Association of Exposure to Endocrine-Disrupting Chemicals During Adolescence With Attention-Deficit/Hyperactivity Disorder–Related Behaviors

Key Points Question Is exposure to endocrine-disrupting chemicals during adolescence a risk factor for behaviors associated with attention-deficit/hyperactivity disorder (ADHD)? Findings In this cross-sectional analysis of data from 205 adolescents participating in a prospective birth cohort study, exposure to select phthalates was associated with an increased risk of ADHD-related behaviors. Associations were strongest for phthalates with antiandrogenic activity. Meaning These findings support the potential importance of exposures to endocrine-disrupting chemicals, especially phthalates, during adolescence as correlates of ADHD-related behaviors.


Introduction
Attention-deficit/hyperactivity disorder (ADHD) is the most common neurobehavioral disorder of childhood, affecting approximately 9.4% of children in the United States. 1 Attention-deficit/ hyperactivity disorder is characterized by difficulty maintaining attention, controlling impulses, and regulating activity level and may be associated with academic and social problems as well as difficulties throughout adulthood. [2][3][4][5] There is growing evidence that exposure to endocrine-disrupting chemicals (EDCs), such as phthalates and phenols, may be associated with ADHD. [6][7][8][9] Phthalates and phenols are widely used in consumer products, including food processing and packaging equipment, personal care products (eg, cosmetics and fragrances), and pharmaceuticals, resulting in ubiquitous exposure. 6,10,11 Although epidemiologic studies have reported associations between prenatal and early childhood exposure to EDCs and ADHD-like behaviors, 8,12,13 few studies have examined the association of exposure to these chemicals during adolescence with ADHD-like behaviors. Similar to the prenatal period, adolescence is a critical time for brain development, characterized by structural and functional changes in the brain as well as the onset of behavioral problems, some of which may be due to hormonal changes. 14 As such, exposure to EDCs during adolescence may be particularly detrimental.
The objective of this study is to examine the association of adolescent exposure to EDCs, specifically phthalates, phenols, and triclocarban, with ADHD-related behaviors. We hypothesize that exposure to certain EDCs during adolescence is particularly detrimental to adolescent behavior given the rapid brain development that occurs during this time.

Study Population
We analyzed data from the New Bedford Cohort, a prospective birth cohort of mother-infant pairs recruited after delivery at a New Bedford, Massachusetts, hospital between 1993 and 1998. 15 The original aim of the study was to examine the association of prenatal organochlorine and metal exposures with subsequent neurodevelopment among children living near the New Bedford Harbor Superfund site. Eligibility requirements included newborns who were vaginally delivered to mothers who were at least 18 years of age and who were available for neonatal examination. In-person neurodevelopmental testing was performed when participants were approximately 15 years of age between 2008 and 2014. Of the 788 newborns enrolled in the New Bedford Cohort, 660 (84%) met eligibility criteria (residence in the study region, absence of catastrophic central nervous system injury or disease, available biomarkers of early chemical exposure, and contact information) for these assessments and, of those newborns, 528 (80%) completed assessments. Midway through the 15-year follow-up, the study added assessment of exposure to EDCs. We invited the 252 participants who were evaluated in this last half of the study to provide spot urine samples at 2 points: during neurodevelopmental assessments and approximately 1 week later (mean duration of time after assessment, 7 days [range, 1-35 days]) (eFigure in the Supplement). This study followed the Brigham and Women's Hospital (Boston, Massachusetts). Written informed parental consent and child assent were obtained from participants. The analysis of deidentified specimens at the Centers for Disease Control and Prevention (CDC) laboratory (Atlanta, Georgia) was determined not to constitute engagement in human participants research.

Urine Sample Collection and EDC Measurement
Urine sample collection and analysis have been described elsewhere. 16 In brief, of the 252 adolescents invited to provide urine samples, 205 (81%) provided at least 1 sample. Urine samples were collected in sterile polypropylene cups and frozen until processing and analysis. 16 It was not possible to analyze all urine specimens, owing to resource limitations; of the 144 adolescents (57%) who provided 2 samples, 60 had each sample analyzed separately, and the mean concentration was used in analyses. For the remaining 84, equal volumes of the 2 samples were pooled for analysis. 16 Specific gravity was measured using a digital refractometer (Pocket PAL-10S; ATACO USA Inc).
Because batch 1 analyses did not include all biomarkers, urinary concentrations of all analytes were available for 178 participants (87%), with the remaining 27 participants missing concentrations of MHINCH, monohydroxyisobutyl phthalate (MHiBP), monoisononyl phthalate (MNP), monohydroxybutyl phthalate (MHBP), bisphenol F (BPF), bisphenol S (BPS), ethylparaben, and triclocarban. The limits of detection for the target analytes ranged from 0.2 to 2.3 μg/L; for concentrations less than the limit of detection, instrument readings were used in data analyses. The study-specific urine quality assurance and quality control samples for both between-batch and within-batch analyses demonstrated the excellent reproducibility of the analytic chemistry methods used. 16

Exposure Assessment
Because people are routinely exposed to multiple EDCs, we created summary exposure measures by combining biomarker concentrations for chemicals with a shared mechanism of action, exposure pathway, or chemical class. As described elsewhere, 16  in units of micromoles per liter) was the molar sum of 2,4-dichlorophenol and 2,5-dichlorophenol. of 0.85 to 0.96 for the CADS and 0.74 to 0.95 for the BASC-2 (hyperactivity and attention problem scales), depending on the reporter. The BASC-2 scales are moderately correlated (r ෂ 0.5-0.6) with the Conners ADHD index, which, in turn, has excellent discriminant validity for ADHD diagnosis. 24,25 Parents and adolescents completed these checklists at the study visit and first urine sample collection, while teachers completed them a median (SD) of 2.5 (6.6) months after the first urine sample collection. The BASC-2 consists of 139 to 176 questions, and the CADS consists of 14 to 15

Covariates
Data on parental demographic charateristics (race/ethnicity, educational level, and income), health history, and tobacco use during pregnancy were collected in a questionnaire administered to mothers approximately 2 weeks after delivery. When the child was 15 years of age, the data on demographic characteristics, smoking habits, and child medical history (including behavioral disorder diagnoses and medication use) were updated via medical record reviews and questionnaires. Selfreported adolescent tobacco, alcohol, and marijuana use was ascertained with the CDC Youth Risk Behavior Survey. Umbilical cord serum levels of polychlorinated biphenyls and dichlorodiphenyldichloroethylene were used as biomarkers of prenatal organochlorine exposures.
Childhood blood lead levels were abstracted from medical records of lead exposure screening; the maximum concentration between 12 and 36 months of age was calculated.
Using a structured diary, adolescents recorded food consumption and personal care product use for the 24 hours prior to urine sample collection. Diary data were used to estimate exposure risk factors during that 24-hour period, including the number of personal care products used and caffeinated beverages consumed, as well as the amount of fast food or canned food consumed.
Height and weight were measured at the time of urine sample collection, and body mass index z scores were calculated using CDC US reference data. 26

Statistical Analysis
Statistical analyses were performed from January 15 to December 31, 2019. Regression diagnostics supported log 2 transformation of urinary biomarker concentrations. We dichotomized outcomes at the 98th percentile, which BASC-2 and CADS guidelines define as indicative of significant behavioral problems. To leverage the multiple indices of ADHD-related behavior, maximize power, and avoid multiple comparisons, we used a repeated-measures analysis design, considering each of the 14 binary indices as a repeated measure reflecting a single underlying outcome. 27 behavioral outcomes, and biomarker concentrations in this analysis. 30 We performed additional sensitivity analyses, including adjusting for umbilical cord serum polychlorinated biphenyls and dichlorodiphenyldichloroethylene, and adjusting for 12-month to 36-month maximum blood lead levels, because exposure to these chemicals hase been associated with ADHD-related behaviors in the New Bedford Cohort. We adjusted for whether adolescents had

Results
The  (Table 1). Individual biomarker concentrations among study participants were similar to those observed in US adolescents in the 2011-2012 National Health and Nutrition Examination Survey. 31 Of the 205 adolescents in this analysis, 82 (40%) had scores consistent with a significant behavioral problem as defined by at least 1 BASC-2 or CADS ADHD-related measure ( Table 3), whereas 39 (19%) had an ADHD diagnosis (Table 2), which is higher than US population estimates (approximately 10%). 1 In covariate-adjusted models, summary measures of phthalates were consistently associated with an increased risk of significant ADHD-related behavior problems (Figure; Table 4)

JAMA Network Open | Environmental Health
Endocrine-Disrupting Chemical Exposure in Adolescence and ADHD    Associations tended to be stronger among male adolescents than among female adolescents, most notably with respect to the ΣDEHP metabolites (P = .002 for interaction), in which a 2-fold increase corresponded to a 1.62 (95% CI, 1.38-1.91) increased risk of significant ADHD-related behavior problems in male adolescents compared with a 1.06 (95% CI, −1.15 to 1.33) increased risk in female adolescents ( Figure; eTable 2 in the Supplement). The remaining sex-specific differences were estimated with less precision.
A 2-fold increase in Σdichlorophenols was also associated with an increased risk of ADHDrelated behaviors (RR, 1.15; 95% CI, 1.01-1.32) (Table 4) When considering the more inclusive threshold for those with possible significant behavioral problems (>85th percentile), we observed similar, albeit slightly attenuated, findings. For example, each 2-fold increase in Σantiandrogenic phthalates was associated with a 1.21 (95% CI, 1.00-1.47) increased risk of possible significant (vs 1.34 for clearly significant) ADHD-related behavior problems (Table 4), and this difference appeared to be more pronounced among male adolescents (eTable 2 in the Supplement).
Associations were similar, but more precise, in models imputing missing data (eTable 4 in the Supplement). Associations were stronger for summary phthalate measures when we excluded adolescents with an ADHD diagnosis. We did not observe substantive changes in results with the other sensitivity analyses, including exclusion of those taking medications (eTables 4 and 5 in the Supplement).

Discussion
In a population with a high prevalence of ADHD-related behaviors, our findings support an association of adolescent exposure to EDCs, particularly select phthalates, with an increased risk of significant ADHD-related behavior problems at exposure biomarker concentrations typical of adolescents in the general US population. These results are consistent with previously reported findings in the New Bedford Cohort associating adolescent EDC exposure, particularly to antiandrogenic phthalates, with increased externalizing behaviors (a BASC-2 composite of hyperactivity, aggression, and conduct problems). 16 Previous studies have reported associations of prenatal and early postnatal exposure to phthalates and BPA with an increase in ADHD-related behaviors. However, to our knowledge, no other published studies have analyzed associations between adolescent EDC biomarker concentrations and ADHD-related behaviors. Adolescence is a critical period for brain development and may be another time of heightened vulnerability to EDC exposure. Several cross-sectional studies (with children of wide age ranges, including some adolescents) have reported associations between increased urinary phthalate and BPA concentrations and ADHD-related behaviors. 5,7,9,32 Experimental animal studies also support the potential for EDC exposures to be associated with adverse behavior in adolescence, including behaviors consistent with ADHD. [33][34][35][36] Furthermore, there is well-established evidence demonstrating the unique association that neurotoxicant exposures (eg, alcohol and organophosphates) during adolescence have with adverse behavior. 37,38 This analysis focused on ADHD-related behaviors as opposed to clinical ADHD diagnosis. By relying on multiple indices of ADHD-related behavior (from the BASC-2 and CADS) from multiple observers (teachers, parents, and self), we were able to leverage a uniquely comprehensive characterization of adolescent behavior, including behaviors that can vary over time (eg, inattention) and thus may be susceptible to variations in exposure risk factors. In particular, combining information from these multiple measures increased the power of our analysis, resulting in more stable estimates of association than would have been possible by examining each behavioral measure individually. In addition, by expressing multiple measures as a single outcome, we did not induce the statistical issues associated with multiple comparisons.

Limitations
This study has some limitations. The assessed EDCs have short elimination half-lives, as well as shortterm variability in exposure, resulting in the potential for nondifferential exposure misclassification, which could bias results toward the null. However, our study collected urine samples at 2 different times from most participants, allowing us to better characterize each participant's mean or usual exposure. Given the cross-sectional design of this analysis, another challenge is the potential for reverse causation, such that existing behavioral problems may alter habits that enhance EDC exposure via changes in, for instance, diet, substance use, and/or personal care product use. [39][40][41][42] However, adjustments for these factors did not affect associations, suggesting that reverse causation from these habits was unlikely. Furthermore, excluding adolescents with a diagnosis of ADHD enhanced the existing associations (eTable 4 in the Supplement). Although data on participants' prenatal EDC exposure are lacking, we do not expect this lack of data would confound associations with adolescent exposure; weak within-person correlations have been reported between prenatal and childhood exposures to phthalates and BPA, and we would expect similar or weaker correlations between prenatal and adolescent time periods. 43,44 In addition, participants were from a small geographical region, which may potentially affect the generalizability of our findings.

Conclusions
Our results support the importance of adolescent exposure to EDCs, particularly phthalates, as a potential risk factor for significant ADHD-related behavior problems. Attention-deficit/hyperactivity disorder is a common and costly neurobehavioral disorder; US health care and education expenditures associated with ADHD in children and adolescents are estimated to be $38 billion to $72 billion annually, while costs associated with ADHD in adults, including lost wages, are much higher, ranging from $143 billion to $266 billion annually. 2,45 The identification of modifiable risk factors for ADHD is of great public health importance. These findings contribute new insights into the potential detrimental neurobehavioral outcomes of EDC exposure during adolescence.