Racial and Ethnic Disparities in Phthalate Exposure and Preterm Birth: A Pooled Study of Sixteen U.S. Cohorts

Preterm birth is a major cause of neonatal mortality and morbidity that may perpetuate impacts on intergenerational health.¹ In the United States, preterm birth rates increased over the past several decades, reaching a peak of 10.5% during the period 2021–2022.^2,3 Furthermore, racial and ethnic disparities in preterm birth are prevalent. In 2021, the highest proportions of preterm births occurred among non-Hispanic Black (14.7%), non-Hispanic Native Hawaiian or other Pacific Islander (12.6%), non-Hispanic American Indian or Alaskan Native (12.3%), and Hispanic or Latino populations (10.2%), with a lower proportion among the non-Hispanic White (9.5%) and Asian (9.2%) populations.² Racial and ethnic disparities in preterm birth are attributable to a number of complex and interrelated factors. Structural racism is widely considered to be the primary upstream cause of racial and ethnic disparities in preterm birth and can take many forms,⁴ such as systemic barriers to economic opportunity,^5,6 residential segregation,^6,7 or increased exposure to psychosocial stressors.^8,9 For example, housing policies at the national and local levels can create residential segregation by race and ethnicity, which can have profound impacts on the place-based resources necessary to promote healthy living among residents (e.g., accessible food stores, employment opportunities, social services, and parks and recreational facilities). Collectively, these forces can shape racial and ethnic inequities in health, including preterm birth.⁶

One understudied pathway by which social and structural determinants can increase preterm birth risk is by increasing disparities in environmental exposures, including synthetic chemicals used in consumer and personal care products.^6,10 There are racial and ethnic disparities in chemical exposures from beauty products,¹¹ as well as from certain processed foods that are likely to have higher contaminant levels.¹² Prenatal exposure to synthetic chemicals, including phthalates, is increasingly considered an important risk factor for preterm birth and its racial and ethnic disparities.^13,14 Phthalates are used extensively in commercial goods, such as personal care products, food packaging materials, and medications.¹⁵ Because of their widespread use, urinary phthalate metabolites are ubiquitous in the U.S. population and may be especially concerning among pregnant individuals.^16–18 Phthalate exposure is hypothesized to be associated with a range of pregnancy complications, including preterm birth,¹⁸ by mechanisms such as dysregulation of biological processes involving inflammation, oxidative stress, endocrine activity, placental development and function, and epigenetic and transcriptomic regulations, among other interrelated and complex processes.^19–22

Nationally representative data has shown racial and ethnic differences in urinary phthalate metabolite concentrations among nonpregnant populations, with non-Hispanic Black women exhibiting the highest exposure levels.^17,23–25 Differences in phthalate exposures may arise from racial and ethnic differences in personal care product use and composition,^11,14,23,26 as well as differences in dietary exposures.^10,12,27,28 These proximate drivers of exposure have been shown to be influenced by systems of power and oppression such as structural racism.¹¹ For example, racial residential segregation impacts food landscapes and dietary behavior such as fast food consumption,²⁸ which has been associated with greater phthalate exposure.^12,27 These associations often exist independent of socioeconomic status. For example, predominately Black residential areas in New York City have higher densities of fast food restaurants than predominately White areas, and high-income Black neighborhoods have exposures similar to those of low-income Black neighborhoods.²⁸

Given the relevance of phthalate exposures to racial health equity, prior cohort studies in the United States have sought to characterize racial and ethnic differences in phthalate exposure during pregnancy but were limited by relatively small sample sizes across racial and ethnic groups.^{14,16,29–31} Additional characterization of exposure disparities in the context of pregnancy is crucial because patterns of consumer product use may change during pregnancy in ways that differ by racial or ethnic background^32,33 and because gestation may be a susceptible period for exposure.

In a pooled analysis of 16 prospective pregnancy cohorts in the United States, we found that hypothetically reducing levels of a urinary phthalate metabolite mixture was associated with fewer preterm births.¹⁸ In extending that work and using the same pooled data, our goal for the present study was to examine the role of racial and ethnic disparities in the relationship between phthalate exposure and preterm birth. First, we sought to characterize racial and ethnic disparities in urinary phthalate metabolite concentrations and examine how hypothetical interventions to remove these exposure disparities could reduce preterm births. Second, we investigated whether associations between urinary phthalate metabolite concentrations and preterm birth varied across racial and ethnic groups in a stratified analysis. This research was addressed with the understanding that co-occurring social and environmental factors, and not underlying genetic differences,⁴ influence the likelihood of phthalate exposures across racial and ethnic groups and lead to differential susceptibility to exposure effects.

In a pooled analysis of more than 6,000 pregnancies, we observed large racial and ethnic disparities in urinary phthalate metabolite concentrations. Specifically, Black and Hispanic/Latina participants had up to 148% and 94% higher average concentrations, respectively, than White participants after adjustment for covariates. Furthermore, g-computation results suggested that the probability of preterm birth among these groups would be lower if they had urinary phthalate metabolite concentrations approximately equal to those among White participants. Our prior study of overall associations showed four phthalate metabolites (MBP, MiBP, MECPP, and MCPP) were individually associated with higher odds of preterm birth.¹⁸ In the present study, we observed some evidence that these odds ratios significantly differed by participant race and ethnicity, where two of the four metabolites (MBP and MiBP) associated with preterm birth in our prior overall analysis exhibited effect estimates that were greater in magnitude for Black and/or Hispanic/Latina participants.¹⁸ These findings suggest that, in the United States, racial and ethnic disparities in phthalate exposure are important contributors to differences in preterm birth. Overall, our findings support the hypothesis that reducing the disparities in exposure to multiple phthalates would reduce preterm births among systematically marginalized racial and ethnic groups.

Preterm birth in the United States, which increased from 9.6% of pregnancies in 2015 to 10.2% in 2019, remains a leading cause of infant death and disability in U.S. children.⁵³ Preterm birth is also a major racial and ethnic health inequity in the United States, where, in 2021, the probabilities of preterm birth for non-Hispanic Black (14.7%), Hispanic (10.2%), and American Indian/Alaska Native (12.3%) pregnancies were all higher than those observed in non-Hispanic White pregnancies (9.5%).² The most cited explanations for these disparities, specifically for the largest gap observed between non-Hispanic Black and non-Hispanic White births, are socioeconomic differences, genetic variation, and smoking, but rigorous investigation has dispelled these hypotheses because racial and ethnic disparities in preterm birth still exist after accounting for these factors.^4,54 In the present analysis, we found that the racial and ethnic disparities in phthalate exposure observed may be important contributors to the parallel disparities in preterm birth. For example, we estimated that jointly reducing exposure to nine phthalates among Black participants in our study to the levels observed among White participants would result in 13% fewer preterm births among Black participants. Correspondingly, the predicted probability of preterm birth among Black participants after the hypothetical intervention more closely resembled that observed among White participants [103 (95% CI: 71, 139) vs. 93 (95% CI: 82, 105) per 1,000 live births, respectively]. Although the preintervention preterm birth probability was lower for Hispanic/Latina participants [75 (95% CI: 65, 86) per 1,000 live births] than White participants, the hypothetical intervention was still estimated to result in an additional 9% (95% CI: $-18\%$, 0.8%) decrease in preterm births. Confidence intervals for the effects of the hypothetical interventions included the null, reflecting statistical uncertainty in estimates. Nevertheless, these data are consistent with the hypothesis that reducing disparities in phthalate exposure would help to mitigate preterm birth rates among key demographic groups in the United States.

Our approach to examining the impact of hypothetical reductions in phthalate exposures on preterm birth exemplifies how we can move beyond simply examining effect modification to investigate environmental contributors to racial health inequities. We note that our study question is about the effects of phthalates, where we contrast the expected number of preterm births that we would observe under two different distributions of phthalates. This question is distinct from a mediation question, where we might ask how much of the causal effect of race is mediated by phthalates. Following previously proposed guidelines,^4,55 our study investigated effect size differences in combination with examination of exposure differences by race and ethnicity. We observed that the differences in preterm birth could be driven, in part, by disparities in phthalate exposure. Furthermore, our novel application of g-computation in this context simultaneously allowed us to: a) estimate the impact of reducing all phthalate metabolites simultaneously (i.e., the mixture effect); and b) visualize probabilities of preterm birth under a scenario where phthalate metabolite levels were equitable across racial and ethnic groups.

One of the assumptions of g-computation, for causal inference, as implemented in our approach, is that outcome models are correctly specified. In our approach, this means that the underlying logistic models are assumed to be correct, and our sensitivity analyses suggest this is appropriate, given limited effect measure modification. As Robins and Hernán note, there are no needed assumptions about the validity of the exposure models.⁵⁶ In our approach, our choice to model exposure based on observed distributions does not affect the validity of our model given the distributional goals we set are, in principle, achievable. Here, we have modeled exposures for the express purpose of quantifying an exposure disparity. Thus, our approach estimates the impact of eliminating the disparity because it is quantified by the difference in GMs of exposures. The additional assumptions necessary are standard causal assumptions, which we previously described in the context of phthalate metabolite mixtures and preterm birth.¹⁸ Perhaps most relevantly, we assume treatment variation irrelevance. For our analysis, this assumption is akin to a “no side-effects” assumption, whereby we assume that we could lower phthalate metabolite levels through an intervention that would not change other determinants of preterm birth.^57,58 This may not be met, because changing phthalate exposure is likely to result in changes to other exposures as well. This highlights the need to study true interventions in an approach with intersectional perspectives.⁵⁹ Overall, however, our hypothetical intervention analysis provides needed information in the path toward achieving health equity through reducing disparities in a ubiquitous class of chemicals with well-document disparities by race and ethnicity.

The racial and ethnic disparities in phthalate exposure we observed in our pooled study are consistent with prior evidence. A study using a large ($n=\mathrm{38,080}$) and representative cross-sectional sample of women living in the United States also observed racial and ethnic differences in urinary phthalate metabolites, including 78% higher MEP concentrations among non-Hispanic Black in comparison with non-Hispanic White participants.⁴⁷ Similar to our sensitivity analysis examining phthalate exposure disparities across educational strata as an indicator of socioeconomic status, that study found racial and ethnic disparities in phthalate metabolite concentrations existed across income levels.⁴⁷ Evidence from prospective pregnancy cohort studies have also provided evidence consistent with our results.^16,29,31 In a recent pilot study of data pooled data from nine U.S. cohorts, adjusted differences in urinary phthalate metabolite concentrations between Hispanic (any race) and non-Hispanic White participants were of similar magnitude to differences observed in our study.¹⁶ For example, concentrations of MEP displayed the largest differences, with 108% higher (95% CI: 19%, 266%) levels among Hispanic in comparison with non-Hispanic White participants.¹⁶ Unlike most previous studies, we examined racial and ethnic differences in phthalate metabolite concentrations after adjustment for covariates. We recognize that many of these factors (education, age in pregnancy, prepregnancy BMI, etc.) result, in part, from racism and potential drivers of exposure disparities. However, we sought to examine exposure disparities that persist for reasons beyond those we were able to measure. Adjustment attenuated associations, indicating that socioeconomic and other factors play a role in the exposure disparities. Of note, attenuation could also be attributable to mediation of associations by the covariates. However, even after adjustment, racial and ethnic differences in phthalate metabolite concentrations were still substantial.

The racial and ethnic disparities in phthalate metabolite levels are likely related to differences in factors such as consumer product use^60,61 and diet.^27,62 We did not have data on such factors among study participants. We expect that these differences are because of disparities in access to products and foods with lower phthalate levels, a consequence of systematic environmental differences attributable to a multitude of factors including structural racism.^11,63 Although our results cannot determine specific interventions that would equitably reduce phthalate metabolite concentrations among Black and Hispanic/Latina participants, we expect that the lower postintervention levels are feasible in society because such distributions were actually observed among White participants. In our study, the largest racial and ethnic disparities were for metabolites derived from phthalate commonly used in personal care products, including metabolites of diethyl, di-n-butyl, and diisobutyl phthalate. These parent chemicals are typically used as fragrance additives, and high levels have been detected in a range of hair products marketed to Black women.⁶⁰ The corresponding use of such products is also higher among non-Hispanic Black women in comparison with non-Hispanic White women,⁶⁴ which may translate to higher urinary phthalate metabolite concentrations.⁶⁵ Future intervention strategies should be developed using the evidence and well-established theories from the fields of environmental justice and health equity.^10,11,26 For example, racism, sexism, and classism intersect in Black hair discrimination, which penalizes Black people, especially Black women, for wearing their hair in natural styles.¹¹ In this scenario, Black women may be pressured to maintain their hair in certain styles to achieve and maintain economic opportunities, and the products used to transform and maintain those hairstyles can have higher levels of phthalates and subsequently increase exposures.^11,26 Explicit consideration of such variation in the cumulative social factors that contribute to exposure risk is critical to advance strategies to reduce and prevent exposures.^10,11

Marginalized racial and ethnic populations may also be more susceptible to phthalate exposures through diet.^12,62,66 Increasing evidence shows that consumption of ultraprocessed foods, including fast food, may increase phthalate exposures.^12,27,62 Ultraprocessed foods are ready-to-eat items that are made with minimal whole foods and provide low nutritional value,⁶⁷ and consumption may be higher among people of color.^68,69 Phthalate levels may be higher in these foods because of the materials used in processing or packaging materials.⁷⁰ Reasons for the higher ultraprocessed food consumption among these groups are multifactorial but likely involve policies on housing and food subsidies, food deserts, and employment inequities.^5–7 Historical and ongoing housing policies create residential segregation along racial and ethnic lines, which adversely impact the food landscape.^6,71 Such policies reduce access to fresh food and increase access to ultraprocessed foods,⁷² thus increasing potential dietary phthalate exposures. Further, food subsidies that keep prices low for ultraprocessed rather than fresh foods can potentiate this issue by increasing access and consumption within lower-income communities.⁶⁸

Real-world interventions to mitigate these diverse sources of exposure could take the form of behavioral interventions,^73,74 regulations, or voluntary consumer market campaigns.^75,76 At the individual level, consumers can attempt to avoid purchasing personal care or food items that may contain phthalates. However, accessibility and cost may impact the availability of “phthalate-free” products,⁷⁷ which could differentially impact marginalized groups.¹⁰ Moreover, it is difficult for consumers to identify phthalate-free goods, even with the aid of consumer guides.⁷⁵ Few consumer guides conduct independent product testing and instead rely on potentially inaccurate or nondescriptive product labels.⁷⁸ In addition, the evidence for the effectiveness of interventions on personal care and food products to reduce phthalate exposures is mixed,^{73,74,79–81} which is likely because of the multifactorial nature of environmental phthalate exposure.⁷⁵ Finally, although individual-level behavior interventions may provide some immediate reduction in phthalate exposures,⁷⁴ it is difficult to determine how to scale such interventions to benefit entire populations.

Regulations and consumer market actions are other ways to reduce disparities in phthalate metabolite concentrations at the population level.⁷⁶ Legislative options could limit the use of phthalates in personal care products, like cosmetics and other beauty products, particularly those intended for women. The U.S. Consumer Product Safety commission now limits the use of phthalates in children’s toys,⁸² but few such restrictions currently exist for products intended for women of reproductive age. However, regulations proposed by public interest groups (e.g., Breast Cancer Prevention Partners, Campaign for Safe Cosmetics) have recently made progress.⁸³ At the end of 2022, the U.S. Congress passed the Modernization of Cosmetics Regulation Act of 2022, which has provisions aimed at reducing exposure to phthalates in consumer products among the public and people working in professional salons. For example, the new law improves the ability of the U.S. Food and Drug Administration(U.S. FDA) to access cosmetic product ingredients in the event of adverse events and requires the U.S. FDA to consider international requirements regarding fragrance allergens when considering new regulations or threshold levels.⁸³ However, among other concerns, public interest groups point out that the new law does not require product ingredient transparency to the public.⁸³ In addition, although the law allows states to maintain previously existing laws banning the use of specific ingredients in cosmetic products, public interest groups have objected to the law’s preemption of states establishing new cosmetic safety regulations that exceed federal regulations.⁸³

The U.S. FDA can regulate phthalates in food and beverages. Recently, the U.S. FDA revoked authorization for 23 of 28 phthalates permitted in food contact and packaging applications.⁸⁴ Although this action may seem promising to reduce phthalate exposures in the United States, there are caveats to the apparent progress. First, the decision to revoke authorization for the 23 phthalates was based on an industry-backed petition, which stated that use of those 23 chemicals had already been discontinued.⁸⁴ In other words, the revoked authorization for these phthalates is unlikely to have additional meaningful impact on dietary exposure. Second, the U.S. FDA did not revoke authorization for the other five phthalates, including DEHP and DEHP replacements, despite petitioning from public interest groups.⁸⁴ The U.S. FDA stated that petitions to remove these additional five compounds were denied because they did not demonstrate that those phthalates were unsafe for the approved food additive uses.⁸⁴ Thus, there is still abundant opportunity for the U.S. FDA to reduce population-level phthalate exposures from food materials.

Our study had several strengths. Combining data from 16 U.S. studies resulted in a large racially and ethnically diverse study population, with large enough sample sizes for Black, Hispanic/Latina, and White groups to estimate exposure disparities and their influence on preterm birth probabilities. In addition, because we included nearly all U.S. studies published before May 2019 with data on prenatal urinary phthalate metabolite concentrations, our results may have greater generalizability than previous studies investigating exposure disparities. Last, our approach to evaluating joint reductions in phthalate metabolites reflects the reality that pregnant populations are exposed to multiple phthalates simultaneously. This approach addresses the longstanding recommendation by the National Academies for regulatory agencies to consider coexposure to multiple phthalates for more accurate cumulative risk assessments.⁸⁵ Further, our study has applied an environmental justice approach with a solution-oriented lens to understand the extent to which a highly prevalent class of chemicals with well-documented racial and ethnic disparities contributes to a persistent health disparity. We acknowledge that this study is an initial step on the path to addressing disparities in phthalate exposures and associated health impacts such as preterm birth, and we hope this provides a framework for future studies to leverage.

Our study also had several limitations. Despite our large sample size, no studies in our dataset focused on recruiting from Asian/Pacific Islander or Native American/Alaska Native populations. Our overall and preterm birth sample sizes for these groups were therefore limited and we were unable to estimate the impacts of hypothetical interventions. These groups deserve attention in future studies, especially because we found that Asian/Pacific Islander participants in our study had elevated exposures to several phthalate metabolites and higher proportions of preterm birth in comparison with White participants. In the United States, those who identify as Native American/Alaska Native are also more likely than those who identify as non-Hispanic White to deliver preterm, and to our knowledge, phthalate exposure levels in this group have not been explored.⁵³ Our sample size also restricted us to evaluating racial and ethnic categories that were broadly defined, obscuring the inherent heterogeneity that exists within each group. For example, we were unable to account for nativity status, acculturation, or immigration status, which are particularly important in the U.S. context for those identifying as Black or Hispanic.^86,87 Further, it has been established that the Asian race category in the United States is heterogenous and should be disaggregated.^88,89 Because of the limited number of harmonized covariates available across the pooled studies,¹⁸ we used education as a primary indicator of socioeconomic status, which would have been improved by including additional information (e.g., income). However, prior studies have shown that maternal education is a strong predictor of U.S. preterm birth disparities,⁹⁰ and results of our sensitivity analysis of exposure disparities across education levels were similar to prior evidence using income strata.⁴⁷

We also recognize there was variation in exposure assessment methods across studies,¹⁸ potentially producing measurement error. However, we adjust for known confounders and chose to average across repeated spot urine samples to improve our exposure characterization.^45,46 In addition, if selection into our study cohorts was strongly related to both phthalate exposure as well as race and ethnicity, that relationship may have biased our findings. However, our prior work demonstrated that associations between phthalate metabolites and preterm birth were not significantly heterogeneous across study cohorts, so this is unlikely to be a major source of bias.¹⁸ Our study may be subject to selection bias or confounding if phthalate exposure caused pregnancy to not result in a live birth.⁹¹ However, given that our central aim was to investigate associations among live births, it is unlikely to be a large source of bias. Last, our pooled analysis was specifically designed to assess the influence of phthalate exposure on preterm birth, which left us unable to address the potential influence of unmeasured environmental chemical and nonchemical exposures that may act independently or together with phthalates to influence racial and ethnic disparities in probabilities of preterm birth, such as other endocrine-disrupting chemicals,⁹² air pollution,⁹³ ambient temperature,⁹⁴ greenspace,⁹⁵ psychosocial stress,⁹⁶ employment inequities,⁹⁷ and access to care.⁹⁸ Future work that incorporates a more complete exposome approach to this research question will likely improve the ability for public health interventions to successfully reduce preterm birth across racial and ethnic groups.

In an analysis harmonizing data from nearly all U.S. cohort studies that measured urinary phthalate metabolites during pregnancy before May 2019, Black and Hispanic/Latina participants had significantly higher urinary concentrations of most phthalate metabolites in comparison with White participants. Hypothetical interventions to reduce these racial and ethnic disparities predicted reductions in preterm births. We also observed some evidence that ORs for associations between phthalate metabolites and preterm birth differed by participant race and ethnicity, where ORs were greatest in magnitude among Black or Hispanic/Latina participants for two of the four phthalate metabolites that were associated with preterm birth in the overall study population. These findings support the need to reduce phthalate exposures among systematically marginalized people.

This research was supported in part by the Intramural Research Program of the National Institutes of Health (NIH)/National Institute of Environmental Health Sciences (NIEHS, ZIAES103321). The project was also supported by the NIEHS (grants P42ES017198 to Dr. Alshawabkeh, P30ES005022 to Dr. Barrett, R21ES031231 to Dr. Bloom, P01ES009605 and R24ES028529 to Dr. Eskenazi, R01ES021369 and U24ES028529-06 to Dr. Holland, R01ES024381 to Dr. Braun, R01ES030078 to Dr. Buckley, P42ES017198 to Dr. Cordero, R01ES022934 to Dr. Dabelea, P30ES010126 and P01ES09584 to Dr. Engel, R01ES013543, R01ES014393, and R01ES08977 to Dr. Factor-Litvak, R01ES009718 to Dr. Hauser, ES013543 to Dr. Herbstman, P30ES023513 to Dr. Hertz-Picciotto, P30ES000002 for Dr. James-Todd, Z01ES103333 to Dr. Jukic, R01ES031591 and P42ES017198 to Dr. Meeker, R01ES031657 to Dr. Messerlian, P01ES022844 and R01ES017500 to Dr. Padmanabhan, T32ES007018 to Ms. Rosen, R01ES0125169 to Dr. Sathyanarayana, R21ES025551 and R24ES028533 to Dr. Schmidt, R01ES016863 and R01ES016863-02S4 to Dr. Swan, P30ES005022 to Dr. Weinberger, P01ES011261 to Dr. Lanphear), NIH (grants UH3OD023251 to Dr. Alshawabkeh, UH3OD023365 to Dr. Hertz-Picciotto, P30ES005022 to Dr. Rich, UH3OD023342 to Dr. Schmidt), U.S. Environmental Protection Agency (grants R82670901 and R827039 to Dr. Engel, R82670901 to Dr. Eskenazi), National Institute of Diabetes and Digestive and Kidney Diseases (grant R01DK076648 to Dr. Dabelea), National Cancer Institute (grant R21CA128382 to Dr. Michels), National Center for Advancing Translational Sciences (grant UL1TR001881 to Dr. Wang), and Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant R21HD058019 to Dr. Weinberger).

The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

The findings and conclusions of this report are those of the authors and do not necessarily represent the official position of the U.S. Centers for Disease Control and Prevention (U.S. CDC). Use of trade names is for identification only and does not imply endorsement by the U.S. CDC, the Public Health Service, or the U.S. Department of Health and Human Services. The analysis of deidentified specimens at the U.S. CDC was determined not to constitute engagement in human subjects research.

J.M.B. reported grants from the National Institutes of Health during the conduct of the study and served as an expert witness for plaintiffs in litigation related to perfluoroalkyl substances–contaminated drinking water for Morgan & Morgan law firm outside the submitted work. T.F.M. reported research support to their institution and equity from NxPrenatal Inc.; serving on the scientific advisory board of and equity from Mirvie Inc.; and serving on the scientific advisory board of and cash payment from Hoffmann-La Roche and Comanche Biopharma.