Prenatal and Childhood Per‐ and Polyfluoroalkyl Substance (PFAS) Exposures and Blood Pressure Trajectories From Birth to Late Adolescence in a Prospective US Prebirth Cohort

Nonstandard Abbreviations and Acronyms

Per‐ and polyfluoroalkyl substances (PFASs) are synthetic chemicals that are widely used in industry and commercial products.¹ PFASs are resistant to biodegradation and can accumulate in the environment due to their high chemical stability.² Human exposures to PFAS are widespread, as it was reported in the US National Health and Nutrition Examination Survey (NHANES) that PFASs were detectable in >98% of all participants.³ PFAS exposures have been linked with adverse health outcomes, including a higher risk of hypertension in human adults^{4, 5, 6, 7} and cardiovascular complications in pregnant people.^{8, 9, 10}

However, evidence on the associations of prenatal and childhood exposures to PFAS with blood pressure (BP) in children and adolescents is limited. BP tracks from childhood to adulthood, and higher childhood BP is associated with risk of hypertension and cardiovascular diseases in adulthood.^{11, 12} Understanding the environmental risk factors for childhood high BP may inform efforts to prevent cardiovascular diseases and thereby support policies to reduce pollutants such as PFASs. Previous studies have provided inconclusive results.^{13, 14, 15, 16, 17} Most studies measured BP at only a single time point before 12 years, making it difficult to understand the long‐term effects of PFAS exposures on BP across the early developmental stages, that is, from birth to late adolescence. Additionally, emerging evidence suggested that PFAS mixtures may show an overall association with BP, more so than exposure to individual PFASs. For example, in the Human Early Life Exposome Project, prenatal and postnatal PFAS exposures were associated with child systolic BP (SBP) when considered as a mixture¹⁷ but not individually.¹⁵ Yet other studies did not examine the effects of PFAS mixtures on childhood BP.

Considering these gaps, we used data from a prospective prebirth cohort to examine the associations between prenatal and midchildhood PFAS exposures with repeated measures and trajectories of SBP and diastolic BP (DBP) from birth to adolescence. We also used a novel mixture approach developed on the basis of the item response theory to examine the overall burden of exposure to PFAS mixtures on BP.¹⁸

Methods

The data, analytic methods, and study materials that support the findings of this study are available from Dr Marie‐France Hivert (mhivert@partners.org) upon reasonable request. Project Viva data‐sharing policies are detailed at https://www.projectviva.org/for‐investigators.

Study Population

Project Viva is a prospective prebirth cohort study. Methods for recruitment and follow‐up have been described in the cohort profile.¹⁹ Between 1999 and 2002, we recruited pregnant women at their first prenatal visit from 8 practices of the Atrius Harvard Vanguard Medical Associates. Women who had multiple gestations, could not answer questions in English, had ≥22 weeks of gestation, or planned to move away from the study area were excluded from participation. Of the 2128 live births in Project Viva, 1380 children had data on prenatal PFASs and ≥1 BP measurement between birth and late adolescence, and 639 had data on midchildhood PFASs and ≥1 BP measurement between midchildhood and late adolescence. We included 1506 unique children in at least 1 analysis (Figure S1).

The study was approved by the institutional review boards of participating institutions. Pregnant women provided written informed consent at enrollment. At follow‐up visits, mothers provided written informed consent and participating children provided either assent (<18 years) or signed consent (≥18 years). We followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline for cohort studies.²⁰

Exposures: Prenatal and Childhood PFASs

We collected maternal plasma samples at enrollment (median, 9.6 weeks of gestation; samples collected between 1999 and 2002) and child plasma samples at the midchildhood visit (median, 7.7 years; samples collected between 2007 and 2010). We stored samples in PFAS‐free cryovial tubes in liquid nitrogen freezers and shipped thawed aliquots to the Division of Laboratory Sciences at the Centers for Disease Control and Prevention for analyses. We previously published the analytical methods for maternal and child samples.^{21, 22, 23} This analysis included PFASs with detectable concentrations in >60% of samples. PFASs that met this threshold in maternal samples were (1) perfluorooctane sulfonate (PFOS), (2) perfluorooctanoate (PFOA), (3) perfluorohexane sulfonate (PFHxS), (4) perfluorononanoate (PFNA), (5) 2‐(N‐ethyl‐perfluorooctane sulfonamido) acetate (EtFOSAA), and (6) 2‐(N‐methyl‐perfluorooctane sulfonamido) acetate (MeFOSAA). PFASs that met this threshold in child samples were (1) PFOS, (2) PFOA, (3) PFHxS, (4) PFNA, (5) MeFOSAA, and (6) perfluorodecanoate (PFDA). The limit of detection (LOD) was 0.2 ng/mL for PFOS and 0.1 ng/mL for all other PFASs in maternal samples and 0.1 ng/mL for all PFASs in child samples. We imputed PFAS concentrations below the LOD with LOD/√2. We log2‐transformed PFAS concentrations in the analyses to render an approximately normal distribution.

Outcomes: Blood Pressure in Childhood and Adolescence

Trained research assistants measured BP at 6 in‐person follow‐up visits: (1) birth, (2) infancy (median, 6.3 months; interquartile range [IQR], 6.0–6.9 months), (3) early childhood (median, 3.2 years; IQR, 3.1–3.3 years), (4) midchildhood (median, 7.7 years; IQR, 7.4–8.4 years), (5) early adolescence (median, 12.9 years; IQR, 12.5–13.7 years), and (6) late adolescence (median, 17.5 years; IQR, 17.2–17.9 years). Protocols and manuals of operations for measuring BP are provided in Data S1. We used biannually calibrated oscillometric automated BP recorders (delivery and infancy visits: Dinamap 8100 [Dinamap, Tampa, FL]; childhood visits: Dinamap Pro100 or Pro200 [Dinamap, Tampa, FL]; adolescence visits: Omron HEM‐907XL [Omron, Bannockburn, IL]) to take 5 BP readings, each 1 minute apart, on the participant's upper arm using appropriate‐sized cuffs. At the delivery and infancy visits, we took BP measurements with the participants quiet or asleep and lying face up in the bassinet or reclined in a research assistant's arm. At the childhood and adolescence visits, we took BP measurements with the participants sitting quietly and with their right arm extended in a manner in which the brachial artery was perpendicular to their heart. The average of all 5 BP readings at each visit was used for analyses.

Covariates

Mothers reported their age, race, ethnicity, household income, height, prepregnancy weight, and smoking status via interviews and questionnaires at enrollment. We extracted data on parity and history of chronic hypertension from the medical records. We calculated prepregnancy body mass index (BMI; kg/m²) as prepregnancy weight (kg) divided by height (m) squared and categorized it as normal or underweight (<25 kg/m²), overweight (25 to <30 kg/m²), and obese (≥30 kg/m²).

We abstracted data on child sex and birth weight from the medical records. We calculated gestational age using established methods described previously²⁴ and derived the sex‐specific birth weight‐for‐gestational age z scores using a US national reference.²⁵ Mothers reported total breastfeeding duration at 12 months and reported children's race and ethnicity at the early childhood visit and household income at the midchildhood visit. At each follow‐up visit, we measured children's length or standing height using methods described previously.²⁶

Statistical Analysis

We used linear regression models to examine associations between individual prenatal PFASs (modeled continuously as log2‐transformed) and BP at each follow‐up visit from birth to late adolescence and between individual midchildhood PFASs (modeled continuously as log2‐transformed) and BP at the midchildhood, early adolescence, and late adolescence follow‐up visits.

We also used linear spline mixed‐effects models to examine associations between individual prenatal PFASs and repeated BP measures from birth to midchildhood and between midchildhood PFASs and repeated BP measures from midchildhood to late adolescence. In this analysis, we categorized PFAS levels into tertiles. We placed 2 knots at 6 and 13 years for the spline, which allowed us to divide the study period into 3 distinct child life stages: (1) infancy to early childhood (0 to <6 years), (2) midchildhood (6 to <13 years), and (3) adolescence (≥13 years); this classification is consistent with our previous work.^{27, 28} We predicted BP values from 0 to 20 years, with all covariates held at their mean values, and plotted the trajectories. To compare differences in BP trajectories by PFAS levels, we used Wald tests to estimate the probability that all product terms of PFAS tertiles and age splines are equal to 0, and we considered P <0.10 as evidence of differences in BP trajectories.

To evaluate PFASs as a mixture, we calculated an exposure burden score for PFAS mixtures using a novel method developed by Liu et al.¹⁸ This method uses item response theory to develop exposure burden scores for PFAS mixtures with biomonitoring data from the 2017 to 2018 NHANES as the reference and creates a composite PFAS burden score that is independent of the health outcome and comparable across studies. We used linear regression models to examine associations of prenatal and childhood PFAS exposure burden scores with BP at each subsequent follow‐up visit.

A priori, we identified confounders defined as covariates expected to be associated with the exposures and outcomes but not on the potential causal pathway. For analyses of prenatal PFASs, we adjusted for the following covariates: maternal age, race and ethnicity (non‐Hispanic White; non‐Hispanic Black; others [including Hispanic, non‐Hispanic Asian, >1 race, and other race; categories combined due to small sample sizes]), household income at the first trimester (>$70 000/year versus not), prepregnancy BMI (normal or underweight, overweight, obesity), nulliparous (yes, no), chronic hypertension (yes, no), and pregnancy smoking status (never smoker, former smoker, smoked during pregnancy). For analyses of childhood PFASs, we adjusted for breastfeeding duration at 12 months, birth weight‐for‐gestational age z scores, child race and ethnicity (same classification as maternal race and ethnicity), and household income at the midchildhood visit (>$70 000/year versus not). For all analyses, we additionally adjusted for child age, sex, and length/height at the time of BP measurement. We used the multiple imputation by chained equations method (10 imputations and each with 10 iterations) to impute missing values for covariates included in the regression models.²⁹ We investigated whether the associations of prenatal and childhood PFAS exposures with BP differed by child sex by including a product term of child sex and log2‐transformed PFASs in the multivariable models. We considered P values for the product terms <0.10 as evidence of effect modification by sex.

We conducted all analyses using Stata version 17.0 (StataCorp, College Station, TX).

Results

Descriptive Results

Table 1 provides characteristics of Project Viva participants included in the prenatal PFAS (n=1380) and midchildhood PFAS (n=639) analyses. Most mothers (>60%) were non‐Hispanic White, had prepregnancy BMI <25 kg/m², graduated from college, had a household income >$70 000/year at the first trimester and midchildhood visits, and did not smoke during pregnancy; a small proportion (2%) had chronic hypertension before pregnancy. Most children (66% for the prenatal and 59% for midchildhood PFAS analyses) were non‐Hispanic White. Participants in this study (n=1506) were comparable to those excluded from this study (n=622), except excluded mothers had higher household income at the midchildhood visit (Table S1). We provided distributions of prenatal and midchildhood PFAS levels in Table 2. The Pearson correlation coefficients for prenatal PFAS pairs ranged from 0.18 (PFNA and EtFOSAA) to 0.72 (PFOS and PFOA), and for midchildhood PFAS pairs ranged from 0.14 (PFNA and PFHxS) to 0.78 (PFOS and PFOA) (Figure S2). Prenatal and midchildhood PFASs were weakly to moderately correlated (Pearson correlation coefficients= −0.01 to 0.41) (Figure S2).

PFASs and BP at Each Follow‐Up Visit

No prenatal PFASs were consistently associated with SBP or DBP across all follow‐up visits from birth to late adolescence (Figure 1, Table S2). After multivariable adjustment, prenatal PFOS was associated with higher SBP in midchildhood and early adolescence, PFOA with higher SBP in midchildhood, EtFOSAA with higher SBP at birth and midchildhood, and MeFOSAA with higher SBP at birth (Figure 1, upper panel). Prenatal PFOS and EtFOSAA were associated with lower DBP in infancy and MeFOSAA with higher DBP in early childhood (Figure 1, lower panel). Prenatal PFHxS and PFNA were not associated with BP at any visit. The associations between prenatal PFNA with late adolescence SBP (P‐interaction=0.04) and DBP (P‐interaction=0.03) were negative in female children and positive in male children, and the association between prenatal EtFOSAA and late adolescence SBP also differed by sex (P‐interaction=0.09); all other associations did not differ by sex (Table S3).

• Download figure
• Download PowerPoint

Similarly, no midchildhood PFASs were consistently associated with SBP or DBP across follow‐up visits from midchildhood to late adolescence (Figure 2, Table S4). In the unadjusted models, midchildhood PFOS, PFOA, PFHxS, and MeFOSAA were all cross‐sectionally associated with lower midchildhood SBP, but the associations were attenuated after adjusting for confounders (Table S4). Midchildhood PFOA was also associated with lower early adolescence SBP in the unadjusted models, but this was also attenuated in the adjusted models. Midchildhood PFNA and PFDA were not associated with BP at any visit. Associations did not differ by sex (Table S5).

• Download figure
• Download PowerPoint

PFAS and BP Trajectories

SBP trajectories from 0 to 20 years did not differ by prenatal PFAS levels (Figure 3, upper panel; P=0.12–0.99). DBP trajectories from 0 to 20 years differed slightly by prenatal PFOA, PFHxS, and PFNA levels (Figure 3, lower panel; P=0.01–0.09). Higher (tertiles 2 and 3 versus tertile 1) prenatal PFOA, PFHxS, and PFNA levels were associated with a higher rate of DBP increase in infancy to early childhood (0 to <6 years) and in adolescence (>13 years).

• Download figure
• Download PowerPoint

SBP trajectories from 6 to 20 years also did not differ by midchildhood PFAS levels (Figure 4, upper panel; P=0.28–0.69). DBP trajectories from 6 to 20 years differed slightly by midchildhood PFHxS and MeFOSAA levels (Figure 4, lower panel; P=0.03–0.08). Higher PFHxS levels (tertiles 2 and 3 versus tertile 1) were associated with a lower rate of DBP increase, and higher MeFOSAA levels (tertiles 2 and 3 versus tertile 1) were associated with a higher rate of DBP increase in midchildhood (6 to <13 years) and a lower rate in adolescence (>13 years).

• Download figure
• Download PowerPoint

Overall PFAS Burden

Prenatal PFAS burden scores were not associated with SBP or DBP at any follow‐up visit in any models (Table S6). A higher midchildhood PFAS burden score was cross‐sectionally associated with lower SBP in the unadjusted models, but the association was attenuated after multivariable adjustment; midchildhood PFAS burden score was not associated with SBP or DBP at the early or late adolescence visits (Table S7).

Discussion

In this large prospective prebirth cohort from eastern Massachusetts, prenatal or midchildhood PFAS exposures were not consistently associated with SBP or DBP across all follow‐up visits from birth to adolescence. We observed that (1) specific individual prenatal PFAS exposures (eg, PFOS, PFOA, EtFOSAA, MeFOSAA) were associated with higher SBP at certain follow‐up visits, such as birth and midchildhood; (2) prenatal PFAS exposures (eg, PFOS, EtFOSAA) were associated with lower DBP in infancy; (3) certain individual prenatal PFAS exposures (ie, PFOA, PFHxS, PFNA) were associated with diverging DBP trajectories from 0 to 20 years; and (4) certain individual midchildhood PFAS exposures (ie, PFHxS, MeFOSAA) were associated with diverging trajectories of DBP from 6 to 20 years. The effect sizes in these differences in BP trajectories were small in magnitude, and no dose‐dependent associations were observed. Prenatal or midchildhood PFAS mixture burden scores did not show significant associations with subsequent SBP or DBP.

PFAS exposures have been linked to higher BP and risk of hypertension in adults and pregnant people, although findings are inconsistent. For example, Min et al⁴ found in 2934 US adults (≥20 years) from the 2003 to 2004 and 2005 to 2006 NHANES that higher serum PFOA levels were associated with higher odds of hypertension (defined as >140/90 mm Hg or with a medical diagnosis of hypertension) in a dose‐dependent fashion; those with PFOA levels at the 80th (versus 20th) percentile had 2.62 times the odds of hypertension. Other studies reported similar associations of PFAS exposures with higher SBP, DBP, and risk of hypertension in 1612 Chinese adults (aged 22–96 years),⁵ 15 786 Italian young adults (aged 20–39 years),⁶ and 1058 US midlife women (aged 42–52 years).⁷ In pregnant people, the Odense Child Cohort (Denmark; n=1436) found positive associations of early‐pregnancy serum PFOA and PFOS with higher SBP and DBP,⁸ and a cross‐sectional study of 674 women from 2 Shanghai hospitals (China) identified several cord plasma PFASs (eg, PFHxS) associated with higher risk of preeclampsia.⁹ Recently, we found in Project Viva (n=1558) that higher early‐pregnancy PFAS exposures increased mean DBP and the odds of gestational hypertension.¹⁰ However, several other studies in adults also reported a general lack of association between PFAS exposures and BP.^{30, 31, 32, 33}

Long‐chain PFASs have long half‐lives in the human body that range from a few years (eg, for PFOS, PFOA) to up to a decade (eg, for PFHxS).^{34, 35} Early‐life exposure routes include transplacental transfer, breastfeeding, and postnatal exposures to other environmental factors (eg, house dust) and consumer products that contain PFAS (eg, carpeting).^{36, 37, 38} The persistence of PFAS exposures during critical developmental periods may affect BP and BP‐related health outcomes later in life. Animal studies have demonstrated that higher prenatal PFAS exposures can lead to higher offspring BP. Rogers et al³⁹ were the first to observe that exposing pregnant Sprague‐Dawley rats to PFOS and PFNA increased offspring BP in a sex‐specific manner (ie, the effects appeared earlier in male than female offspring). Dangudubiyyam et al⁴⁰ also found in Sprague‐Dawley rats that pregnancy exposure to PFOS increased BP and impaired vascular relaxation mechanisms in the adult offspring, although no sex‐specific effects were observed.

The lack of consistent associations of prenatal and childhood PFAS exposures with child BP across time points in our study aligns with findings from a limited number of similar population studies. Geiger et al¹³ did not find cross‐sectional associations of serum PFOA or PFOS levels with SBP, DBP, or risk of hypertension (defined as age‐, height‐, and sex‐specific SBP or DBP ≥95th percentile) in 1655 adolescents aged 12 to 18 years from the 1999 to 2000 and 2003 to 2008 NHANES. Similarly, Manzano‐Salgado et al¹⁴ found no associations of maternal first‐trimester plasma PFHxS, PFOS, PFOA, or PFNA with BP (calculated as the mean of SBP and DBP) in ~1000 children aged 4 or 7 years in the INMA (Infancia y Medio Ambiente, Environment and Childhood) birth cohort in Spain. However, 2 cross‐sectional studies found positive associations for specific PFASs. In a study of 2251 adolescents aged 12 to 20 years from the 2003 to 2012 NHANES, Ma et al⁴¹ found that higher serum PFOS levels (but not perfluorooctanoate, PFHxS, or perfluorononanoate) were associated with higher DBP (but not SBP). In another study of 48 obese children aged 8 to 12 years from Dayton, Ohio, Khalil et al found that serum perfluorononanoate levels (but not PFOA, PFOS, or PFHxS) were associated with higher SBP (but not DBP).⁴² Importantly, these studies differed in research questions, with Manzano‐Salgado et al focusing on prenatal PFAS exposures¹⁴ and the other 3 investigating childhood PFAS exposures.^{13, 41, 42}

Our study adds to the limited literature on the effects of both prenatal and childhood PFAS exposures on repeated measures of BP and longitudinal BP trajectories. In the Human Early Life Exposome Project, a consortium of 6 European birth cohorts, prenatal or childhood PFAS exposures were not individually associated with BP in the environmental‐wide association analysis (aged 6–12 years; n=1277).¹⁵ In a reanalysis of the Human Early Life Exposome Project data (n=1101), Papadopoulou et al¹⁷ found that exposure to a mixture of both prenatal and childhood PFAS was associated with higher SBP (but not DBP) when the mixture levels were ≥50th percentile; however, the validity of this approach is less clear because children are not simultaneously exposed to both prenatal and postnatal PFAS in real‐life settings. As shown in our analyses, the correlations between prenatal and midchildhood PFAS levels are low (ρ=−0.01 to 0.18). In the HOME (Health Outcomes and Measures of the Environment) Study (Cincinnati, Ohio), Li et al found that pregnancy (measured at ~16 weeks and in cord blood) and childhood (aged 3, 8, and 12 years) serum PFHxS, but not PFOA, PFOS, or PFNA, was associated with higher SBP at 12 years (n=221).¹⁶ In our study, prenatal PFHxS was also marginally associated with higher SBP in infancy, midchildhood, and early adolescence; and higher DBP in midchildhood and early adolescence (Figure 1, Table S2). It is important to note that although no PFAS was consistently associated with BP at every time point in our study, and no specific time point showed all PFASs associated with higher BP, the PFAS–BP associations were generally in the positive direction (especially for prenatal PFASs). Combined with prior findings from Papadopoulou et al¹⁷ and Li et al,¹⁶ this suggests that early‐life PFAS exposures might affect BP, although the magnitudes may be small.

The mechanisms underlying the associations between early‐life PFAS exposures and offspring BP remain unclear, but possible pathways may involve oxidative stress, inflammation, and impaired vasodilation.^{16, 17} Regarding oxidative stress, a pilot study in a subset of Boston Birth Cohort participants (n=39) showed that higher cord plasma PFOS and perfluoroheptane sulfonic acid were moderately correlated with higher levels of cord plasma 8‐hydroxy‐deoxyguanosine (r=0.41–0.46),⁴³ a well‐established biomarker for oxidative stress,⁴⁴ suggesting that in utero PFAS exposures might increase fetal oxidative stress. As for inflammation, Papadopoulou et al reported that prenatal perfluorooctanoate exposure was positively associated with the proinflammatory biomarker interleukin‐1β in childhood plasma samples collected at a mean age of 8 years.¹⁷ Additionally, early‐life PFAS exposures may indirectly affect BP through child obesity and excessive adiposity, which are major risk factors for childhood hypertension.⁴⁵ PFASs can influence maternal and infant thyroid hormones (although the directions and magnitudes vary across individual PFASs),^{46, 47} and this may increase the long‐term risk of both childhood obesity and elevated BP.^{48, 49} PFAS can also bind to and activate peroxisome proliferator‐activated receptors that regulate lipid metabolism, placental functions, and fetal and postnatal growth.⁵⁰ This binding could alter developmental adipogenesis and adipocyte programming, leading to obesogenic effects.⁵¹

Distinct associations of prenatal and childhood PFAS exposures with SBP and DBP observed in our study may also reflect different underlying mechanisms. In this study, prenatal exposures to PFOS, PFOA, PFNA, EtFOSAA, and MeFOSAA were associated with (or marginally associated with) lower DBP in infancy, but these PFASs were not associated with SBP at this time point (Figure 1, Table S2). As discussed above, previous studies have also observed prenatal PFAS exposure associations specifically with SBP or DBP (but not both), although no consistent patterns were observed.^{16, 17, 41, 42} SBP and DBP may have different early‐life influences, but the mechanisms remain unclear. A recent study in the Boston Birth Cohort (n=902) identified 48 cord blood metabolites that were specifically associated with DBP in childhood and adolescence (but not with SBP) after false‐discovery‐rate adjustment, and cord metabolites consistently demonstrated stronger associations with DBP than with SBP.²⁸ In the context of this current analysis, early‐life determinants, including environmental factors such as PFASs, may differentially affect SBP and DBP, necessitating further research to elucidate the distinct biological mechanisms involved.

Our study has limitations. First, this is an observational study. Although we adjusted for a comprehensive set of confounders, residual and unmeasured confounding may still exist. Second, this study included a subset (n=1506) of all Project Viva participants (n=2128), and the samples for the prenatal PFAS (n=1380) and midchildhood PFAS (n=639) analyses were different. Yet maternal and child characteristics were comparable between these groups (Table 1, Table S1). Third, we measured maternal PFASs in early pregnancy and postnatal PFASs in midchildhood. However, the critical windows of susceptibility for PFAS exposures on BP could be in other periods (eg, late pregnancy or early infancy). Fourth, most participants are non‐Hispanic White (69%) and are from eastern Massachusetts, and the generalizability of findings to other demographic groups may be limited. Finally, false‐positive findings are possible due to multiple comparisons; however, we focused our interpretation on the overall patterns and consistency of the observed associations, rather than solely relying on statistical significance or P values.⁵²

Our study has strengths and innovations. First, we used data from Project Viva, a large pre‐birth cohort that has been continuously followed up for >20 years. The well‐maintained cohort infrastructure and the longitudinal, research‐quality BP measures allowed us, for the first time, to model BP trajectories over the first 20 years of life by prenatal and midchildhood PFAS exposures. Second, we applied a novel mixture method to estimate the overall PFAS burden on BP.¹⁸ The PFAS levels in Project Viva are comparable to those found in NHANES during the same sample periods (prenatal blood samples were collected between 1999 and 2002, and midchildhood blood samples between 2007 and 2010).^{3, 47} While the blood levels of long‐chain PFASs, particularly PFOS and PFOA, have decreased significantly since the early 2000s,⁵³ our study's PFAS burden scores were calculated by mapping our data with the 2017 to 2018 NHANES database. This approach enabled us to compare levels with other studies that collected PFAS data more recently or used the same reference database and thus increased the generalizability of our study findings.¹⁸

In short, in this large (n=1506) prospective prebirth cohort of mothers and their children from eastern Massachusetts, we found associations of prenatal and childhood PFAS exposures with BP at specific time points between birth and late adolescence but no consistent associations across all time points or PFAS types. Other studies, such as cohorts in the Environmental influences on Child Health Outcomes consortium,⁵⁴ should replicate our findings and investigate if other critical windows of susceptibility exist for PFASs on child BP.

Sources of Funding

The Project Viva study is supported by grants from the National Institutes of Health (R01HD034568, UG3OD023286, R24ES030894). Dr James‐Todd is supported by a grant from the National Institutes of Health (P30ES000002). The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders or the publishers.

Disclosures

None.

Footnotes