Impact of Skin Care Products on Phthalates and Phthalate Replacements in Children: the ECHO-FGS

Abstract

Background:

Phthalates and their replacements have been implicated as developmental toxicants. Young children may be exposed to phthalates/replacements when using skin care products (SCPs).

Objectives:

Our objective is to assess the associations between use of SCPs and children’s urinary phthalate/replacement metabolite concentrations.

Methods:

Children (4–8 years old) from the Environmental Influences on Child Health Outcomes-Fetal Growth Study (ECHO-FGS) cohort provided spot urine samples from 2017 to 2019, and mothers were queried about children’s SCP use in the past 24 h ($n=906$). Concentrations of 16 urinary phthalate/replacement metabolites were determined by liquid chromatography–tandem mass spectrometry ($n=630$). We used linear regression to estimate the child’s use of different SCPs as individual predictors of urinary phthalate/replacement metabolites, adjusted for urinary specific gravity, age, sex assigned at birth, body mass index, and self-reported race/ethnic identity, as well as maternal education, and season of specimen collection. We created self-organizing maps (SOM) to group children into “exposure profiles” that reflect discovered patterns of use for multiple SCPs.

Results:

Children had lotions applied (43.0%) frequently, but “2-in-1” hair-care products (7.5%), sunscreens (5.9%), and oils (4.3%) infrequently. Use of lotions was associated with 1.17-fold [95% confidence interval (CI): 1.00, 1.34] greater mono-benzyl phthalate and oils with 2.86-fold (95% CI: 1.89, 4.31) greater monoethyl phthalate (MEP), 1.43-fold (95% CI: 1.09, 1.90) greater monobutyl phthalate (MBP), and 1.40-fold (95% CI: 1.22, 1.61) greater low-molecular-weight phthalates (LMW). Use of 2-in-1 haircare products was associated with 0.84-fold (95% CI: 0.72, 0.97) and 0.78-fold (95% CI: 0.62, 0.98) lesser mono(3-carboxypropyl) phthalate (MCPP) and MBP, respectively. Child’s race/ethnic identity modified the associations of lotions with LMW, oils with MEP and LMW, sunscreen with MCPP, ointments with MEP, and hair conditioner with MCPP. SOM identified four distinct SCP-use exposure scenarios (i.e., profiles) within our population that predicted 1.09-fold (95% CI: 1.03, 1.15) greater mono-carboxy isononyl phthalate, 1.31-fold (95% CI: 0.98, 1.77) greater mono-2-ethyl-5-hydroxyhexyl terephthalate, 1.13-fold (95% CI: 0.99, 1.29) greater monoethylhexyl phthalate, and 1.04-fold (95% CI: 1.00, 1.09) greater diethylhexyl phthalate.

Discussion:

We found that reported SCP use was associated with urinary phthalate/replacement metabolites in young children. These results may inform policymakers, clinicians, and parents to help limit children’s exposure to developmental toxicants. https://doi.org/10.1289/EHP13937

Introduction

Phthalates are endocrine-disrupting chemicals,¹ and children’s exposure has been associated with differences in body composition,² neurodevelopment,^2–4 and pulmonary and immune function.^5,6 Children are exposed to phthalates and phthalate replacement compounds by using personal care products (PCPs), including applying skin care products (SCPs), as well as ingesting food and beverages contaminated by plastic packaging, inhaling and ingesting contaminated dust, and using medications and vitamins.⁷ Previous studies of exposure sources have mostly focused on adult or prenatal maternal exposure to phthalates/replacements.⁸ However, because of greater perfusion, hydration, and surface area to body mass ratio than adults, young children are more vulnerable to dermal exposures than adults.⁹ Yet, there are only a few studies available^10–12 to characterize sources of exposure to phthalates/replacements among young children in the US or how the patterns of exposure may vary by racial/ethnic identity and sex assigned at birth.

A few previous studies reported associations between greater urinary phthalate concentrations and recent use of plastics for storing and heating foods and beverages and recent use of PCPs, including SCPs, in US children.^10–12 Experimental and dosing studies demonstrated dermal phthalate transfer.^13,14 Studies of US women reported differences in the use of plastics to store/prepare food and in the use of PCPs by women with different racial and ethnic identities.^15–17 Other work has shown greater urinary phthalate concentrations among women using PCPs, including SCPs, marketed to women of color compared to those marketed to white women.^18–21 However, to our knowledge, there are no US data available to estimate the contributions of different SCPs to urinary phthalate/replacement concentrations in young children with different racial/ethnic identities and sexes. Furthermore, previous studies focused on phthalate exposure associated with use of individual PCPs, which may not inform on exposure to phthalates/replacement compounds associated with using a mixture of multiple products.²²

We used the Environmental Influences on Child Health Outcomes-Fetal Growth Study cohort (ECHO-FGS), a multicenter US-based study of children’s health outcomes and prenatal environmental exposures, to estimate sources of exposure to phthalates/replacements through use of SCPs in young children. We hypothesized a priori that more frequent use of SCPs would be associated with greater urinary phthalate/replacement metabolite concentrations that varied by the child’s race/ethnic identity and sex assigned at birth and that patterns of using multiple SCPs would be associated with different urinary phthalate/replacement metabolite concentrations. We clarify that race and ethnicity are social constructs and are used here as a proxy for individual and systemic lived experiences resulting from prior and ongoing historical processes, based on phenotype that is presumably reflected in racial and ethnic identity.^23,24

Methods

Study Population

This was a clinical center–based retrospective cohort study with multiple participating clinics from across the US. The current analysis includes 630 children. As described in a prior publication,²⁵ we reenrolled mothers and children from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) Fetal Growth Studies (FGS). That study enrolled 2,802 women aged 18 to 40 years at 12 clinical centers across the US from July 2009 to January 2013, as described in detail in a previous publication.²⁶

The ECHO-FGS cohort followed up the children of mothers enrolled at 10 of the 12 original NICHD-FGS study sites located in Alabama, California, Delaware, Illinois, New Jersey, New York, and South Carolina (Figures S1 and S2), as described previously.²⁵ Briefly, we recruited 1,116 mother–child pairs from NICHD-FGS into ECHO-FGS from May 2017 through April 2019, when children were 4–8 years of age. All NICHD-FGS mother–child pairs were eligible to enroll at the 10 participating ECHO-FGS clinical sites (which included $n=\mathrm{2,373}$ mother–child pairs). ECHO-FGS participants completed a clinical examination, provided a nonfasting spot urine sample for analysis, and the parent/guardian (usually the mother) completed study questionnaires ($n=\mathrm{1,033}$). Emails and text messages with weblinks to REDCap questionnaires were sent in advance of the clinical examination, and the parent/guardian entered the data directly into REDCap survey management software online.²⁷ Most participants (74.1%) completed the study questionnaires within 24 h before urine collection. Questionnaires queried the parent/guardian about sociodemographic factors, diet, physical activity, daycare and school history, early life history, social communication, medical history and medications, home and neighborhood environments, and behavior. The following question about SCP use was completed by 906 participants:

In the PAST 24 HOURS, please list all lotions, cosmetics, shampoos, and soaps that were applied to your child’s skin (i.e., hands, face, body, etc.) Please be specific whenever possible, list product type and brand name or generic name. State ‘None’ if you did not use any.

In total, 630 ECHO-FGS children had completed questionnaire data and a urine specimen for chemical analysis and are included in the current analysis. Children, or their parent/guardian, self-reported the child’s racial/ethnic identity as non-Hispanic black (NHB), non-Hispanic white (NHW), Hispanic, or Asian/Pacific Islander (PI). Urine was analyzed for specific gravity using a digital handheld refractometer after collection, processed, aliquoted, and frozen at $-80°\mathrm{C}$ for laboratory analysis. As we previously noted,²⁵ written informed consent was obtained from the parent or legal guardian of each enrolled child, and depending on age and state regulations, child assent was also obtained. The study was approved by a central institutional review board (IRB) at Columbia University Medical Center and the IRBs of Christiana Care Health Systems, Northwestern University, Medical University of South Carolina, University of Alabama—Birmingham, Columbia University, Long Beach Memorial Medical Center, New York Hospital—Queens, Saint Peter’s University Hospital, University of California Irvine, and Fountain Valley Hospital.

Exposure Assessment

We classified the reported products applied to children’s skin, SCPs, into 14 “SCP-type” categories. Based on the reported frequencies of use, we classified SCPs as the following: lotion, oil (primarily hair), soaps (including body wash and hand soap), sunscreen, ointments, hand sanitizer, shampoos, conditioners, “other” haircare products, 2-in-1 haircare products, 3-in-1 hair/bodycare products, lip products, deodorant, and “other” SCPs. We also classified SCPs into four “formulary-type” categories based on ingredients: phthalate-free or not phthalate-free, paraben-free or not paraben-free, medicated or not medicated, and “organic” or not organic. These four formulary-type categories were based primarily on manufacturer and labeling information augmented by SkinSAFE (https://www.SkinSafeProducts.com; HER Inc./Mayo Clinic), a publicly available database that collates ingredients for more than 40,000 commercially available SCPs.^28,29 We repeated the classification of SCP use and formulary type in a random sample of 10% of participants with 99.44% agreement, demonstrating high reliability of our SCP-categorization strategy.

We also employed the self-organizing map (SOM) algorithm as an unsupervised learning tool to group child participants into distinct SCP-use categories based on the similarities in their exposure profiles for reported use of the 18 aforementioned SCP types and formulary types. SOM is a statistical learning algorithm that applies learning strategies, similar to cluster analysis and multidimensional scaling, to identify and project multidimensional feature patterns as nodes on an organized 2-dimensional grid, often referred to as the “map.”^30,31 Nodes, which we define as “exposure profiles,” reflect patterns of the combined frequency of using different SCPs within our study population and are projected on the map. We selected the number of nodes (i.e., exposure profiles) for our map based on the intersection of the minimum Akaike’s information criterion (AIC) value and the maximum adjusted coefficient of determination (${\mathrm{R}}^2$) value as we desired a parsimonious set of distinct groupings.

Outcome Assessment

Children’s urine samples were shipped to the Children’s Health Exposure Analysis Resource (CHEAR) Hub at the University of Michigan (Ann Arbor, MI, USA) on dry ice. Urinary phthalate/replacement metabolites were measured using a method based on the US Centers for Disease Control and Prevention (CDC) method.³² Urinary metabolites were determined by liquid chromatography–tandem mass spectrometry (LC-MS/MS) using a Thermo Scientific Transcend TXII Turbulent Flow system (Thermo Fisher Scientific) interfaced with a Thermo Scientific Quantiva triple quadrupole mass spectrometer using multiple reaction monitoring in the negative mode. Two or three quality control (QC) samples from each of two 19-sample QC pools were run with each batch of 10 participant specimens. Method limits of detection (LOD) were determined as three times the standard deviation of sample blanks (Table 1). The method was validated using acceptance criteria similar to those established within the CDC method.³²

Table 1.

Distributions of specific-gravity corrected urinary phthalate/replacement metabolite concentrations (ng/mL) among children of ECHO-FGS participants, 2017–2019, overall and by child’s race/ethnic identity.

Race/ethnicity grouping	Metabolite	LOD	$n$ (%) > LOD	$\text{Mean}\pm \text{SD}$	Minimum	25th percentile	50th percentile	75th percentile	Maximum
Overall ($n=630$)a	MBzP	0.02	627 (99.5)	$11.8\pm 29$	0.36	2.26	4.99	10.20	459.00
	MCiNP	0.01	276 (43.8)	$0.27\pm 0.35$	$-0.09$	0.13	0.18	0.28	3.51
	MCiOP	0.01	601 (95.4)	$1.38\pm 5.35$	0.00	0.42	0.64	1.19	127.00
	MCOCH	0.2	474 (75.2)	$0.97\pm 2.11$	$-0.03$	0.20	0.36	0.83	30.00
	MCPP	0.05	626 (99.4)	$1.78\pm 3.71$	$-0.03$	0.67	1.07	1.74	59.50
	MECPP	0.02	630 (100)	$14.7\pm 17.7$	0.95	6.05	9.55	16.05	173.00
	MECPTP	0.2	630 (100)	$123\pm 244$	2.80	26.37	49.40	111.36	3,187.00
	MEHHP	0.2	630 (100)	$28.5\pm 47$	1.69	10.64	17.80	29.61	762.00
	MEHHTP	0.2	630 (100)	$10.3\pm 18.8$	0.22	2.33	4.40	9.10	246.00
	MEHP	1	269 (42.7)	$1.37\pm 2.38$	$-0.71$	0.41	0.81	1.63	45.20
	MEOHP	0.01	630 (100)	$6.42\pm 9.4$	0.23	2.44	4.16	6.50	131.00
	MEP	0.1	630 (100)	$81\pm 349$	1.21	9.00	18.20	43.85	5,024.00
	MHNCH	0.2	584 (92.7)	$2.15\pm 6.38$	$-0.07$	0.40	0.78	1.67	107.00
	MiBP	0.01	630 (100)	$20.1\pm 62$	0.41	6.12	10.40	18.37	1,381.00
	MiNP	0.01	9 (1.4)	$0.03\pm 0.24$	$-0.25$	0.00	0.00	0.00	5.53
	MBP	0.2	630 (100)	$18.4\pm 29.7$	0.73	7.05	12.20	20.01	428.00
	DINCHb	—	—	$0.01\pm 0.03$	0.00	0.00	0.00	0.01	0.43
	DEHPb	—	—	$0.17\pm 0.25$	0.01	0.07	0.11	0.18	3.62
	DEHTPb	—	—	$0.43\pm 0.85$	0.01	0.09	0.18	0.40	10.80
	LMWb	—	—	$0.59\pm 1.84$	0.03	0.14	0.25	0.43	26.00
	HMWb	—	—	$0.66\pm 0.91$	0.06	0.23	0.37	0.70	11.00
NHB ($n=200$)	MBzP	0.02	199 (99.5)	$17.4\pm 45.6$	0.04	3.14	7.01	15.12	459.00
	MCiNP	0.01	110 (55)	$0.34\pm 0.47$	$-0.04$	0.15	0.22	0.34	3.51
	MCiOP	0.01	193 (95)	$1.16\pm 1.26$	0.04	0.47	0.72	1.30	8.07
	MCOCH	0.2	164 (82)	$1.46\pm 2.59$	$-0.03$	0.24	0.44	1.71	17.20
	MCPP	0.05	199 (99.5)	$2.05\pm 4.62$	0.00	0.76	1.13	1.93	59.50
	MECPP	0.02	200 (100)	$15.5\pm 16.3$	0.95	6.52	10.10	17.57	103.00
	MECPTP	0.2	200 (100)	$155\pm 300$	3.80	37.56	69.10	130.24	3,187.00
	MEHHP	0.2	200 (100)	$33.2\pm 40.9$	1.69	12.40	21.40	37.46	284.00
	MEHHTP	0.2	200 (100)	$11.8\pm 18.5$	0.44	3.54	5.71	11.62	154.00
	MEHP	1	98 (49)	$1.48\pm 1.76$	$-0.24$	0.46	0.95	1.95	10.10
	MEOHP	0.01	200 (100)	$7.17\pm 8.43$	0.23	2.87	4.81	7.98	56.70
	MEP	0.1	200 (100)	$150\pm 534$	1.54	20.64	39.00	82.58	5,024.00
	MHNCH	0.2	193 (96.5)	$3.49\pm 7.76$	0.05	0.60	1.06	3.46	61.80
	MiBP	0.01	200 (100)	$17.3\pm 23.9$	0.41	6.22	9.68	17.53	225.00
	MiNP	0.01	0 (0)	$0.01\pm 0.06$	$-0.25$	0.00	0.00	0.00	0.31
	MBP	0.2	200 (100)	$23\pm 43.4$	0.73	8.08	14.10	24.68	428.00
	DINCHb	—	—	$0.02\pm 0.03$	0.00	0.00	0.01	0.02	0.24
	DEHPb	—	—	$0.19\pm 0.22$	0.01	0.08	0.13	0.22	1.51
	DEHTPb	—	—	$0.54\pm 1.03$	0.01	0.14	0.25	0.49	10.80
	LMWb	—	—	$0.96\pm 2.78$	0.05	0.20	0.35	0.68	26.00
	HMWb	—	—	$0.82\pm 1.08$	0.09	0.28	0.53	0.91	11.00
Hispanic ($n=175$)	MBzP	0.02	174 (99.4)	$12.6\pm 21.7$	0.16	2.45	5.43	11.07	148.00
	MCiNP	0.01	71 (40.6)	$0.25\pm 0.33$	$-0.02$	0.13	0.18	0.26	3.00
	MCiOP	0.01	168 (96)	$2.12\pm 9.87$	0.03	0.44	0.65	1.26	127.00
	MCOCH	0.2	137 (78.3)	$0.93\pm 2.46$	0.06	0.22	0.40	0.74	30.00
	MCPP	0.05	173 (98.9)	$1.93\pm 4.65$	$-0.03$	0.61	1.08	1.79	49.80
	MECPP	0.02	175 (100)	$17.5\pm 20.5$	3.05	6.83	10.50	19.08	173.00
	MECPTP	0.2	175 (100)	$109\pm 205$	2.80	21.94	41.50	86.82	1,618.00
	MEHHP	0.2	175 (100)	$31.7\pm 46.5$	2.30	11.57	18.70	35.51	511.00
	MEHHTP	0.2	175 (100)	$8.71\pm 14.4$	0.41	2.17	3.72	7.91	101.00
	MEHP	1	85 (45.9)	$1.64\pm 3.62$	$-0.27$	0.49	0.97	1.81	45.20
	MEOHP	0.01	175 (100)	$7.32\pm 10.2$	0.50	2.74	4.45	8.02	107.00
	MEP	0.1	175 (100)	$62.7\pm 149$	1.21	11.10	18.40	40.00	1,452.00
	MHNCH	0.2	164 (93.7)	$2.13\pm 8.27$	0.13	0.44	0.87	1.52	107.00
	MiBP	0.01	175 (100)	$22.7\pm 46.5$	1.04	6.80	11.30	19.39	415.00
	MiNP	0.01	3 (1.7)	$0.06\pm 0.43$	$-0.09$	0.00	0.00	0.00	5.53
	MBP	0.2	175 (100)	$17.1\pm 18$	1.28	7.74	13.10	20.38	151.00
	DINCHb	—	—	$0.01\pm 0.03$	0.00	0.00	0.00	0.01	0.43
	DEHPb	—	—	$0.20\pm 0.27$	0.02	0.07	0.12	0.21	2.83
	DEHTPb	—	—	$0.38\pm 0.71$	0.02	0.08	0.15	0.31	5.48
	LMWb	—	—	$0.50\pm 0.81$	0.04	0.16	0.26	0.44	7.58
	HMWb	—	—	$0.64\pm 0.78$	0.07	0.25	0.36	0.69	5.66
NHW ($n=166$)	MBzP	0.02	166 (100)	$8.01\pm 10.3$	0.21	2.21	4.15	8.39	62.90
	MCiNP	0.01	67 (40.4)	$0.23\pm 0.26$	$-0.09$	0.12	0.18	0.27	2.20
	MCiOP	0.01	157 (94.6)	$1.05\pm 1.48$	0.09	0.40	0.60	0.95	11.60
	MCOCH	0.2	119 (71.7)	$0.66\pm 1.01$	0.02	0.20	0.33	0.64	8.98
	MCPP	0.05	165 (99.4)	$1.52\pm 1.64$	0.15	0.71	1.09	1.72	14.80
	MECPP	0.02	166 (100)	$11.1\pm 15.1$	1.85	5.29	7.33	12.85	167.00
	MECPTP	0.2	166 (100)	$117\pm 251$	3.16	24.66	41.70	88.10	1,757.00
	MEHHP	0.2	166 (100)	$22.1\pm 60$	1.72	8.78	13.40	20.37	762.00
	MEHHTP	0.2	166 (100)	$11.4\pm 24.8$	0.48	2.07	3.71	9.33	246.00
	MEHP	1	41 (24.7)	$9.87\pm 1.16$	$-0.71$	0.34	0.61	0.98	10.70
	MEOHP	0.01	166 (100)	$5.07\pm 10.6$	0.38	2.19	3.15	5.14	131.00
	MEP	0.1	166 (100)	$23.3\pm 41.8$	2.10	6.83	11.50	20.52	304.00
	MHNCH	0.2	153 (92.2)	$1.27\pm 2.42$	0.09	0.35	0.60	1.18	22.10
	MiBP	0.01	166 (100)	$22\pm 107$	1.58	5.47	9.35	17.05	1,381.00
	MiNP	0.01	1 (0.6)	$0.01\pm 0.07$	$-0.13$	0.00	0.00	0.00	0.71
	MBP	0.2	166 (100)	$12.2\pm 10.3$	1.42	5.90	9.86	15.92	86.50
	DINCHb	—	—	$0.01\pm 0.01$	0.01	0.00	0.00	0.01	0.10
	DEHPb	—	—	$0.13\pm 0.29$	0.01	0.06	0.09	0.13	3.62
	DEHTPb	—	—	$0.42\pm 0.88$	0.02	0.09	0.14	0.33	6.52
	LMWb	—	—	$0.27\pm 0.56$	0.03	0.10	0.17	0.26	6.67
	HMWb	—	—	$0.59\pm 0.96$	0.07	0.21	0.31	0.55	6.59
Asian/PI ($n=89$)	MBzP	0.02	88 (98.9)	$4.39\pm 5.1$	0.15	1.65	2.53	5.28	36.50
	MCiNP	0.01	28 (31.5)	$0.22\pm 0.21$	0.04	0.12	0.16	0.24	1.22
	MCiOP	0.01	83 (93.3)	$1.03\pm 1.67$	0.00	0.35	0.55	0.98	12.80
	MCOCH	0.2	54 (60.7)	$0.56\pm 1.4$	0.01	0.13	0.24	0.58	127.00
	MCPP	0.05	89 (100)	$1.34\pm 1.42$	0.26	0.57	0.86	1.38	9.29
	MECPP	0.02	89 (100)	$14.4\pm 18.1$	2.38	7.09	10.10	15.68	147.00
	MECPTP	0.2	89 (100)	$87.3\pm 128$	3.16	19.55	40.60	99.41	783.00
	MEHHP	0.2	89 (100)	$23.4\pm 27.8$	2.94	11.47	17.30	22.75	220.00
	MEHHTP	0.2	89 (100)	$8.23\pm 13.4$	0.22	1.95	3.43	7.63	76.50
	MEHP	1	45 (50.6)	$1.53\pm 2.13$	$-0.45$	0.47	1.00	1.76	14.30
	MEOHP	0.01	89 (100)	$5.46\pm 6.75$	0.87	2.60	3.85	5.57	48.80
	MEP	0.1	89 (100)	$68.7\pm 403$	1.73	4.99	9.58	21.29	3,775.00
	MHNCH	0.2	74 (83.2)	$0.84\pm 0.97$	$-0.07$	0.25	0.55	1.11	6.01
	MiBP	0.01	89 (100)	$17.9\pm 18.5$	2.52	6.80	11.10	20.41	108.00
	MiNP	0.01	5 (5.6)	$0.07\pm 0.18$	$-0.12$	0.00	0.00	0.00	0.97
	MBP	0.2	89 (100)	$22.3\pm 32.5$	1.35	6.89	12.70	20.41	192.00
	DINCH	—	—	$0.00\pm 0.01$	0.00	0.00	0.00	0.01	0.06
	DEHPb	—	—	$0.15\pm 0.19$	0.02	0.08	0.11	0.15	1.44
	DEHTPb	—	—	$0.31\pm 0.46$	0.01	0.07	0.15	0.35	2.80
	LMWb	—	—	$0.53\pm 2.08$	0.05	0.12	0.24	0.35	19.50
	HMWb	—	—	$0.49\pm 0.50$	0.06	0.20	0.30	0.55	3.01

Note: —, no data; DEHP, diethylhexyl phthalate; DEHTP, di-2-ethylhexyl terephthalate; DINCH, di-isononyl cyclohexane-1,2-dicarboxylate; ECHO-FGS, Environmental Influences on Child Health Outcomes-Fetal Growth Study; HMW, high molecular weight; LMW, low molecular weight; LOD, limit of detection; MBzP, mono-benzyl phthalate; MCiNP, mono-carboxy isononyl phthalate; MCiOP, mono-carboxy isooctyl phthalate; MCOCH, cyclohexane-1,2-dicarboxylic acid mono carboxyisooctyl ester; MCPP, mono (3-carboxypropyl) phthalate; MECPP, mono (2-ethyl-5-carboxypentyl) phthalate; MECPTP, mono-2-ethyl-5-carboxypentyl terephthalate; MEHHP, mono (2-ethyl-5-hydroxyhexyl) phthalate; MEHHTP, mono-2-ethyl-5-hydroxyhexyl terephthalate; MEHP, mono ethyl hexyl phthalate; MEOHP, mono (2-ethyl-5-oxohexyl) phthalate; MEP, monoethyl phthalate; MHNCH, cyclohexane-1,2-dicarboxylic acid mono hydroxyisononyl ester; MiBP, mono-isobutyl phthalate; MiNP, mono-isononyl phthalate; MBP, mono-n-butyl phthalate; NHB, non-Hispanic black; NHW, non-Hispanic white; PI, Pacific Islander; SD, standard deviation.

^a The $n$ is different from the previous tables because this table includes only children with analysis of urinary phthalates/replacements.

^b Units are nmol/L.

Concentrations of 16 urinary phthalate/replacement metabolites were determined in children’s urine including: mono-benzyl phthalate (MBzP), mono-carboxy isononyl phthalate (MCiNP), mono-carboxy isooctyl phthalate (MCiOP), cyclohexane-1,2-dicarboxylic acid mono carboxyisooctyl ester (MCOCH), mono (3-carboxypropyl) phthalate (MCPP), mono-(2-ethyl-5-carboxypentyl) phthalate (MECPP), mono-2-ethyl-5-carboxypentyl terephthalate (MECPTP), mono-(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), mono-2-ethyl-5-hydroxyhexyl terephthalate (MEHHTP), monoethylhexyl phthalate (MEHP), mono-(2-ethyl-5-oxohexyl) phthalate (MEOHP), monoethyl phthalate (MEP), cyclohexane-1,2-dicarboxylic acid mono hydroxyisononyl ester (MHNCH), mono-isobutyl phthalate (MiBP), mono-isononyl phthalate (MiNP), and mono-n-butyl phthalate (MBP). We defined composite di-isononyl cyclohexane-1,2-dicarboxylate (DINCH) as the sum of MCOCH and MHNCH; composite diethylhexyl phthalate (DEHP) as the sum of urinary MECPP, MEHHP, MEHP, and MEOHP; and composite di-2-ethylhexyl terephthalate (DEHTP) as the sum of MECPTP and MEHHTP on a molar basis expressed as nanomoles per liter. We dropped one extreme outlying observation from DINCH. Low-molecular-weight phthalates (LMW) are frequently employed as solvents, emulsifiers, and fragrance carriers in PCPs, whereas high-molecular-weight phthalates (HMW) are often used as plasticizers in polyvinyl chloride (PVC) and plastic packaging.^33–36 We estimated the total urinary LMW concentration as the sum of MEP, MiBP, and MBP, and HMW as the sum of MBzP, MCiNP, MCiOP, MCPP, MECPP, MECPTP, MEHHP, MEHHTP, MEHP, MEOHP, and MiNP on a molar basis expressed as nanomoles per liter. To minimize bias, we used instrument-reported phthalate/replacement metabolite values less than the LOD read directly from the machine during data analysis.³⁷

For descriptive analysis, we standardized urinary metabolite concentrations to urine-specific gravity to account for differences in spot urine volume/dilution using $\text{Pc}=\mathrm{P}\left[\left({\text{SG}}_{\mathrm{m}}–1\right)/\left(\text{SG}–1\right)\right]$, where P was the measured urinary phthalate concentration, Pc was the SG-corrected phthalate concentration (ng/mL), ${\text{SG}}_{\mathrm{m}}$ was the median urinary SG in the study population, and SG was the observed specific gravity for an individual urine specimen.³⁸ We used the unstandardized urinary phthalate/replacement metabolite concentrations for regression analysis and included urine-specific gravity as a covariate in the models.³⁹

Statistical Analysis

We characterized the frequencies and distributions of sociodemographic factors, SCP-use variables, and urinary phthalate/replacement metabolite concentrations. We natural log transformed the urinary phthalate/replacement metabolite concentrations after adding 1.00 to normalize the distributions. We estimated pairwise Spearman correlation coefficients between the SCP-use variables.

To limit the overall number of independent hypothesis tests, we implemented a bivariate association screening procedure to inform further tests of the associations between SCP-use variables and children’s urinary phthalate/replacement metabolite concentrations in multiple regression models. We estimated partial Spearman correlation coefficients for each SCP-use and urinary phthalate/replacement metabolite pair, adjusted for urinary specific gravity (Table S1). We selected SCP-use–urinary phthalate/replacement metabolite pairs with $r>|0.07|$ (corresponding to $p$-value of $<0.10$) for further analysis using multiple linear regression models.

We estimated the association between the selected SCP-use variable–urinary phthalate/replacement metabolite pairs using individual general linear regression models, adjusted for the child’s urinary specific gravity, age (years), sex assigned at birth (female/male), race/ethnic identity (NHB, Hispanic, NHW, or Asian/PI), and body mass index (BMI, ${\mathrm{kg}/\mathrm{m}}^2$), as well as maternal education (completed secondary school yes or no) and season of urine collection (fall, winter, spring, or summer). The covariates were selected for adjustment as confounders based on hypothesized relationships with PCP use and urinary phthalate/replacement concentrations according to the literature and using a directed acyclic graph (Figure S3).^40,41 We adjusted for BMI to accommodate reported associations with urinary phthalates/replacements,⁴² some of which are moderately lipophilic, and potential differences in the volume of the product applied to the skin. We exponentiated the regression coefficients and 95% confidence limits to express the results as relative differences in the geometric mean levels of urinary phthalates/replacement metabolites. We further tested for effect modification of the SCP-phthalate/replacement associations according to child’s race/ethnic identity and sex assigned at birth by including cross-product terms with SCP-use variables and all other model covariates and interpreted the results using stratified models.⁴³ Finally, we used the four SOM-identified SCP-use categories (i.e., exposure profiles) as a categorical predictor of urinary phthalate/replacement levels, with profile 3 as the reference category.

The R (version 4.1.3; R Foundation for Statistical Computing) statistical software ecosystem was used for statistical analysis. We excluded $n=12$ missing maternal education, leaving $n=618$ in the analysis. We defined statistical significance as a $p$-value of $<0.05$ for main and a $p$-value of $<0.10$ for modification effects. We also adjusted the type 1 error rate using a Bonferroni approach based on the largest effective number of tests of each predictor in single predictor and effect modification models as $0.05/14=0.004$ and for SCP-exposure profile category models as $0.05/21=0.002$.⁴⁴

Results

ECHO-FGS Participants With and Without Urinary Phthalates/Replacements

We previously reported that women in the ECHO-FGS were more likely to identify as NHB or NHW and less likely to identify as Hispanic or Asian/PI than other participants from the original NICHD-FGS cohort.²⁵ Compared to ECHO-FGS children without analysis of urinary phthalate/replacements ($n=276$), those with urinary analysis ($n=630$) were modestly younger ($\text{mean}=6.75$ vs. 6.95 years; $p=0.01$), less likely to identify as NHW (26.35% vs. 39.13%; $p=0.002$), and more likely to identify as Asian/PI (14.12% vs. 10.87%), NHB (31.75% vs. 26.45%), or Hispanic (27.78% vs. 23.55%), although they were of similar sex assigned at birth (51.90% vs. 49.28% female; $p=0.47$). However, as shown in Table S2, reported use of SCPs was similar between ECHO-FGS children with and without urinary phthalates/replacements, with the exceptions of less frequent use of phthalate-free ($p=0.002$) and paraben-free ($p=0.04$) SCPs among those with compared to those without urinary analysis.

Distributions of Sociodemographic Factors

The distributions of sociodemographic factors among children of ECHO-FGS participants ($n=906$) are described in Table 2, overall and according to child’s race/ethnic identity. Children were 4–8 years of age, approximately half male (51%), and identified as NHB (30%), NHW (30%), Hispanic (26%), or Asian/PI (13%). Children in the NHW group were older ($\text{mean}=7.48$ vs. 6.75 years overall) and children in the Asian/PI group were younger ($\text{mean}=6.15$ years) than other children. Children in the Hispanic group had greater BMI ($\text{mean}=17.70$ vs. $16{.91\mathrm{kg}/\mathrm{m}}^2$ overall), and children in the Asian/PI group had lesser BMI ($\text{mean}=15{.94\mathrm{kg}/\mathrm{m}}^2$) than other children. Maternal education and household income also varied according to racial/ethnic identity. Mothers of children in the NHB group had the highest frequency of low income (44.3% vs. 22.0% overall), whereas mothers in the NHW group had the highest frequency of high income (75.7% with high income vs. 40.2% overall). We found heterogeneity in the time/season of urine collection/study completion associated with child’s race/ethnic identity, likely due in part to differences in the source populations of children at the different study enrollment sites. The distributions of sociodemographic factors were similar between male and female children (Table S3).

Table 2.

Distributions of sociodemographic factors among children of ECHO-FGS participants, enrolled 2017–2019, overall and by child’s race/ethnic identity.

Characteristic	Overall ($n=906$)	NHB ($n=273$)	Hispanic ($n=240$)	NHW ($n=274$)	Asian/PI ($n=119$)	$p$-Valuea
Age (years) ($\text{mean}\pm \text{SD}$)	$6.75\pm 1.00$	$6.30\pm 0.87$	$6.89\pm 0.93$	$7.48\pm 0.70$	$6.15\pm 0.96$	$<0.0001$
Sex assigned at birth [$n$ (%)]
Female	443 (48.9%)	134 (49.1%)	124 (51.7%)	130 (47.4%)	55 (46.2%)	0.73
Male	463 (51.1%)	139 (50.9%)	116 (48.3%)	144 (52.6%)	64 (53.7%)	—
BMI (${\mathrm{kg}/\mathrm{m}}^2$) ($\text{mean}\pm \text{SD}$)	16.91 (2.65)	16.78 (2.74)	17.70 (3.18)	16.76 (2.01)	15.94 (1.81)	$<0.0001$
Maternal education [$n$ (%)]b
Less than or up to high school	169 (19.0%)	83 (26.5.4%)	88 (32.2%)	9 (2.9%)	16 (13.7%)	$<0.0001$
More than high school	719 (80.97%)	273 (73.5%)	240 (67.8%)	274 (97.1%)	119 (86.3%)	—
Household income [$n$ (%)]c
High	327 (40.2%)	34 (13.9%)	41 (19.4%)	193 (75.7%)	59 (57.3%)	$<0.0001$
Middle	307 (37.8%)	102 (41.8%)	112 (53.1%)	59 (23.1%)	34 (33.0%)	—
Low	179 (22.0%)	108 (44.3%)	58 (27.5%)	3 (1.2%)	10 (9.7%)	—
Season of specimen collection [$n$ (%)]d
Spring	239 (37.9%)	64 (32.0%)	77 (44.0%)	47 (28.3%)	51 (57.3%)	0.0002
Summer	188 (29.8%)	67 (33.5%)	41 (23.4%)	57 (34.3%)	23 (25.8%)	—
Winter	132 (21.0%)	42 (21.0%)	35 (20.0%)	43 (25.9%)	12 (13.5%)	—
Autumn	71 (11.3%)	27 (13.5%)	22 (12.6%)	19 (11.4%)	3 (3.4%)	—
Study site (location) [$n$ (%)]
Christiana Care Health Systems (Delaware)	248 (27.4%)	93 (34.1%)	48 (20.0%)	91 (33.2%)	15 (13.4%)	0.0005
Northwestern University (Illinois)	145 (16.0%)	20 (7.3%)	24 (10.0%)	77 (28.1%)	12 (20.29%)	—
Medical University of South Carolina (South Carolina)	139 (15.3%)	39 (14.3%)	6 (2.5%)	88 (32.1%)	2 (30.3%)	—
University of Alabama—Birmingham (Alabama)	116 (12.8%)	105 (038.5%)	10 (4.2%)	1 (0.4%)	0 (0.0%)	—
Columbia University (New York)	103 (11.4%)	7 (2.6%)	81 (33.8%)	4 (1.5%)	4 (9.2%)	—
Long Beach Memorial Medical Center (California)	54 (6.0%)	7 (2.6%)	54 (20.4%)	7 (2.6%)	14 (12.6%)	—
New York Hospital—Queens (New York)	40 (4.4%)	2 (0.7%)	2 (0.8%)	0 (0.0%)	34 (38.2%)	—
Saint Peter’s University Hospital (New Jersey)	37 (4.1%)	0 (0.0%)	28 (11.7%)	6 (2.2%)	2 (2.5%)	—
University of California—Irvine/FVH (California)	24 (2.6%)	0 (0.0%)	16 (6.7%)	0 (0.0%)	6 (6.7%)	—

Note: Proportions may not sum to 100% due to rounding error. —, no data; BMI, body mass index; ECHO-FGS, Environmental Influences on Child Health Outcomes-Fetal Growth Study; FVH, Fountain Valley Hospital; NHB, non-Hispanic black; NHW, non-Hispanic white; PI, Pacific Islander; SD, standard deviation.

^a $p$-Value for difference between child’s race/ethnic identity using ${\mathrm{\chi }}^2\text{-test}$ for categorical variables and analysis of variance for continuous variables.

^b $n=18$ missing.

^c $n=93$ missing.

^d$n=276$ missing urine specimen.

Distributions of SCPs Used

The frequencies of SCPs reported used in the past 24 h are presented in Table 3, overall and according to children’s race/ethnic identity. Soaps (including body wash and hand soap) were used most frequently, as 68.5% of children used at least one soap product, although were used by fewer children in the Hispanic (61.3%) and Asian/PI (58.6%) groups. Lotions were also used frequently overall (46.0%), with at least one lotion employed by 66.6% of children in the NHB group but only 20.8% of children in the NHW group. Shampoo (36.3%) and hair conditioner (16.7%) were used by a substantial number of children, whereas fewer children used ointments (8.7%), 2-in-1 haircare products (7.5%), sunscreen (5.9%), oils (4.3%), deodorant (3.9%), other haircare products (3.8%), other SCPs (3.8%), lip products (3.6%), hand sanitizer (2.4%), and 3-in-1 hair/body care products (1.8%).

Table 3.

Frequencies of SCP use among children of ECHO-FGS participants, 2017–2019, overall and by child’s race/ethnic identity ($n=906$).

SCP	$n$ (%) Overall ($n=906$)	$n$ (%) NHB ($n=273$)	$n$ (%) Hispanic ($n=240$)	$n$ (%) NHW ($n=274$)	$n$ (%) Asian/PI ($n=119$)	$p$-Valuea
Lotions	—	—	—	—	—	0.0005
0	489 (60)	91 (33.3)	110 (45.8)	217 (79.2)	71 (59.7)	—
1	390 (43)	165 (60.4)	123 (51.3)	55 (20.1)	47 (39.5)	—
$2+$	27 (3)	17 (6.2)	7 (2.9)	2 (0.7)	1 (0.8)	—
Oils	—	—	—	—	—	$<0.0001$
0	861 (95.7)	245 (89.7)	231 (96.3)	267 (97.4)	118 (99.2)	—
$1+$	45 (4.3)	28 (10.3)	9 (3.8)	7 (2.6)	1 (0.8)	—
Soaps (including body wash and hand soap)	—	—	—	—	—	$<0.0001$
0	285 (31.5)	57 (20.9)	93 (38.8)	86 (31.4)	49 (42.2)	—
1	531 (58.6)	202 (74)	131 (54.6)	143 (52.2)	55 (46.2)	—
$2+$	90 (9.9)	14 (5.1)	16 (6.7)	45 (16.4)	15 (12.4)	—
Sunscreen	—	—	—	—	—	$<0.0001$
0	841 (93.7)	267 (97.8)	230 (95.8)	240 (87.6)	104 (87.4)	—
$1+$	65 (5.9)	6 (2.2)	10 (4.2)	34 (12.04)	15 (12.6)	—
Ointments	—	—	—	—	—	$<0.0001$
0	827 (91.3)	231 (84.6)	224 (93.3)	264 (96.4)	108 (90.8)	—
$1+$	79 (8.7)	42 (15.4)	16 (6.7)	10 (3.6)	11 (9.2)	—
Hand sanitizer	—	—	—	—	—	0.20
0	884 (97.6)	269 (98.5)	236 (98.3)	263 (96)	116 (97.5)	—
1	22 (2.4)	4 (1.5)	4 (1.7)	11 (4)	3 (2.5)	—
Shampoo	—	—	—	—	—	$<0.0001$
0	577 (63.7)	246 (90.1)	137 (57.1)	125 (45.6)	69 (58)	—
$1+$	329 (36.3)	27 (9.9)	103 (42.9)	149 (54.4)	50 (42)	—
Conditioner	—	—	—	—	—	$<0.0001$
0	755 (83.3)	263 (96.3)	205 (85.4)	183 (66.8)	104 (87.4)	—
$1+$	151 (16.7)	10 (3.7)	35 (14.6)	91 (33.2)	15 (12.6)	—
Other hair products	—	—	—	—	—	0.06
0	872 (96.2)	267 (97.8)	234 (97.5)	257 (93.8)	114 (95.8)	—
$1+$	34 (3.8)	6 (2.2)	6 (2.5)	17 (6.2)	5 (4.2)	—
2-in-1	—	—	—	—	—	$<0.0001$
0	838 (92.5)	272 (99.6)	226 (94.2)	239 (87.2)	101 (84.9)	—
$1+$	68 (7.5)	1 (0.4)	14 (5.8)	35 (12.8)	18 (15.1)	—
3-in-1	—	—	—	—	—	0.0005
0	890 (98.2)	273 (100)	236 (98.3)	262 (95.6)	119 (100)	—
1	16 (1.8)	0 (0)	4 (1.7)	12 (4.4)	0 (0)	—
Lip products	—	—	—	—	—	0.36
0	873 (96.4)	259 (94.9)	235 (97.9)	264 (96.4)	115 (96.6)	—
1	33 (3.6)	14 (5.1)	5 (2.1)	10 (3.6)	4 (3.4)	—
Deodorant	—	—	—	—	—	0.21
0	871 (96.1)	259 (94.9)	229 (95.4)	265 (96.7)	118 (99.2)	—
$1+$	35 (3.9)	14 (5.1)	11 (4.6)	9 (3.3)	1 (0.8)	—
Other SCPs	—	—	—	—	—	0.06
0	872 (96.2)	267 (97.8)	234 (97.5)	257 (93.8)	114 (57.8)	—
$1+$	34 (3.8)	6 (2.2)	6 (2.5)	17 (6.2)	5 (4.2)	—
Organic SCPs	—	—	—	—	—	0.0004
0	751 (82.9)	239 (87.5)	223 (92.9)	195 (71.2)	94 (79)	—
1	120 (13.2)	33 (12.1)	15 (6.3)	56 (20.4)	16 (13.4)	—
$2+$	35 (3.9)	1 (0.4)	2 (0.8)	23 (8.4)	9 (7.6)	—
Phthalate-free SCPs	—	—	—	—	—	0.0004
0	234 (25.8)	62 (22.7)	71 (29.6)	58 (21.2)	43 (36.1)	—
1	307 (33.9)	105 (38.5)	83 (34.6)	86 (31.4)	33 (27.7)	—
2	240 (26.5)	86 (31.5)	50 (20.8)	78 (28.5)	26 (21.8)	—
3	89 (9.8)	19 (7)	28 (11.7)	30 (10.9)	12 (10.1)	—
$4+$	36 (4)	1 (0.4)	8 (3.3)	22 (7)	5 (4.2)	—
Paraben-free SCPs	—	—	—	—	—	$<0.0001$
0	188 (20.8)	48 (17.6)	59 (24.6)	44 (16.1)	37 (31.1)	—
1	259 (28.6)	106 (38.8)	62 (25.8)	64 (23.4)	27 (22.7)	—
2	236 (26)	86 (31.5)	65 (27.1)	65 (23.7)	20 (16.8)	—
3	129 (14.2)	23 (8.4)	38 (15.8)	47 (17.2)	21 (17.6)	—
$4+$	94 (10.4)	10 (3.7)	16 (6.7)	64 (19.7)	14 (11.8)	—
Medicated SCPs	—	—	—	—	—	0.005
0	856 (94.5)	256 (93.8)	233 (97.1)	262 (95.6)	105 (88.2)	—
1	50 (5.5)	17 (6.2)	7 (2.9)	12 (4.4)	14 (11.8)	—

Note: Proportions may not sum to 100% due to rounding error. —, no data; ECHO-FGS, Environmental Influences on Child Health Outcomes-Fetal Growth Study; NHB, non-Hispanic black; NHW, non-Hispanic white; PI, Pacific Islander; SCP, skin care product.

^a $p$-Value for difference between child’s race/ethnic identity using ${\mathrm{\chi }}^2\text{-test}$.

We found that relatively few children used SCPs labeled as organic (17.1%) and medicated (5.5%), but most children had used at least one SCP labeled as phthalate-free (74.2%) or paraben-free (79.2%). However, children in the NHB group used the fewest numbers of phthalate-free and paraben-free SCPs (7.4% and 12.1% used at least three phthalate-free and paraben-free SCPs, respectively), whereas children in the NHW group used the most phthalate-free and paraben-free SCPs (17.9% and 36.9% used at least three phthalate-free and paraben-free SCPs, respectively).

We also found differences in SCP use according to child sex assigned at birth, with significantly greater use of 2-in-1 haircare products among males than females but significantly greater use of lotions, hair conditioners, phthalate-free SCPs, and paraben-free SCPs among females than males (Table S4). Most SCP variables were correlated with one another, albeit weakly (Table S5). However, we found moderate to strong positive correlations for reported use of shampoo and conditioner and for use of phthalate-free and paraben-free SCPs with lotions, soaps (including body wash and hand soap), shampoo, and conditioner, and for medicated SCPs with ointments.

Distributions of Urinary Phthalate/Replacement Metabolites

The distributions of children’s urinary phthalate/replacement metabolite concentrations are provided in Table 1, overall and according to race/ethnic identity. Most urinary phthalate/replacement metabolites were detected frequently ($>50\%$), although we detected MCiNP (43.8%), MEHP (42.7%), and MiNP (1.4%) infrequently. We measured the greatest median overall concentrations for MECPTP ($49.40\text{\hspace{0.17em}ng}/\mathrm{mL}$), MEP ($18.20\text{\hspace{0.17em}ng}/\mathrm{mL}$), MEHHP ($17.80\text{\hspace{0.17em}ng}/\mathrm{mL}$), MBP ($12.20\text{\hspace{0.17em}ng}/\mathrm{mL}$), and MiBP ($10.40\text{\hspace{0.17em}ng}/\mathrm{mL}$). Children in the NHB group had greater median urinary MBzP ($7.01\text{\hspace{0.17em}ng}/\mathrm{mL}$), MECPTP ($69.10\text{\hspace{0.17em}ng}/\mathrm{mL}$), MEP ($39.00\text{\hspace{0.17em}ng}/\mathrm{mL}$), MEHHP ($21.40\text{\hspace{0.17em}ng}/\mathrm{mL}$), and MBP ($14.10\text{\hspace{0.17em}ng}/\mathrm{mL}$) than others, whereas children in the Hispanic ($11.30\text{\hspace{0.17em}ng}/\mathrm{mL}$) and Asian/PI ($11.10\text{\hspace{0.17em}ng}/\mathrm{mL}$) groups had higher median urinary MiBP than others. Children in the NHB group also had higher median urinary composite DEHTP (0.25 vs. $0.18\text{\hspace{0.17em}nmol}/\mathrm{L}$ overall), LMW (0.35 vs. $0.25\text{\hspace{0.17em}nmol}/\mathrm{L}$ overall), and HMW (0.53 vs. $0.37\text{\hspace{0.17em}nmol}/\mathrm{L}$ overall) metabolites, whereas the sum of urinary DEHP metabolites was lower among the children in the NHW group than for the other racial/ethnic identity groups (0.09 vs. $0.11\text{\hspace{0.17em}nmol}/\mathrm{L}$ overall). Median urinary phthalate/replacement concentrations were similar between children assigned male and female sex at birth, with similar variabilities suggested by the differences between the 75th percentiles and 25th percentiles of the concentrations (Table S6).

Associations Between Individual SCPs and Urinary Phthalate/Replacement Metabolites

Table 4 describes the relative differences in children’s geometric mean urinary phthalate/replacement metabolite concentrations as dependent variables per unit difference in the selected SCP-use variables as independent variables (units for each SCP are described in Table 3). Use of body lotion was associated with a 1.17-fold [95% confidence interval (CI): 1.00, 1.34] higher geometric mean urinary MBzP concentration, and use of oil was associated with 2.86-fold (95% CI: 1.89, 4.31), 1.43-fold (95% CI: 1.09, 1.90), and 1.40-fold (95% CI: 1.22, 1.61) greater geometric mean urinary MEP, MBP, and LMW concentrations, respectively. In contrast, use of 2-in-1 haircare products was associated with 0.84-fold (95% CI: 0.72, 0.97) and 0.78-fold (95% CI: 0.62, 0.98) lesser geometric mean urinary MCPP and MBP concentrations, respectively. Urinary phthalate/replacement concentrations in SCP-users were similar between those who used at least one phthalate-free SCP and those who did not (Table S7). Similarly, the SCP formulary-type variables were not associated with urinary phthalate/replacement metabolite levels when adjusted for confounding in linear regression models (Table S8).

Table 4.

Associations using linear regression between use of individual SCPs as independent variables and select natural log-transformed urinary phthalate/replacement metabolite concentrations as dependent variables among children of ECHO-FGS participants, adjusted for confounding factors ($n=618$).

SCP/Metabolite	Ratio of geometric means (95% CI)	$p$-Value	SCP/metabolite	Ratio of geometric means (95% CI)	$p$-Value
Lotion	—	—	Shampoo	—	—
MBzP	1.17 (1.00, 1.34)	0.03	MECPP	0.98 (0.87, 1.10)	0.59
MEP	1.05 (0.99, 1.23)	0.56	MEHHP	0.96 (0.84, 1.10)	0.54
MHNCH	0.96 (0.87, 1.05)	0.35	MEHP	1.01 (0.92, 1.09)	0.88
MBP	1.04 (0.94, 1.16)	0.43	MEOHP	0.96 (0.86, 1.08)	0.51
DINCH	1.00 (0.99, 1.00)	0.24	MHNCH	0.98 (0.88, 1.09)	0.71
LMW	1.00 (0.95, 1.06)	0.96	DINCH	1.00 (1.00, 1.01)	0.87
Oils	—	—	DEHP	1.01 (0.99, 1.04)	0.33
MBzP	1.16 (0.81, 1.67)	0.43	Conditioner	—	—
MEP	2.86 (1.89, 4.31)	$<0.0001$	MCPP	1.06 (0.95, 1.19)	0.27
MBP	1.43 (1.09, 1.90)	0.01	2-in-1	—	—
LMW	1.40 (1.22, 1.61)	$<0.0001$	MBzP	0.79 (0.60, 1.06)	0.12
Soaps (including body wash and hand soap)	—	—	MCiNP	—	0.34
MECPTP	1.11 (0.96, 1.29)	0.15	MCPP	0.84 (0.72, 0.97)	0.02
DEHTP	1.03 (0.98, 1.08)	0.20	MECPP	0.84 (0.68, 1.02)	0.08
HMW	1.04 (0.99, 1.09)	0.07	MECPTP	0.79 (0.56, 1.12)	0.18
Sunscreen	—	—	MEHHP	0.87 (0.69, 1.10)	0.26
MCOCH	0.89 (0.76, 1.04)	0.15	MEHHTP	0.83 (0.63, 1.09)	0.20
MCPP	1.11 (0.94, 1.30)	0.21	MEOHP	0.89 (0.73, 1.08)	0.25
MHNCH	0.84 (0.69, 1.04)	0.15	MEP	0.74 (0.53, 1.04)	0.09
MiBP	1.17 (0.90, 1.48)	0.08	MBP	0.78 (0.62, 0.98)	0.03
MBP	1.14 (0.89, 1.47)	0.28	DEHP	0.97 (0.92, 1.02)	0.19
DINCH	1.00 (0.99, 1.00)	0.21	DEHTP	0.95 (0.85, 1.05)	0.29
Ointments	—	—	LMW	0.93 (0.83, 1.04)	0.20
MCOCH	1.06 (0.47, 1.22)	0.38	HMW	0.92 (0.83, 1.02)	0.13
MECPTP	1.23 (0.89, 1.68)	0.21	3-in-1	—	—
MEP	0.99 (0.73, 1.35)	0.95	MECPTP	0.51 (0.26, 1.19)	0.06
MHNCH	1.06 (0.90, 1.29)	0.40	DEHTP	0.84 (0.68, 1.04)	0.10
MiNP	0.99 (0.95, 1.02)	0.34	HMW	0.82 (0.66, 1.02)	0.07
DINCH	1.08 (0.99, 1.01)	0.95	Deodorant	—	—
DEHTP	1.03 (0.93, 1.13)	0.57	MEP	1.19 (0.77, 1.83)	0.44
Shampoo	—	—	Other	—	—
MBzP	1.05 (0.89, 1.23)	0.60	MEHP	0.89 (0.71, 1.10)	0.26

Note: Each row represents the results of an individual linear regression model. All models adjusted for child’s urinary specific gravity, age (years), race/ethnic identity (NHB, Hispanic, NHW, or Asian/PI), sex assigned at birth (female/male), body mass index (${\mathrm{kg}/\mathrm{m}}^2$), maternal education (completed secondary school yes or no), and season of urine collection (fall, winter, spring, or summer). SCP units are described in Table 3. —, no data; CI, confidence interval; DEHP, diethylhexyl phthalate; DEHTP, di-2-ethylhexyl terephthalate; DINCH, di-isononyl cyclohexane-1,2-dicarboxylate; ECHO-FGS, Environmental Influences on Child Health Outcomes-Fetal Growth Study; HMW, high molecular weight; LMW, low molecular weight; MBP, mono-n-butyl phthalate; MBzP, mono-benzyl phthalate; MCiNP, mono-carboxy isononyl phthalate; MCiOP, mono-carboxy isooctyl phthalate; MCOCH, cyclohexane-1,2-dicarboxylic acid mono carboxyisooctyl ester; MCPP, mono (3-carboxypropyl) phthalate; MECPP, mono (2-ethyl-5-carboxypentyl) phthalate; MECPTP, mono-2-ethyl-5-carboxypentyl terephthalate; MEHHP, mono (2-ethyl-5-hydroxyhexyl) phthalate; MEHHTP, mono-2-ethyl-5-hydroxyhexyl terephthalate; MEHP, mono ethyl hexyl phthalate; MEOHP, mono (2-ethyl-5-oxohexyl) phthalate; MEP, monoethyl phthalate; MHNCH, cyclohexane-1,2-dicarboxylic acid mono hydroxyisononyl ester; MiBP, mono-isobutyl phthalate; MiNP, mono-isononyl phthalate; NHB, non-Hispanic black; NHW, non-Hispanic white; PI, Pacific Islander; SCP, skincare product.

The associations between individual SCPs and children’s urinary phthalate/replacement metabolite concentrations were different according to child’s race/ethnic identity group (Table S9). Specifically, use of body lotion was associated with a 1.10-fold (95% CI: 1.02, 1.19) greater geometric mean LMW among children in the NHW group but with a 0.83-fold (95% CI: 0.70, 0.98) lesser geometric mean LMW among children in the Asian/PI group, and with a null association among the children in the NHB and Hispanic groups (p-value for modification $=$ 0.06). However, this difference was not statistically significant after multiple testing correction. While use of oils was associated with higher urinary MEP and LMW levels among all children, the magnitudes were greatest among children in the Asian/PI (290.03-fold, 95% CI: 39.65, 2,100.65 and 16.44-fold, 95% CI: 10.59, 25.53, respectively) and Hispanic (3.78-fold, 95% CI: 1.50, 9.49 and 1.35-fold, 95% CI: 1.04, 1.77, respectively) groups (p-value for modification $=$ 0.0001 and $1.18\times {10}^{-10}$, respectively). However, the effect estimate for children in the Asian/PI group was very imprecise because few children ($n<5$) reported use of oils. Sunscreen was associated with 1.59-fold higher urinary MCPP in the Hispanic group (95% CI: 1.09, 2.33) but was close to a null association in other groups, although without statistical significance after correction for multiple testing error (p-value for modification $=$ 0.09). Use of ointments was similarly associated with a 2.07-fold (95% CI: 1.07, 4.01) higher urinary MEP levels among the children in the Hispanic group and with lesser urinary MEP concentrations among other children (p-value for modification $=$ 0.09) but was nonsignificant after correction for multiple testing. Finally, children using conditioner in the NHB group had a 1.38-fold (95% CI: 0.97, 1.95) higher urinary MCPP level, but associations were close to the null in the other racial/ethnic groups (p-value for modification $=$ 0.07), and the difference was not statistically significant after correction for multiple testing error.

Differences in associations between individual SCPs and children’s urinary phthalate/replacement metabolite were also modified by child’s sex assigned at birth (Table S10). Among children assigned male at birth, use of ointments was associated with higher urinary MCOCH (1.18-fold, 95% CI: 0.99, 1.41; $p$-value for modification $=$ 0.08), MECPTP (1.55-fold, 95% CI: 1.02, 2.35; $p$-value for modification $=$ 0.09), MHNCH (1.29-fold, 95% CI: 1.03, 1.62; $p$-value for modification $=$ 0.03), and DEHTP (1.12-fold, 95% CI: 0.99, 1.27; $p$-value for modification $=$ 0.05), although not among females. However, the differences were not statistically significant after multiple testing correction. In contrast, use of oils was associated with 1.69-fold (95% CI: 1.39, 2.05; $p$-value for modification $=$ 0.004) higher urinary LMW among female children.

Associations Between SCP-Use Exposure Profile Categories and Urinary Phthalate/Replacement Metabolites

We implemented SOM to reduce complex patterns of variability across 18 SCP-use variables into a small number of distinct categories defined by weight vectors referred to as “codes,” that are projected onto a two-dimensional representation known as the “map” as shown in Figure 1 (numeric data can be found in Table S11, which summarizes the codes for each exposure profile). Profile 1 corresponds to moderate SCP use focused on greater use of shampoo and conditioner, soap (including body wash and hand soap), and products labeled as paraben-free and phthalate-free; Profile 2 also corresponds to moderate overall SCP use but focused on greater use of shampoo without conditioner, soap (including body wash and hand soap), sunscreen, products labeled as paraben-free and phthalate-free, and medicated products; Profile 3 corresponds to low overall SCP use relative to other profiles; and Profile 4 corresponds to higher overall PCP use relative to other profiles. Categories are defined based on the frequencies of participants’ responses to each SCP-use variable normalized to the range of responses in the population so that minimum and maximum reported values are represented by 0.0 and 1.0, respectively “codes,” and illustrated using radial bars to indicate average frequency of use. Longer bars correspond to more frequent use. For example, lotions contributed to exposure profiles 4 (code $=$ 0.312, indicating an average response that was 31.2% of the maximum reported frequency used in the study population), 2 (code $=$ 0.299), 3 (code $=$ 0.267), and 1 (code $=$ 0.218). Oils played an important role in exposure profile 4 (code $=$ 0.236, indicating an average response that was 23.6% of the maximum reported frequency used in the study population) but was a much smaller contributor to exposure profiles 3 (code $=$ 0.048), 2 (code $=$ 0.033), and 1 (code $=$ 0.027). Shampoo contributed substantially to exposure profiles 2 (code $=$ 1.000, indicating an average response equal to the maximum reported frequency used in the study population) and 1 (code $=$ 0.997), modestly to 3 (code $=$ 0.255), but not to 4 (code $=$ 0.000, indicating an average response equal to the minimum reported frequency used) in the study population.

Figure 1.

Exposure continuum map describing frequencies of children of ECHO-FGS participants categorized into SCP-use exposure profiles using a self-organizing map ($n=906$) (numeric data in Table S11). Note: Exposure profile categories are defined based on the frequencies of participant responses to each SCP-use question and illustrated using radial bars labeled with letters indicating the average reported frequencies normalized to the range of reported frequencies in the study population; longer bars correspond to greater use. For example, exposure profiles 2 and 4 include children that reported greater use of lotions than children in exposure profiles 1 and 3. Cond., hair conditioner; Deod., deodorant; ECHO-FGS, Environmental Influences on Child Health Outcomes-Fetal Growth Study; Hair, hair products other than shampoo, conditioner, or oil; Hand., hand sanitizer; Lip., lip products; Med., medicated products; Oils, hair oil; Organic, products using only organic ingredients; Par-fr., paraben-free products; Pht-fr., phthalate-free products; SCP, personal care product; Soap, bar soap, liquid soap, and body wash; Sun., sunscreen; 2–1s, 2-in-1 shampoo and hair conditioner products; 3–1s, 3-in-1 body wash, shampoo, and hair conditioner products.

We subsequently used the SOM-based SCP-use exposure profile assignments to estimate the relationships between categorical exposure profiles and urinary phthalate/replacement concentrations in general linear regression as shown in Figure 2 (numeric data can be found in Table S12). We used SCP-use exposure profile 3, the most prevalent exposure profile, and the lowest exposure, as the reference category, and adjusted for confounding factors. Children with SCP exposure profiles 1 and 2 had 1.09-fold higher (95% CI: 1.03, 1.15) and 1.06-fold higher (95% CI: 1.00, 1.13) geometric mean urinary MCiNP concentrations, respectively, compared to those with exposure profile 3 (Figure 2B). Children with exposure profile 2 had 1.31-fold (95% CI: 0.98, 1.77) higher geometric mean urinary MEHHTP (Figure 2I), and children with SCP exposure profile 1 had a 1.13-fold (95% CI: 0.99, 1.29) higher geometric mean level of MEHP (Figure 2J), compared to profile 3, although not statistically significant. Similarly, exposure profile 1 was associated with a 1.04-fold (95% CI: 1.00, 1.09) higher geometric mean level of urinary DEHP (Figure 2R), albeit again not statistically significant.

Figure 2.

Relationships between SCP-use exposure profile categories and urinary phthalate/replacement metabolite concentrations (ng/mL or nmol/L for composite sums) among children of ECHO-FGS participants ($n=618$) (numeric data in Table S12). Note: Multiple linear regression models adjusted for child’s urinary specific gravity, age (years), race/ethnic identity (NHB, Hispanic, NHW, or Asian/PI), sex assigned at birth (female/male), and body mass index (${\mathrm{kg}/\mathrm{m}}^2$), maternal education (completed secondary school yes or no), and season of urine collection (fall, winter, spring, or summer). We estimated the effect of membership in exposure category 1, 2, or 4 compared to exposure category 3 on the geometric mean urinary phthalate/replacement concentration (ng/mL or nmol/L for composite sums). The points on the graph represent the effect estimates for the relative differences in geometric mean concentrations, and the lines represent the 95% confidence intervals. CI, confidence interval; DEHP, diethylhexyl phthalate; DEHTP, di-2-ethylhexyl terephthalate; DINCH, di-isononyl cyclohexane-1,2-dicarboxylate; ECHO-FGS, Environmental Influences on Child Health Outcomes-Fetal Growth Study; LMW, low molecular weight; HMW, high molecular weight; MBP, mono-n-butyl phthalate; MBzP, mono-benzyl phthalate; MCiNP, mono-carboxy isononyl phthalate; MCiOP, mono-carboxy isooctyl phthalate; MCOCH, cyclohexane-1,2-dicarboxylic acid mono carboxyisooctyl ester; MCPP, mono (3-carboxypropyl) phthalate; MECPP, mono (2-ethyl-5-carboxypentyl) phthalate; MECPTP, mono-2-ethyl-5-carboxypentyl terephthalate; MEHHP, mono (2-ethyl-5-hydroxyhexyl) phthalate; MEHHTP, mono-2-ethyl-5-hydroxyhexyl terephthalate; MEHP, mono ethyl hexyl phthalate; MEOHP, mono (2-ethyl-5-oxohexyl) phthalate; MEP, monoethyl phthalate; MHNCH, cyclohexane-1,2-dicarboxylic acid mono hydroxyisononyl ester; MiBP, mono-isobutyl phthalate; MiNP, mono-isononyl phthalate; NHB, non-Hispanic black; NHW, non-Hispanic white; PI, Pacific Islander; SCP, skincare product.

Discussion

In this retrospective cohort investigation, we found associations between reported use of SCPs and young children’s urinary phthalate/replacement metabolite concentrations. Specifically, use of lotions, oils, and sunscreens were associated with higher urinary levels of primarily LMW metabolites. We also found that child’s race/ethnic identity modified the associations between SCP use and children’s urinary phthalate/replacement metabolites. Use of lotion was associated with higher urinary LMW in NHWs, oils with higher LMW in Asians/PIs and Hispanics, sunscreen and ointments with greater MCPP and MEP in Hispanics, and conditioner with greater MCPP in NHBs. Sex-assigned at birth modified associations as well, as ointment use was associated with higher levels of urinary phthalate replacement metabolites in males and oils with higher LMW in females. Mixtures of moderate overall SCP use, defined as exposure profiles, were associated with higher concentrations of primarily HMW urinary phthalate/replacement metabolites, though not for high overall SCP use. These results suggest that recently reported differences in exposure to endocrine-disrupting chemicals used in personal care and beauty products marketed to minoritized groups begin at an early age^18,19,45 and that patterns of SCP use make different contributions to children’s phthalate/replacement exposure than isolated SCPs.

Children’s Use of SCPs and Urinary Phthalates/Replacement Metabolite Levels

Children participating in our study used SCPs frequently in the 24 h preceding the time of the study questionnaire via parental report. We found that most children had applied at least one type of soap (including body wash and hand soap) and lotion, and a substantial proportion of children had also used shampoo. However, oils, sunscreen, ointments, hand sanitizer, hair conditioner, other haircare products, 2-in-1 and 3-in-1 haircare and hair/bodycare products, lip products, and deodorant had been used by $<10\%$ of children.

While previous studies described patterns of using PCPs, including SCPs, among adolescents, adults, infants, and toddlers, very few data are available to characterize SCP use among young US children, and the sample sizes of previous studies were limited.^11,46–50 Approximately 74% of 4- to 6-year-olds and 80% of 7- to 9-year-olds reported having ever used makeup (cosmetic) and body products intended for application to hair, skin, and nails in a small nonrepresentative 2012–2022 survey of US children.⁵¹ Older, female, and children of Hispanic-identifying parents used PCPs more frequently; 12% of the children used products on a daily basis. In our study, we found that a greater proportion of children (68.5%) had used SCPs but less frequently among children in the Hispanic group (65%) than among the children in the NHW (78%) and Asian/PI (70%) groups. In contrast to the prior study, we did not target cosmetic products, which may account in part for the discrepant results.⁵¹ Similar to our results, a 2006–2009 study of children in northern California, mostly $\le 5\text{\hspace{0.17em}years}$ of age, reported more frequent use of hair conditioner, lotion, facial cleanser, lip balm, and sunscreen among females than males, and differences in the maternal frequency of PCP use according to race identity.⁵² We found similar differences in reported SCP use by sex assigned at birth, as well as differences according to race/ethnic identity group.

Urinary phthalate/replacement metabolite levels were generally lower in our study population than reported for a representative sample of US children 4–8 years of age in 2017–2018.⁵³ Geometric mean urinary MBzP, MCiNP, MCiOP, MCOCH, MCPP, MECPP, MEHHTP, MEHP, MEOHP, MHNCH, and MiNP were less in this study population than for US children, although geometric mean MECPTP, MEP, MiBP, and MBP were similar, and geometric mean MEHHP was greater as shown in Table S13. We also found higher urinary phthalate/replacement metabolite levels among children in the NHB group in our study than in other race/ethnic identity groups, which is similar to the patterns reported by biomonitoring studies of US adults.^54–56

Associations Between SCPs and Children’s Urinary Phthalate/Replacement Metabolites

Previous studies reported associations between using PCPs, including SCPs, and urinary phthalate metabolites in adults, but few studies were conducted among young children in North America.^10,57 The aforementioned 2006–2009 study from California reported null associations for urinary MEP, MBP, and MiBP levels and use of deodorant, shampoo, “other” haircare products (including conditioner), soaps, lotion, and suntan lotion in the past 24 h, adjusted for child sex and age, parental education level and race, urinary creatinine, and the total number of PCPs used.¹⁰ However, lip balm use was associated with a 1.33-fold (95% CI: 1.02, 1.74) higher geometric mean urinary level of MEP. We found similar null results for MEP, MBP, and MiBP with use of the analogous SCPs. We did not find an association for use of lip products in our study population, but MBzP was associated with lotion use, MEP and MBP were associated with use of oils, and MCPP and MBP were associated with use of 2-in-1 haircare products. A 2010 study of children 8–13 years of age in Mexico found a significant association between boys’ lotion use in the past 48 h and higher urinary MEHP ($n=53$), girls’ use of hair conditioner and deodorant and higher urinary MEP ($n=55$), and girls’ use of haircare products other than shampoo, conditioner, hair cream, and hair spray/gel and higher urinary MBP, MiBP, and MCPP.⁵⁷ We too found that child’s sex assigned at birth modified the associations between SCPs and urinary phthalates, in which use of oils was associated with higher urinary LMW among females, and the associations between use of ointments and greater urinary MCOCH, MECPTP, MHNCH, and DEHTP were limited to males. Differences in the results between the studies may be attributed in part to different formularies in products sold on Mexican and US shelves, older children participating in the study from Mexico, and changes in manufacturer practices between the earlier 2010 study from Mexico and the current study conducted in 2017–2018.⁵⁸

We found different associations between children’s use of some SCPs and urinary phthalate/metabolite concentrations among racial/ethnic identity groups. The differences were mostly limited to LMW phthalate metabolites, which are reported to be associated with use of PCPs in adults and children.⁸ We found higher urinary LMW among the NHW group using body lotion. We also found substantially elevated MEP and LMW in the Asian/PI and Hispanic groups reporting recent use of oils, although not among the NHB group for whom use was most prevalent. The differences may be driven by the specific product used and differences in application. Using hair oils has been associated with racial differences in pregnancy outcomes in a recent study,⁵⁹ suggesting a potential to contribute to race/ethnic health disparities in adults.¹⁹ However, our results were limited by the small number of children who reported using oils. A larger study with more comprehensive exposure information will be necessary for a more definitive interpretation of this result. Use of hair conditioner was associated with higher urinary MCPP among the NHB group, a metabolite of di-n-octyl phthalate (DnOP). DnOP is a HMW phthalate incorporated into polyvinyl chloride (PVC) plastics, which can be used in product packaging and may be a developmental neurotoxicant.^42,60 It is tempting to speculate that DiNP migrating from plastic packaging in hair conditioner brands marketed to NHB populations potentiated exposure among children in our study.^13,14,45 However, we were unable to test this hypothesis in our data so a future study with collection and direct testing of SCPs will be necessary to more clearly interpret this result.

While preliminary, these results suggest that children in different race/ethnic groups experienced different exposure to specific phthalates using SCPs, potentially placing them at greater risk for associated developmental health effects.^2–6,33 The differences may be related to brand availability and preferences, methods and timing of product application, and/or the frequency of use by different race/ethnic groups.

We applied SOM to examine usage patterns across multiple SCPs and identify the children at greatest risk for exposure to phthalates/replacements via SCPs. Previous studies have reported associations for children’s urinary phthalate metabolites with total numbers of PCPs used^57,61,62 or classified according to PCP type or intended use.^10,51,57 An attractive feature of SOM is that it places more similar SCP-use exposure profiles closer in proximity and more dissimilar SCP-use exposure profiles farther apart, giving insight to profile relationships. Here, our results revealed that most children ($n=710$, 78%) in our population had used SCPs in small amounts (profile 3) (Figure 1). The remaining 22% consisted of smaller subgroups of children with higher frequency use of different combinations of multiple SCPs (profiles 1, 2, and 4) (Figure 1). SCP-use exposure profile 1, characterized by moderate SCP use overall, especially shampoo, conditioner, deodorant, and paraben-free products, and SCP-use exposure profile 2, also characterized by moderate SCP use overall but more frequent use of soaps (including body wash and hand soap), sunscreen, shampoo, and medicated SCPs, were associated with higher urinary MCiNP. MCiNP is a metabolite of di-isononyl phthalate (DiNP), which is a potentially neurotoxic phthalate used in PVC plastics.^4,42 In contrast, we did not find associations between SCPs and urinary MCiNP using the single-SCP predictor regression models. Still, we measured MCiNP above the LOD in less than half of the children, so this result requires confirmation in a future study.

Similarly, we found that SCP-use exposure profiles 1 and 2 were associated with higher urinary DEHP metabolites and MEHHTP. DEHP is also used in PVC plastics and has evidence of developmental neurotoxicity and endocrine disruption.^4,63 MEHHTP is a metabolite of DEHTP, a recently developed DEHP replacement compound.⁶⁴ While DEHTP exposure has been growing in the US population, few human toxicity data are available to assess the potential health risks to young children.⁶⁵ Despite reflecting high overall SCP use, exposure profile 4 was not associated with urinary phthalate levels, underscoring the importance of investigating distinct permutations of multiple SCPs used as potentially important sources of children’s exposure to phthalates/replacements.

Dietary sources are espoused as the primary route of human exposure to HMW phthalates/replacements, including DiNP, DEHP, and DEHTP,³³ which are not typically reported as ingredients in SCPs.^33–36 However, our results are consistent with DiNP, DEHP, and DEHTP migrating from plastic packaging into the SCPs applied to children’s skin, which exposed our study population.^13,14 Furthermore, the associations between SOM-derived SCP-use exposure profiles were not revealed in single-SCP predictor regression models; thus, our results underscore the importance of interrogating complex SCP-use patterns to identify phthalate/replacement exposure risks. We did not, however, find associations between SCP-use exposure profile categories and LMW urinary phthalates that are commonly reported ingredients in SCPs^33–36 and that we detected using the single-SCP predictor regression models, suggesting that using both the single-SCP predictors and SCP-use exposure profiles were complementary approaches.

Strengths and Limitations

Our study has several strengths, including a large sample size of young children, a group that has received little attention, a racially/ethnically diverse study population that allowed for racial/ethnic effect modification analyses, open-ended collection of data about SCPs used, a comprehensive panel of phthalate/replacement metabolites measured in urine, and both single SCP-predictor and SCP-use exposure profile-based approaches to identify predictors of children’s urinary phthalate/replacement metabolites. In addition to the metabolites of prevalent phthalate diesters, we measured metabolites of DINCH and DEHTTP, newer products increasingly used to replace HMW phthalates in PVC and other plastics applications.^66,67

However, our study also had limitations. Importantly, not all study questionnaires were completed at the time of urine collection; 64.7% were completed on the day of urine collection, 74.6% were completed within 24 h, 77.5% were completed within 48 h, and 83.4% were completed within 1 week before urine collection. This may have misclassified exposure in some children because urinary phthalates/replacements have short half-lives and high day-to-day variation.⁶⁸ We do not anticipate that questionnaire completion timing would be differential according to children’s use of SCPs or urinary phthalate/replacement concentrations. However, this may have biased results toward the null hypothesis and contributed to our unexpected null results between SCP-use exposure profile categories and LMW phthalate metabolites.

In defining exposure categories, we “binned” reported SCPs by product type and by product formulary (ingredients) to facilitate the statistical analysis. We used the SkinSAFE database to augment the information on product labels in creating formulary categories. However, SkinDeep (https://www.ewg.org/skindeep/), DailyMed (https://dailymed.nlm.nih.gov/dailymed/index.cfm), the Consumer Products Information Database (https://www.whatsinproducts.com/types/similar_product/1/2/home), and other online sources are also available to consumers. A recent study reported lower error rates in SkinSafe ingredient lists relative to some other databases, although product labels were most accurate.²⁸ Thus, future studies are needed to confirm our results using different online SCP ingredient databases to augment product label information. Furthermore, we queried products applied to skin only, which may be confounded by using other PCPs that are also sources of exposure to phthalates/replacements and may be applied in conjunction with SCPs. Perfumes and other fragrances may be used by children⁵² and have been associated with urinary phthalates/replacements in adults.^21,69 We also did not collect data about motivations for product use, which will be important to design interventions that reduce children’s exposure to phthalates/replacements. However, we investigated SCPs labeled as organic, phthalate-free, and paraben-free. Consumers are motivated in part by health concerns when purchasing SCPs labeled as organic, although the organic label is not regulated for use on SCPs by the US Food and Drug Administration, and product labels may be inaccurate.^70,71

Despite a large number of study participants, we were unable to evaluate differences in associations by child’s race/ethnic identity according to clinical study site, in particular for less represented regions like the southeastern US. Most participants completed a secondary school degree, which similarly precluded a stratified analysis by maternal education, which may drive product choice.⁵² Thus, larger regional studies will be necessary to further understand the nature of phthalate/replacement exposure disparities among children using SCPs. We also conducted multiple statistical tests, and a Bonferroni correction procedure to the type 1 error suggested that some findings may have occurred by chance.⁷² A future study will be necessary to confirm our results. Finally, we did not consider lifestyle factors and clinical factors associated with exposure to phthalates/replacements, such as diet.⁷³ Food allergies related to skin conditions may confound associations,⁷⁴ while skin conditions, such as atopic dermatitis, may impact the selection of SCPs and absorption of phthalates. Few children used medicated SCPs in our study, and we did not find an association with urinary phthalate/replacement metabolites. However, clinical skin conditions should be considered to assess the impact and the associated cost/benefit of using SCPs in a potentially vulnerable population. To confirm our findings, a future investigation should collect data on all PCPs used over several days, ideally in a prospective fashion, coupled to multiple urine collections in a larger sample to more accurately and reliably estimate the associations between use of SCPs and urinary phthalate/replacement levels among young children.

Conclusions

We found that reported use of SCPs was associated with specific urinary LMW and HMW phthalate/replacement metabolite concentrations in children and that the associations differed according to race/ethnic identity and sex assigned at birth. We also found that specific exposure combinations of using multiple SCPs were associated with higher urinary concentrations of HMW phthalates and their replacements. These results suggest that SCPs may be an important source of exposure to both LMW and HMW phthalates and their replacements in children 4 to 8 years of age. We investigated the differences between self-reported race/ethnic groups to identify potential inequities in exposure to phthalates/replacements among US children. While we could not ascertain the etiologies of the differences in this study, the results can inform future studies to disentangle components of structural racism as a mechanism that is likely to drive differences.²³ The results can also promote discussions among policymakers that regulate manufacture and packaging of SCPs to eliminate endocrine-disrupting chemical exposure disparities among children, eliminate targeted marketing of endocrine-disrupting chemical-containing products toward black and Hispanic children, and underscore the importance of concurrent use of multiple SCPs as a source of exposure to endocrine-disrupting chemicals, rather than focus on individual products. Furthermore, these results can help clinicians and advocacy groups to advise parents and guardians on product choices and use to limit children’s exposure to potentially hazardous phthalates/replacements. However, given study limitations, the results require confirmation.

Conclusions and opinions are those of the individual authors and do not necessarily reflect the policies or views of EHP Publishing or the National Institute of Environmental Health Sciences.