Prenatal phthalate exposures and adiposity outcomes trajectories: a multivariate Bayesian factor regression approach
Phuc Nguyen, Stephanie M. Engel, Amy H. Herring · 2025
Read Prenatal phthalate exposures and adiposity outcomes trajectories: a multivariate Bayesian factor regression approach by Phuc Nguyen, Stephanie M. Engel, Amy H. Herring (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-02518.txt.
Introduction
Phuc Nguyen, Stephanie Engel, and Amy Herring study whether chemical exposure before birth shapes child body fat in a 2025 paper. The chemicals are phthalates, found in many plastics, and animal work hints they may raise obesity risk. Urine tests from late pregnancy covered 382 mothers in a New York study.
Their children, 180 of them, were measured from age four to nine for weight and fat. A new Bayesian model grouped the chemicals into two latent factors and tracked effects over time by sex. In boys, exposure linked to lower fat at young ages but higher fat after age seven. In girls, no clear link appeared.
Catalog record
| CREATOR | Phuc Nguyen, Stephanie M. Engel, Amy H. Herring |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 47 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-02518
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Prenatal phthalate exposures and adiposity outcomes trajectories: a multivariate Bayesian factor regression approach Phuc Nguyen, Stephanie M. Engel, Amy H. Herring Abstract We aim to assess the longitudinal effects of prenatal exposure to phthalates on the risk of childhood obesity in children aged 4 to 7, with potential time-varying and sex-specific effects. Multiple body-composition-related outcomes, such as BMI z-score, fat mass percentage, and waist circumference, are available in the data. Existing chemical mixture analyses often look at these outcomes individually due to the limited availability of multivariate models for mixture exposures. We propose a multivariate Bayesian factor regression that handles multicollinearity in chemical exposures and borrows information across highly correlated outcomes to improve estimation efficiency. We demonstrate the proposed method’s utility through simulation studies and an analysis of data from the Mount Sinai Children’s Environmental Health Study. We find the associations between prenatal phthalate exposures and adiposity outcomes in male children to be negative at early ages but to become positive as the children get older. Key Words : Bayesian factor model, multivariate regression, longitudinal data, chemical mixtures. 1 Introduction The increasing prevalence of childhood obesity is a relevant public health concern. The 2017-2018 National Health and Nutrition Examination Surveys (NHANES) estimated that 19.3% of children aged 2-19 years have obesity and another 16.1% are overweight ( Fryar et al. 2020 ) . The prevalence of obesity among children and adolescents has increased by over threefold in the last 30 years ( Committee on Accelerating Progress in Obesity Prevention and others 2012 ) . Obese children are at greater risk for several health conditions, including obesity in adulthood, type 2 diabetes, heart disease, arthritis, various cancers, and a shorter lifespan. Studies suggest that approximately 70% of obese children face a significant risk of heart diseases in adulthood ( Gollust et al. 2013 ) . Some estimates suggest that one-third of children born today (and half of Latino and black children) are expected to develop type 2 diabetes at some point in their lives ( Narayan et al. 2003 ) . Childhood obesity also carries negative implications for mental health, as it is linked to an increased risk of being bullied ( Gollust et al. 2013 ) . Environmental chemical exposures have been found to increase the risk of obesity in children by interfering with the body’s hormones, also known as the endocrine system. Currently, over 800 chemicals are known to be endocrine-disrupting. At least 33 epidemiological studies on humans have established a connection between developmental exposure to environmental chemicals and weight gain among children during their later years ( National Academies of Sciences, Engineering, and Medicine and others 2016 , Vrijheid et al. 2020 ) . Exposures to endocrine-disrupting chemicals are also common in newborn babies and pregnant women ( National Academies of Sciences, Engineering, and Medicine and others 2016 ) . The human body is most vulnerable to the long-lasting effects of environmental chemical exposure during development in utero and early childhood ( National Academies of Sciences, Engineering, and Medicine and others 2016 ) . For example, exposures to obesogens, such as cigarette smoke, and diethylstilbestrol, during gestational or postnatal periods have been shown to associate with childhood obesity, even after the chemicals had been eliminated ( National Academies of Sciences, Engineering, and Medicine and others 2016 ) . Phthalates are a group of chemicals commonly used in consumer products, such as toys, food packaging, or cosmetics, that are known to have endocrine-disrupting properties. Exposure to phthalates is often approximated by measurement of concentrations of phthalate metabolites in urine ( Kim and Park 2014 ) . In vivo and in vitro studies suggest that exposure to phthalates, notably di-2-ethylhexyl phthalate (DEHP) metabolites, may promote obesity, and the effect may depend on gender and exposure period, with fetal development a potential critical window ( Kim and Park 2014 , National Academies of Sciences, Engineering, and Medicine and others 2016 ) . There is also a large body of literature on observational studies of phthalate exposures in humans ( Golestanzadeh et al. 2019 ) . Cross-sectional studies using NHANES have found associations, including age, gender, and racial differences in the associations, between phthalate mixtures and adiposity outcomes ( Trasande et al. 2013 , Buser et al. 2014 , Wu et al. 2020 , Berger et al. 2021 , Seo et al. 2022 ) . Because pregnancy is a vulnerable period of exposure, many prospective studies have examined how prenatal exposures to phthalates affect newborns and young children. They also observed adverse effects on adiposity outcomes and sex-specific effects, though phthalate components identified as critical differed between studies ( Buckley et al. 2016a , Vafeiadi et al. 2018 , Berger et al. 2021 ) . However, how the health effects of prenatal exposures to phthalates change as a child grows up, for instance, whether the effects dampen or intensify over time, is less well-studied. While existing approaches shed light on various aspects of the mixtures’ health effects over time, they also have unique limitations. Studies that fitted single-metabolite models with time-varying effects ( Harley et al. 2017 ) failed to account for the fact that subjects were exposed to all phthalates metabolites simultaneously. Many studies fitted multi-metabolites linear mixed-effects models with interactions between chemicals and age, or fractional polynomials of age ( Yang et al. 2018 , Ye’elah et al. 2021 ) . Linear mixed models are not well-suited to model the health effects of highly correlated phthalate mixtures. They can produce unstable estimates under minor changes to model specification in the presence of multicollinearity ( Bobb et al. 2015 ) . A common workaround is to take the molar sum of correlated groups such as DEHP ( Buckley et al. 2016b , Harley et al. 2017 , Ye’elah et al. 2021 ) , low molecular weight phthalates, or high molecular weight phthalates ( Trasande et al. 2013 , Buser et al. 2014 , Ye’elah et al. 2021 ) , or preprocess the exposures with principal component analysis (PCA) ( Maresca et al. 2016 ) . However, analyzing the molar sum implies that we believe all components within the sum are equally important to predicting the outcome. Additionally, PCA finds directions that explain the most variability in the mixtures, not necessarily those most highly associated with the outcome. It is desirable to have models that infer groups of chemicals most correlated to the outcome. Heggeseth et al. 2019 used more advanced methods, including generalized additive models (GAM), growth mixture models (GMM), and functional PCA (FPCA) with random forest (RF), to find non-linear effects of prenatal exposures on BMI trajectories up to puberty. These methods also have unique limitations for estimating time-varying effects. GAM is flexible and can model nonlinear effects and interactions at each follow-up time point, however, it does not directly model effects over time ( Heggeseth et al. 2019 ) . GMM infers latent group membership for each subject and estimates group-specific, potentially nonlinear, outcome trajectories and their relationships to exposures ( Jung and Wickrama 2008 ) . As a result, it would be overly complex if the focus is on estimating a single average relationship over time between the exposures and outcomes in the study population. Due to GMM’s