Associations of environmental chemical exposures measured in personal silicone wristbands with sociodemographic factors, COVID-19 restrictions, and child respiratory health

Abstract

Background:

Although human biomonitoring of environmental chemicals has been considered a gold standard, these methods can be costly, burdensome, and prone to unwanted sources of variability that may cause confounding. Silicone wristbands have recently emerged as innovative passive samplers for measuring personal exposures.

Methods:

In a pilot study from 2019-2021 involving 55 children aged 5-9 years in Seattle and Yakima, Washington, we utilized silicone wristbands to explore associations of sociodemographic variables and COVID-19-related restrictions, including school closures, with exposures to numerous chemicals including brominated and organophosphate ester (OPE) flame retardants, polychlorinated biphenyls, polycyclic aromatic hydrocarbons (PAHs), phthalates, and pesticides. We additionally conducted the first analysis testing silicone wristband chemicals as predictors of child wheeze, individually and in mixtures via logistic weighted quantile sum regression (WQS).

Results:

Among 109 semi-volatile organic compounds measured, we detected 40 in >60% of wristbands worn by children continuously for an average of 5 days. Chemicals were generally positively correlated, especially within the same class. Male sex and increasing age were linked with higher exposures across several chemical classes; Hispanic/Latino ethnicity was linked with higher exposures to some phthalates and OPEs. COVID-19 restrictions were associated with lower wristband concentrations of brominated and triaryl OPE flame retardants. Each one-decile higher WQS exposure index was suggestively associated with 2.11-fold [95% CI: 0.93-4.80] higher odds of child wheeze. Risk of child wheeze was higher per 10-fold increase in the PAH chrysene (RR=1.93[1.07-3.49]), the pesticide cis-permethrin (3.31[1.23-8.91]), and di-isononyl phthalate (DINP) (5.40[1.22-24.0])

Conclusions:

Our identification of demographic factors including sex, age, and ethnicity associated with chemical exposures may aid efforts to mitigate exposure disparities. Lower exposures to flame retardants during pandemic restrictions corroborates prior evidence of higher levels of these chemicals in school versus home environments. Future research in larger cohorts is needed to validate these findings.

Introduction

Accurate assessment of personal chemical exposures is critical to the advancement of environmental health sciences. Human biomonitoring of environmental chemicals has been traditionally considered as a gold standard,¹ but many factors including costs, lab methods, and biospecimen storage and transport procedures can limit our ability to identify biomarkers that are sensitive, specific, and easy to measure.² Biospecimen collection and storage under the proper conditions can be challenging and costly.³ Of particular concern for vulnerable populations including children, the collection of biospecimens may require invasive procedures such as an intravenous blood draw. Additionally, biomarkers for non-persistent chemicals with short half-lives may be unreliable without multiple samples and unreasonably short intervals between biospecimen collections.⁴

Beyond the challenges of sample collection, biomarker identification and analysis can also pose significant difficulties. Identifying biomarkers of an external exposure requires a comprehensive understanding of the exposure’s toxicokinetic profile in humans. A xenobiotic could have no known or measurable biomarkers in humans, or it could be rapidly metabolized into multiple compounds.³ Even when a good biomarker is identified, features other than the exposure of interest, such as intra-individual variation in metabolism and differences in other personal characteristics,² may substantially impact biomarker concentrations and confound epidemiologic studies. For instance, although urinary biomarkers are typically adjusted based on hydration status (by correcting for specific gravity or creatinine), it is uncommon to account for urinary flow rate, which could confound epidemiologic studies through its effects on exposure assessment and health outcomes.⁵ While biomarkers are often suitable for exposure assessment, there are instances where alternative methods to measure personal exposures are necessary and may be more accurate.

To measure continuous personal exposures without the use of internal biomarkers, studies have employed active and passive samplers carried in air monitoring backpacks.⁶ However, these devices can be expensive for researchers and burdensome for study participants. Smaller and less burdensome passive diffusion samplers have also been utilized for workplace air monitoring, but are typically designed to measure only one or a few compounds.^3,7-11 More recently, silicone wristbands have been used as novel personal passive samplers to measure exposure to volatile and semi-volatile organic compound (VOC and SVOC) contaminants.^3,9-15 Silicone wristbands are inexpensive, non-invasive, and easier to wear compared to other personal exposure measurement devices. Rather than requiring different techniques to assess different chemicals, silicone wristbands facilitate the simultaneous measurement of a wide range of chemicals with different physicochemical properties.³ Additionally, wristbands provide an average exposure measure over the period they are worn without the need for repeated biomonitoring events.¹⁶

Silicone wristbands have been employed in human environmental exposure assessments for a wide range of chemicals,^3,9-15 including brominated flame retardants (BFRs), legacy flame retardants that have been phased-out of production; organophosphate esters (OPEs), used as modern alternative flame retardants; polychlorinated biphenyls (PCBs), a class of chemicals used as heat transfer fluids in electrical equipment that have been phased out because of their endocrine disrupting properties; polycyclic aromatic hydrocarbons (PAHs), products of partial combustion found in air pollution and tobacco smoke; phthalates, ubiquitous endocrine disrupting chemicals used as plasticizers; and various pesticides. Studies show associations of exposure to OPEs,^17,18 PCBs,¹⁹ PAHs,^20-24 phthalates,^25-27 and pesticides^28,29 with respiratory health outcomes including asthma, wheeze, and spirometry. However, many prior studies may be limited by their exposure assessment methods. For example, some PAHs have short half-lives, on the order of hours in humans.³⁰ Spot urine PAHs have been shown to correlate reasonably well with 48 hour wristband measurements,³¹ but spot urine PAHs may not accurately reflect exposures over longer periods of time.^32,33 In addition to advantages over PAH biomarkers, wristbands may also more accurately assess personal PAH exposures compared to air monitoring backpacks by being in closer proximity to PAH point sources, incorporating dermal exposure, and selectively capturing the bioavailable PAH fraction.⁶ Thus, in addition to possible sources of confounding discussed above, certain chemical exposures could be particularly prone to misclassification when assessed using biomarkers, which could result in bias toward null associations with health outcomes. Despite their advantages, we are unaware of any studies that have explored associations of environmental chemicals measured using personal silicone wristbands with child respiratory outcomes.

Beyond uncovering links between chemical exposures and adverse health, personal silicone wristband studies can also be utilized to identify demographic, socioeconomic, behavioral, and environmental predictors of these chemical exposures. Government- and self-imposed restrictions during the COVID-19 pandemic provide a unique context for natural experiments to examine these factors. For example, school closures during the pandemic could have resulted in increased exposure to chemicals associated with the home environment but reduced exposure to chemicals associated with the school environment and traveling to school. Flame retardants, for instance, have been found at higher concentrations in schools and cars compared with the home.^34,35

In addition to environmental determinants of exposure, differences in chemical exposures according to sociodemographic variables could also influence morbidity and be studied using personal silicone wristbands. For instance, personal silicone wristband studies have uncovered higher levels of environmental chemicals among low socioeconomic status (SES) individuals.^36,37 The identification of sociodemographic predictors of chemical exposures could help identify vulnerable populations, thereby supporting efforts to mitigate exposure disparities that may influence long term health.

Using a pilot sample of 55 children with silicone wristband chemical exposure data in Washington state, we examined 1) associations of sociodemographic variables with chemical exposures, 2) whether participation in wristband sampling before vs. during COVID-19 associated restrictions, including school closures, was linked with altered chemical levels, and 3) associations of chemical exposures measured in silicone wristbands with child wheeze and asthma.

Methods

Study population

The Global Alliance to Prevent Prematurity and Stillbirth (GAPPS) biorepository was launched by the Seattle Children’s Hospital in 2007 to study the impact of adverse birth outcomes. Funded through the National Institutes of Health sponsored Environmental influences on Child Health Outcomes prenatal and early childhood pathways to health (ECHO PATHWAYS) consortium, eligible GAPPS participants were re-contacted to participate in longitudinal follow-up in the PATHWAYS-GAPPS cohort study. Eligibility criteria included delivery in Seattle, WA at the University of Washington or Swedish Medical Center, or delivery in Yakima, WA at the Yakima Valley Memorial Hospital; availability of at least one pregnancy urine sample; initial GAPPS enrollment and completion of questionnaire; and child 4–7 years of age. More than 600 mothers and children have been re-enrolled into PATHWAYS-GAPPS. Study protocols were approved by the Institutional Review Boards of the Seattle Children’s Research Institute and the University of Washington.

Silicone wristband exposures

A subsample of 55 PATHWAYS-GAPPS children participated in the silicone wristband pilot, either at the age 4-6 or the age 8-10 study visit. Commercially available silicone wristbands (24hourwristbands.com, Houston, TX, USA) were prepared following previously published methods.^38-41 Wristbands were pre-cleaned through two 12 h Soxhlet extractions using first 1:1 ethyl acetate/hexane (v/v) and then 1:1 ethyl acetate/methanol (v/v), then dried in a vacuum oven at room temperature. When dried, cleaned wristbands were wrapped in combusted aluminum foil (i.e., baked at 450 °C) and stored in zipper storage bags at room temperature until distribution to study participants.

Silicone wristbands were given to participants at study visits or shipped to families along with questionnaires. Children were instructed to wear the silicone wristband continuously for 5 days, and to not remove it for any activity, including showering, swimming, or sleeping. Families were asked to report the date and time they opened the airtight wristband packaging and put the wristband on their child’s wrist, the date and time when they removed the wristband from their child’s wrist and placed it into the provided return packaging, and the amount of time, if any, during which the wristband was not worn by the child. We used these data to determine how long the wristbands were exposed to air in the children’s environments, how long the wristbands were worn by children, and the season.

The methods for targeted measurement of SVOCs in silicone wristbands are described in detail elsewhere.^38-41 Briefly, an 0.75-g piece was taken from every wristband, weighed, and spiked with isotopically labeled internal standards. Wristband pieces were extracted using 1:1 hexane:dichloromethane. A Thermo Scientific SpeedVac Concentrator was used to concentrate extracts to 1.0 mL before being purified using a packed glass column containing 8g of Florisil resin. Lastly, 1.0 mL extracts were transferred into auto sampler vials (ASV) for instrumental analysis. Samples were analyzed using a Q Exactive GC hybrid quadrupole-Orbitrap GC–MS/MS system (Thermo Scientific) in full scan electron ionization mode. These chemicals include: 11 PCBs, 33 OPEs, 10 phthalates and their alternatives, 16 pesticides, 25 PAHs, and nicotine (Supplemental Table 1). Ten Polybrominated Diphenyl Ethers (PBDEs) and 5 novel brominated flame retardants (NBFRs) were quantified using a Q Exactive GC hybrid quadrupole-Orbitrap GC–MS/MS system (Thermo Scientific) operated in negative chemical ionization mode (Supplemental Table 1).

Isotopically labelled recovery standards were spiked into each sample extract prior to MS analysis to calculate recovery of the internal standards (Supplemental Table 2). Lab processing blanks (solvent only, n = 6) and field blanks (unworn wristbands, n = 1 from Yakima and 2 from Seattle) were analyzed alongside the wristband samples for quality assurance and control. All sample values were blank-subtracted using the average level measured in all blanks (if detected). No difference was observed between field and lab blanks. Method detection limits (MDLs) were determined by calculating three times the standard deviation of all blank levels and then dividing by the mass of the wristband. If the analyte was not detected in a blank, the MDL was determined using a signal to noise ratio of 10 on the instrument.

Non-detects were imputed with the MDL/sqrt(2) and concentrations were calculated as chemical mass divided by wristband section mass. These concentrations were normalized by the number of days exposed to air in the child’s environment (the difference between the date and time the airtight wristband packaging was opened and the date and time when the wristband was placed into the provided return packaging), and log₁₀ transformed. Measurements of tris(1-chloro-2-propyl) phosphate (TCPP) isomers, TCPP1 (tris(chloroisopropyl) phosphate), TCPP2 (bis(2-chloro-1-methylethyl)(2-chloropropyl) phosphate), and TCPP3 (bis(2-chloropropyl)(2-chloro-1-methylethyl) phosphate), on each wristband were summed to a single value (hereinafter ‘TCPP’). Only 40 analytes with >60% detection were included in downstream analyses.

Socioeconomic and demographic factors

We collected covariate data from caregivers via questionnaire and from medical records. These included child sex, child race, child ethnicity, child age when the wristband was worn, household income at maternal enrollment, and maternal education at maternal enrollment. Household income at enrollment was reported categorically and modeled continuously according to the midpoint of each category.

COVID-19 pandemic restrictions

We used the date of school closures to operationalize restrictions related to the COVID-19 pandemic. In response to the COVID-19 pandemic, Governor Jay Inslee ordered all K-12 schools in Washington State to close effective March 17^th, 2020, and they would remain closed for the remainder of the academic year.⁴² We used the date and time data from the wristband questionnaire to categorize children as having worn wristbands before March 17, 2020, while schools were still open, or after this date, while the majority of schools were closed. In this study sample, the no pandemic restrictions group included children wearing their wristband as early as November 15, 2019 and as late as March 15, 2020, while the pandemic restrictions group included children wearing their wristband as early as June 25, 2020 and as late as March 1, 2021.

Child wheeze and asthma

Child respiratory outcomes were assessed using the International Study of Asthma and Allergies in Childhood (ISAAC) questionnaire. Caregivers responding to the questionnaire were asked if the child ever had wheezing or whistling in the chest at any time in the past (ever wheeze) or if the child ever had asthma (ever asthma). These questions were administered at two separate study visits scheduled at 4-6 and 8-10 years of age. Children were categorized as having ever asthma or ever wheeze if there were affirmative responses at either or both visits.

Statistical analyses:

We computed pairwise Pearson correlations among log₁₀ transformed chemical concentrations. We calculated descriptive summary statistics for covariates and created z-score sum variables for chemical exposures by expressing all chemicals in terms of standard deviations away from the mean (i.e., scaling and centering), and summing these z-scores for all chemicals and by chemical class. In complete case analyses, variables were modeled continuously or categorically according to the groupings in Table 1. We examined associations of child age, child sex, child race, child ethnicity, household income, maternal education, study site, and pandemic restrictions with chemical levels in silicone wristbands in separate multivariable linear regression models for each chemical class z-score sum and individual chemical analyte. We utilized multivariable models to adjust for confounding and maximize precision. Race and ethnicity were considered as proxies for social, political, and structural factors contributing to health inequities such as residential segregation and differential access to resources including healthcare. Models were also adjusted for the season when wristbands were worn as a precision variable.

Table 1:

Descriptive statistics of the study sample

	Overall (N=55)
Study visit
Age 4-6	25 (45.5%)
Age 8-10	30 (54.5%)
Child sex
Female (reference)	31 (56.4%)
Male	24 (43.6%)
Child age (years)
Mean (SD)	7.3 (1.4)
Range	5.0 - 9.5
Child race
N-Miss	3
White (reference)	42 (80.8%)
Asian	<5 (<8%)
American Indian or Alaska Native	<5 (<8%)
Multiple Race	5 (9.6%)
Other	<5 (<8%)
Child ethnicity
N-Miss	2
Not Hispanic or Latino (reference)	44 (83.0%)
Hispanic or Latino	9 (17.0%)
Household income (USD)
N-Miss	1
Mean (SD)	65972 (26548)
Range	10000 - 90000
Maternal education
Graduated college or technical school (reference)	29 (52.7%)
High school completion	10 (18.2%)
Some graduate work or graduate/professional degree	16 (29.1%)
Site
Seattle (reference)	38 (69.1%)
Yakima	17 (30.9%)
Season wristband worn
Fall (reference)	26 (47.3%)
Winter	21 (38.2%)
Summer	8 (14.5%)
COVID-19 pandemic restrictions
Before March 17, 2020 (schools open, reference)	7 (12.7%)
After March 17, 2020 (majority of schools closed)	48 (87.3%)
Ever had wheezing or whistling in the chest at any time in the past
N-Miss	15
No (reference)	27 (67.5%)
Yes	13 (32.5%)
Ever had asthma at any time in the past
N-Miss	15
No (reference)	36 (90.0%)
Yes	4 (10.0%)

We employed logistic weighted quantile sum regression^43,44 to assess associations between the entire mixture of chemicals with >60% detection and child wheeze. Using the gWQS R package,⁴⁵ we implemented random subset WQS, which has been shown to be more effective than bootstrap parameter estimation for high-dimensional mixtures.⁴⁶ The chemical mixture was transformed into deciles while using the same data for model training and validation. We performed 1,000 iterations with separate models to evaluate both positive and negative associations. We followed the same approach for mixtures analysis by chemical class, only including chemical classes with more than 2 individual analytes.

We additionally examined chemical exposures as predictors of child ever wheeze in separate models for each individual chemical analyte. We utilized Poisson regression models with robust standard errors to obtain relative risks (RRs) for child wheeze. WQS and Poisson models were adjusted for confounders, defined as potential predictors of chemical exposures and child wheeze, which included season along with the sociodemographic variables discussed above. We followed a similar approach for child ever asthma, although there were only 4 cases in our study sample. Owing to model convergence failures due to perfect predictions of this outcome, we present exploratory models for asthma that are not adjusted for covariates.

Results

Descriptive data

Among 55 children in our study sample, the majority were female (56.4%), self-identified as White (80.8%), self-identified as not Hispanic or Latino (83.0%), and wore their silicone wristbands while schools were closed due to the COVID-19 pandemic (87.3%) (Table 1). Among the 40 individual sub-sample with respiratory outcome data, ever wheeze was reported for 13 (32.5%) and ever asthma was reported for 4 (10.0%) children. The majority (69.1%) of children came from the urban Seattle site, and descriptive statistics according to study site are summarized in Supplemental Table 3.

Silicone wristbands were exposed to air in the participant’s environments for an average [range] of 5.10 [4.36-7.00] days, which was not appreciably different from the average of 5.05 [4.26-7.00] days that children wore their wristbands. Among 109 chemicals measured, 40 were detected in >60% of wristbands (Figure 1A, Supplemental Tables 1 and 2). Pairwise Pearson correlations revealed 482 (61.8%) statistically significant (P<0.05), positive correlations among chemicals, while there were no significant negative correlations (Figure 1B). Moreover, correlations were generally stronger between chemicals of the same class (Figure 1B). Distributions were approximately normal for most of the chemical class z-score sums (Figure 1C).

Figure 1: Characterization of chemicals measured in personal silicone wristbands.

Distributions of individual chemicals in silicone wristbands shown via boxplots with median and 1.5 interquartile range (IQR) whiskers (A). Pearson correlations among chemicals (× indicates non-significant correlation) (B). Histograms of chemical class z-score sums (C).

Associations of sociodemographic factors with chemical exposures

Compared with females, male child sex was associated with 14.6 [95% CI: 1.39 to 27.9] higher z-score sum of all chemicals, 3.93 [95% CI: 1.17 to 6.68] higher z-score sum of alkyl OPEs, and 3.67 [95% CI: 1.12 to 6.21] higher z-score sum of PAHs (Figure 2). Each one-year increase in child age was associated with 7.01 [95% CI: 0.887 to 13.1] higher z-score sum of all chemicals, 2.54 [95% CI: 0.925 to 4.15] higher z-score sum of triaryl OPEs, and 0.823 [95% CI: 0.321 to 1.32] higher z-score sum of pesticides (Figure 2). Evaluating chemicals individually revealed positive associations of male child sex with several alkyl OPEs, PBDEs, NBFRs, PAHs, and phthalates (Figure 3A). Elevated child age was positively associated with exposure to several individual triaryl OPE, PBDE, pesticide, and phthalate chemicals (Figure 3B). Compared with non-Hispanic/Latino children, Hispanic/Latino ethnicity was positively associated with TiBP (alkyl OPE), BBP (phthalate), and DEHP (phthalate) (Figure 3C). Compared with the urban Seattle study site, exposure to B4tBPPP (triaryl OPE) was higher among children in the rural Yakima study site (Figure 3D). No individual chemicals were significantly negatively associated with these demographic contrasts. Higher household income was significantly associated with lower BBP (phthalate) exposure, while maternal education was not significantly associated with any individual chemicals (Supplemental Figure 1).

Figure 2: Associations of sociodemographic factors with chemical z-score sums.

Associations of sociodemographic factors with z-score sums (of all chemicals and by chemical class). Coefficients and 95% confidence intervals were estimated from separate models for each chemical class. Multivariable linear regression models included the following covariates: child sex, child age, child ethnicity, child race, study site, household income, maternal education, season, and COVID-19 school closure status. Complete case analyses (N=51). Significant defined as P<0.05.

Figure 3: Associations of sociodemographic factors with individual chemical exposures.

Associations of male vs. female child sex (A), child age (B), Hispanic/Latino vs. non-Hispanic/Latino ethnicity (C), and rural Yakima vs. urban Seattle study site (D) with levels of individual chemicals measured in silicone wristbands. Coefficients and 95% confidence intervals were estimated from separate models for each individual chemical. Multivariable linear regression models included the following covariates: child sex, child age, child ethnicity, child race, study site, household income, maternal education, season, and COVID-19 school closure status. Complete case analyses (N=51). Significant defined as P<0.05.

Associations of COVID-19 related restrictions with chemical exposures

Wearing the silicone wristband after March 17, 2020, while the majority of schools were closed owing to COVID-19 restrictions, was associated with a lower z-score sum of triaryl OPEs (−7.94; 95% CI: −14.4 to −1.44) compared with wearing the wristband before this date (Figure 4A). Furthermore, wearing the silicone wristband after March 17, 2020, while the majority of schools were closed, was significantly associated with lower levels of three individual triaryl OPEs (2IPPDPP, B4tBPPP, and TPHP) and lower levels of one individual NBFR (BEHTBP) chemical (Figure 4B).

Figure 4: Associations of COVID-19 pandemic restrictions (participation after versus before March 17, 2020) with chemicals exposures.

Association of COVID-19 related pandemic restrictions, operationalized as child participation after versus before March 17, 2020, the date of Washington State school closures, with chemical class z-score sums (A) and levels of individual chemicals (B) measured in silicone wristbands. Coefficients and 95% confidence intervals were estimated from separate models for each individual chemical or z-score sum. Multivariable linear regression models included the following covariates: child sex, child age, child ethnicity, child race, study site, household income, maternal education, season, and COVID-19 restrictions group. Complete case analyses (N=51). Significant defined as P<0.05.

Associations of chemical exposures with child wheeze and asthma

Each one-unit increase in the WQS regression index, which corresponded to moving up 10 percentiles in the distribution of exposure, was suggestively associated with 2.11-fold [95% CI: 0.93 to 4.80] higher odds of child wheeze (Figure 5A). While this association constraining WQS in the positive direction was borderline significant (P=0.0738), the negatively constrained WQS regression provided no evidence for a link between higher exposure and lower odds of wheeze (P=0.385). Multiple different classes of chemicals were among the top weighted contributors to the mixture effect, including phthalates, pesticides, alkyl OPEs, PBDEs, and PAHs. The top three chemicals were the phthalate DEHP and the pesticides cis- and trans-permethrin (Figure 5A).

Figure 5: Associations of entire chemical mixture and individual chemicals with child ever wheeze.

Random subset logistic weighted quantile sum (WQS) regression with 1,000 iterations and separate models to evaluate both positive and negative associations of the chemical mixture with odds ratios (ORs) for child ever wheeze (A). Relative risks of ever wheeze associated with levels of individual chemicals (B). Relative risks and 95% confidence intervals were estimated from multivariable Poisson regressions with robust standard errors, in separate models for each individual chemical. All models were adjusted for the following covariates: child sex, child age, child ethnicity, child race, study site, household income, maternal education, season, and COVID-19 school closure status. Complete case analyses in sub-sample with respiratory outcome data (N=40). Significant defined as P<0.05.

WQS analyses by chemical class (per 10 percentile higher rank in the exposure distribution) revealed suggestively higher odds of child wheeze associated with higher exposure to the alkyl OPE mixture (1.80; 95% CI: 0.95 to 3.41) and the phthalate mixture (1.86; 95% CI: 0.96 to 3.62) (Supplemental Figure 2A). The negatively constrained WQS regressions provided no evidence for links between higher exposures and lower odds of wheeze (Supplemental Figure 2B). In analyses of individual chemicals (per 10-fold increase in exposure), we uncovered higher RRs of ever wheeze associated with the PAH chrysene (1.93; 95% CI: 1.07 to 3.49), the pesticide cis-permethrin (3.31; 95% CI: 1.23 to 8.91), and the phthalate DINP (5.40; 95% CI: 1.22 to 24.0) (Figure 5B). In a sensitivity analysis further adjusting for exposure to secondhand smoke, which was only reported for two children, these three associations remained significant, and no other individual chemicals were significantly associated with child wheeze (data not shown).

Exploratory models for associations between chemical exposures and child ever asthma that were not adjusted for covariates are presented in Supplemental Figure 3. As with ever wheeze, each one-unit increase in the unadjusted WQS regression index was suggestively associated with higher odds of ever asthma, and the phthalate DEHP and the pesticides cis- and trans-permethrin were among the top contributors to the mixture effect (Supplemental Figure 3A). In analyses of individual chemicals, we uncovered higher unadjusted RRs of ever asthma associated with higher levels of the alkyl OPE TCEP, BDEs 209 and 47, the pesticides cis- and trans-permethrin, and the phthalate DEHP (Supplemental Figure 3B).

Discussion

In this Washington state cohort study, we characterized exposure to numerous chemicals measured in a pilot sample of silicone wristbands worn continuously over 5 days by 55 children aged 5-9 years. Examining demographic factors, we found that male child sex and older child age were associated with higher exposure to many classes of chemicals, while Hispanic/Latino child ethnicity and participation at the rural Yakima study site (compared with urban Seattle) were associated with elevated exposures to just a few chemicals. Additionally, exposures to flame retardant chemicals including triaryl OPEs and NBFRs were suggestively lower among children wearing their wristbands during the COVID-19 pandemic restrictions, which included school closures. Finally, higher levels of three individual chemicals, cis-permethrin (pesticide), DINP (phthalate), and chrysene (PAH) were associated with higher risks of child ever wheeze, and we observed suggestive but not statistically significant evidence that higher exposure to the entire chemical mixture was also linked with child wheeze. Owing to our small sample size, this study should be considered as a pilot requiring replication. Results may be more robust with larger sample sizes.

Silicone wristbands were worn continuously for 5 days, providing a cumulative metric that may more effectively assess exposure compared with single time-point measurements that are typical of biomarker studies. While biomarker studies can enhance exposure assessment via multiple measurements, repeated sampling can be an economic or logistical challenge, impose burdens on participants, and may be prone to unwanted variability from factors such as time of day. Additionally, common substrates used for certain classes of chemicals could miss a large fraction of an individual’s total exposure. For example, high molecular weight PAHs are primarily excreted in the feces, a rarely utilized substrate in PAH biomonitoring studies.⁴⁷ Moreover, we found a putative link between the high molecular weight PAH chrysene and child wheeze, an association that we may not have detected with urinary PAH biomarkers.

Our results both corroborate and contradict prior associations between various demographic factors and chemical exposures. Ethnic disparities in phthalate exposures observed here are consistent with prior evidence – phthalate exposures were higher among Hispanic compared with non-Hispanic White participants in prior pooled analyses of 9⁴⁸ and 16⁴⁹ USA cohorts. On the other hand, the higher chemical exposures among males observed here conflicts with prior work. For instance, studies utilizing the National Health and Nutrition Examination Survey (NHANES) have shown higher levels of organophosphate flame retardants⁵⁰ and phthalates in females.⁵¹ Among adults, higher personal care product use among women could explain higher chemical levels. Among children, however, personal care products may have a lesser influence over chemical exposures. Additionally, these NHANES analyses characterized chemical in urine, which could account for disparate results. Although many studies have compared chemical measurements between wristbands and biospecimens and observed significant positive correlations, these relationships vary by study and among different subpopulations, and could depend on the sources and routes of exposure.³

We found lower levels of several triaryl OPEs and one NBFR, BEHTBP, in silicone wristbands worn by children after March 17, 2020, when Washington State closed schools in response to the pandemic. BEHTBP as well as triaryl OPEs including triphenyl phosphate (TPHP) and TPHP analogs with varying degrees of aryl isopropylation are components of one of the most common flame retardant mixtures, Firemaster^® 550, which is incorporated into various products to reduce the risk of fires.^52-54 One study detected Firemaster^® 550 components in 100% of dust samples collected from 40 early childhood education facilities in California, USA,⁵⁵ and implicated the foam in napping pads as a major source of exposure.^55,56 Additionally, previous reports have found higher levels of flame retardant chemicals in schools and cars compared with the home.^34,35 Thus, in our study, not going to school and spending less time in the car during the pandemic could have resulted in lower exposures to triaryl OPEs and NBFRs. Note, however, that we cannot rule out contributions from other behavioral and environmental changes during the COVID-19 pandemic, or other factors that change over time irrelevant of COVID-19. Additionally, the pandemic restrictions group included children wearing their wristband as late as March 1, 2021. To the best of our knowledge, children attending public schools in our sampling area had not returned to school by this date, but it is possible that children attending private schools in our study had already returned to in-person educational activities. Flame retardant chemical exposure research in children has predominantly focused on dust and air samples from the home, while far fewer studies have analyzed samples from cars, schools, daycares, or public facilities.⁵⁷ Our findings and the results of other studies highlight the need to further investigate sources of flame retardant chemical exposures in these settings and their consequences for children’s health.

To the best of our knowledge, this is the first study showing links between personal chemical exposures measured in silicone wristbands and respiratory health. Studies assessing chemical exposures using other methods have reported associations of childhood phthalate exposures with allergic diseases and adverse respiratory outcomes including asthma and wheeze.^58-64 However, findings from epidemiologic studies have been somewhat inconsistent and vary according to different phthalate congeners, with the most robust associations reported for high molecular weight phthalates such as DEHP, DINP and their metabolites.^63,64 Here, in individual chemical models, DINP was the only phthalate significantly associated with child wheeze. Furthermore, DEHP was the top weighted chemical in the WQS models showing suggestive associations between 1) the entire set of silicone wristband chemicals and 2) the overall phthalate mixture with child wheeze. An example of regrettable substitution, decreases in DEHP over time have been observed as the chemical has been replaced with alternative plasticizers that may be equally hazardous, including DEHT and DINP.^65-67 Among the phthalate and phthalate alternatives measured here, we observed the highest wristband concentrations for DEHT and DINP, followed by DEHP.

We also uncovered higher risk of ever wheeze associated with exposure to cis-permethrin, a pyrethroid insecticide. A prior systematic review and meta-analysis encompassing 101,353 participants from 11 countries found a positive association between occupational exposure to pesticides and obstructive pulmonary diseases.⁶⁸ Several studies have also reported associations between prenatal and postnatal pyrethroid pesticides measured in biospecimens and personal air samplers with adverse respiratory health.^69-74

We found that a higher silicone wristband concentration of the PAH chrysene was associated with higher risk of child wheeze. To the best of our knowledge, only four longitudinal studies have explored the link between prenatal PAHs and childhood asthma,²³ and evidence for a causal link between PAH exposure and adverse respiratory health remains inconclusive. For instance, one study in northern Manhattan, USA found that asthma was not associated with exposure to PAH alone, but was associated with combined prenatal exposure to PAH and environmental tobacco smoke.⁷⁵ A larger multi-site analysis across five USA cities suggested an adverse association of prenatal PAH exposure with child asthma among females only.²³ Links between prenatal PAHs and adverse respiratory health have also been reported in two Polish cohorts.^20,76,77 Studies of postnatal exposure are also uncommon. One study found a positive link between PAHs measured via ambient air pollution monitoring and wheeze among approximately 300 children with asthma in California, USA.²¹ In a study in the Czech Republic, higher ambient PAH exposure was associated with higher rates of bronchitis in 1,133 children.²⁴ Another study in central Poland found an association between child urinary PAHs and increased risk of food allergy.⁷⁷ Additional studies are needed to evaluate potential links between postnatal PAH exposures and child respiratory and allergic diseases.

Although there were some differences, chemical profiles in silicone wristbands here were generally similar to previously published results for these compounds.^38,78,79 Discrepancies in chemical levels and detection rates between different silicone wristband studies could be attributable to differences in study populations, time trends, geography, and/or chemical analysis methods (see Young et al. (2021) and Yin et al. (2024) for detailed comparisons of silicone wristband chemical profiles across studies). For example, a 2016 silicone wristband analysis in adult participants from North Carolina, United States, detected many PBDEs in 100% of samples, including BDEs 28, 47, 66, 85, 99, 100, 153, 154, and 209.⁸⁰ In our study, detection rates for these PBDE congeners ranged from 12.73-94.55%. Most of the children in our study wore their silicone wristbands after March 17, 2020. This later timing could explain our lower detection rates of PBDEs, which were phased out of production in the Unites States from 2004-2013.⁸¹ Even lower detection rates were found in a 2018-2019 study of Italian and French adults, where most BDE congeners were detected in fewer than 10% of silicone wristbands.⁸² As another example, dichlorodiphenyldichloroethylene (DDE), the primary breakdown product of the insecticide dichlorodiphenyltrichloroethane (DDT), was detected in 3.64% of the child silicone wristbands here. A similar 4.4% detection rate in silicone wristbands was found in children from North Carolina,⁸³ while another study in adult populations in the United States, the United Kingdon, India, and China found much higher detection rates from 38-81%.⁷⁹ Possibly explaining these higher detection rates in adults compared with children, DDE may persist in the environment and bioaccumulate in humans owing to its lipophilicity and long half-life.⁸⁴ However, another study in Uruguayan children detected DDE in 100% of child silicone wristbands, and further analysis revealed that some children had recent DDT exposure.⁸⁵ Thus, determinants of these chemical exposures is multifaceted, and may be jointly driven by factors such as age and other demographic variables, geographic region, and time.

Our study should be considered in the context of several limitations. Participants were highly educated and all came from Washington state, possibly limiting generalizability. Owing to our small sample size and lack of false discovery rate control, these results should be treated as hypothesis generating and requiring replication. Small sample sizes may have especially impacted our COVID-19 restrictions analysis, as only 7 children wore their wristbands before the COVID-19 related school closures in Washington State. Moreover, while we uncovered several associations of chemical exposures with child ever wheeze, we did not have sufficient asthma cases for analysis. Future studies should explore associations between chemicals measured in silicone wristbands and a broader range of respiratory outcomes including asthma and lung function measured objectively via spirometry. Additionally, pesticide-wheeze associations could be confounded by the presence of pets in the home – pets are linked with increased allergic respiratory morbidity, and pets positively correlate with pesticide exposures (possibly owing to pesticide flea and tick treatments).⁸⁶ Our respiratory outcomes analyses could also be subject to reverse causality, as participants wore their wristbands over the five days following the study visit. This limitation could be mitigated by the fact that silicone wristband measures reflect average exposures over time worn, better capturing typical exposure levels experienced by children compared with spot biomarker measurements. Our targeted method measured a limited number of SVOCs, while an untargeted method could have been used to capture more numerous and potentially novel chemicals. Our targeted method was insensitive for certain analytes of interest, such as nicotine. However, in general, targeted methods can capture lower levels of exposure, a considerable strength in epidemiologic studies of non-occupational exposures. Finally, our results could also be affected by residual confounding.

Highlights.

• First study of associations between chemical exposures measured in personal silicone wristbands and child respiratory health.

• Higher chrysene, cis-permethrin, and di-isononyl phthalate were associated with higher relative risks of child wheeze.

• Logistic weighted quantile sum regression revealed suggestive positive association between the entire mixture of 40 chemicals with child wheeze.

• COVID-19-related school closures were associated with lower wristband concentrations of brominated and triaryl organophosphate ester flame retardants.