, 1.50	0.15	−1.19, 1.50	−0.77	−2.19, 0.65	−0.49	−1.85, 0.88
DPHP	0.47	−0.67, 1.61	−0.04	−1.20, 1.12	0.86	−0.33, 2.05	0.20	−0.90, 1.31	0.56	−0.74, 1.87

Abbreviations: FSIQ, Full-scale intelligence quotient; CI, confidence interval.

In component-wise BKMR, which examined pregnancy averaged maternal urinary OPE metabolites in association with child FSIQ with adjustment for covariates, PIPs for all OPEs were low (BCEP: 18.3%, BDCIPP: 5.3%, DPHP: 6.7%, and DNBP: 7.0%). PIPs below 50% generally reflect the low importance of the chemical in model iterations.⁴⁸ In univariate predictor-response functions, increased concentrations of the four OPE metabolites while holding other OPEs at their median were not associated with changes in child FSIQ (Figure 1). Bivariate response plots, which show the predictor-response functions of each OPE metabolite while holding a second OPE metabolite at its 10^th, 50^th, or 90^th percentile, show parallel functions, indicative of no multiplicative interactions between OPEs (Figure 2). The overall risk summaries for various quantiles of exposure to a mixture of OPE metabolites are all null (Supplemental Figure 2).

Figure 1:

Univariate predictor-response plots for specific gravity corrected and ln-transformed concentrations of maternal urinary OPE metabolites in association with child full-scale IQ at age 8 years. Models are adjusted for maternal race, maternal body mass index, maternal IQ, and household income. Z represents concentrations of the OPE in question while holding other OPEs at their median level. H(z) represents the function for the univariate relationship between Z and child full-scale IQ after adjustment for covariates, and the shaded area represents the 95% CI for h(z).

Figure 2:

Bivariate exposure-response plots for specific gravity corrected and ln-transformed concentrations of maternal urinary OPE metabolites in association with child full-scale IQ at age 8 years. Quantiles of Exposure2 are fixed at 0.1, 0.5, and 0.9. Models adjusted for maternal race, maternal body mass index, maternal IQ, and household income.

4. Discussion

In this study of 233 mother-child dyads in the Cincinnati, Ohio area, maternal urinary OPE metabolite concentrations during pregnancy were not associated with child intelligence at age 8 years, except for a modest positive association of BCEP with full-scale IQ in GEE models, which was robust to sensitivity analyses. However, BKMR models suggest that none of the four OPE metabolites examined were associated with child intelligence when considering the chemical mixture. In previous work within the same birth cohort, we observed that maternal urinary BCEP was associated with season and concentrations of its parent compound, TCEP, in household dust.¹¹ Further, TCEP concentrations were the most strongly associated of any OPE with maternal race, and it was the most strongly associated of any OPE with maternal race, and it was the only OPE to be associated with maternal education.⁴⁹ Given these complex relationships, it is possible that residual confounding may play a role in the slight positive association we observed between BCEP and FSIQ, especially as we did not observe any significant associations in BKMR results. Child sex did not significantly modify the observed associations, and maternal serum ΣPBDE concentrations did not confound the relationships between OPEs and child intelligence.

Others have also investigated the associations of OPE metabolites during pregnancy and child cognition with mixed results. The CHAMACOS study of a low-income Hispanic group of pregnant women in California observed 2.9- and 3.9-point decreases in FSIQ and Working Memory at age 7 years, respectively, in association with 10-fold increases in maternal urinary DPHP.²³ The PIN study of mainly non-Hispanic white and well-educated women based in North Carolina examined maternal urinary DPHP and BDCIPP during pregnancy and did not find any significant associations with child cognition at age 2-3 years.²⁵ Another study by Liu et al., based in Wuhan, China, observed significant associations between a two-fold increase in maternal urinary BDCIPP and the Mental Development Index of the Bayley Scales of Infant Development (ß = −5.75 points). However, they did not observe any significant associations with maternal urinary DPHP.²⁴ Further, the PIN study and the Liu et al. study observed some evidence of effect measure modification by child sex, but, similar to our findings, the CHAMACOS study did not. Both the CHAMACOS study and the Liu et al. study assessed OPE exposure using the molar sum of OPEs, but no study yet to our knowledge has applied any advanced mixture-modeling methods, such as BKMR, to explore associations with child neurodevelopment.

Compared to the CHAMACOS study, the PIN study, and the Liu et al. study, our study has the highest reported median DPHP concentrations (1.37 – 1.76 μg/L) and the second-highest BDCIPP concentrations (0.66 – 0.87 μg/L), behind the PIN study, and the Liu et al. study had the lowest overall BDCIPP and DPHP concentrations.^23–25 No other study reported BCEP or DNBP concentrations for comparison. The PIN study, whose results were most similar to our own, also reported the most similar population demographics, with the majority of mothers having greater than a high school education and being non-Hispanic white.

Other factors may also have contributed to the discrepancy in findings between studies. Our study was the only one based in the United States to include repeated exposure measurements during pregnancy, which likely helped reduce exposure misclassification. Upon comparison of the methods of our study with the other studies of urinary OPE metabolites during pregnancy and child IQ, only the CHAMACOS study used the same intelligence assessment tool. The PIN Study used the MacArthur-Bates Communicative Development Inventories (MBCDI) and the Mullen Scales of Early Learning (MSEL), while the Liu et al. study used the Chinese revision of the Bayley Scales of Infant Development.^24,25 Finally, the children included in our study are the oldest: CHAMACOS children were age 7 years, PIN children were age 2 to 3 years, and the Liu et al. children were age 2 years. Intelligence stabilizes as children age, with intelligence measured at age 7 years and later being more predictive of adult IQ than intelligence measured at pre-school ages.⁵⁰ Although the overall findings of this study were null, we still recommend further investigation of this topic due to the discrepant findings in the existing literature and the need to explore the epidemiology of OPEs and child intelligence in diverse populations.

Our study does have several limitations. As mentioned previously, residual confounding cannot be ruled out. Additionally, our sample size was modest after the 8-year follow-up. Study attrition contributed to the modest sample size and may have resulted in selection bias among the participants eligible for this study. However, the socio-demographics of the participants who remained in the study until the 8-year follow-up were similar to those of the original participants, suggesting that selection bias due to loss-to-follow-up is likely minimal.³¹ Additionally, our study may have limited generalizability to other populations who experience different mixtures of OPEs due to manufacturing or regulatory dissimilarities between geographic regions.

The study design and methodology also have several strengths. We obtained three measurements of maternal urinary OPEs during pregnancy, the greatest of any published study on OPE metabolites and child neurodevelopment thus far. OPEs have a short half-life in humans, which can substantially impair researchers’ ability to infer the effects of exposure when multiple measurements are not available.¹⁰ We also measured IQ in a small portion of the children at a later time period (n=12), allowing us to increase our sample without a high risk of bias, given that IQ is relatively stable over time.^36,50 While we did not observe any time period-specific effects, the short half-life of OPEs in the body necessitates frequent biomonitoring to accurately assess exposure patterns.¹⁰ Additionally, our statistical methodology was robust, employing a multiple imputation approach, GEE models for repeated exposure measurements and sensitivity analyses, and BKMR analysis to explore effects of OPE exposure mixture.

5. Conclusions

Prenatal concentrations of OPE biomarker, including BCEP, BDCIPP, DPHP, and DNBP, were not adversely associated with child cognition at 8 years of age, as measured by the WISC-IV in this study. Associations were not modified by child sex and were not confounded by maternal ΣPBDE serum concentrations. As the studies to assess the association between OPE metabolites and child intelligence are still scarce, more studies are necessary to explore this research question further and include prenatal and postnatal exposures.

Highlights.

• We measured urinary OPE metabolites during pregnancy and intelligence at 8 years.

• Maternal OPEs were not associated with child intelligence in multivariable models.

• Mixtures of OPEs were not associated with child intelligence.

• PBDEs did not confound associations, and child sex was not an effect modifier.