Environmental Exposure to Pyrethroid Pesticides in a Nationally Representative Sample of U.S. Adults and Children: the National Health and Nutrition Examination Survey 2007-2012

Abstract

Pyrethroids are an important class of insecticides, and thousands of tons of these compounds are used in the United States every year. This study characterized exposures to pyrethroids and assessed demographic, socioeconomic, and lifestyle factors that modulate pyrethroid exposure using data from the National Health and Nutrition Examination Survey (NHANES) 2007-2012, a nationally representative survey of the non-institutionalized population of the United States. Urinary levels of commonly used biomarkers of pyrethroid exposure, including 3-phenoxybenzoic acid (3-PBA), 4-fluoro-3-phenoxybenzoic acid (F-PBA), and cis-dibromovinyl-dimethylcyclopropane carboxylic acid (DBCA), were determined by liquid chromatography-tandem mass spectrometry. The detection rate of 3-PBA, a nonspecific metabolite of several pyrethroids, was 78.1% in adults (N = 5,233) and 79.3% in children (N = 2,295). The detection rates of all other pyrethroid metabolites were <10 %. The median urinary level of 3-PBA in adults was 0.47 μg/L (interquartile range, 0.14-1.22 μg/L). For children, the median urinary level was 0.49 μg/L (interquartile range, 0.17-1.29 μg/L). Age, gender, family income-to-poverty ratio (PIR), levels of physical activity, alcohol intake, and body mass index were associated with 3-PBA levels in adults. In children, age, gender, race/ethnicity, and PIR were associated with 3-PBA levels. 3-PBA levels also differed significantly across NHANES cycles, with higher levels observed in NHANES 2011-2012. Geometric mean 3-PBA levels in U.S. adults were 0.41 μg/L in NHANES 2007-2008, 0.41 μg/L in NHANES 2009-2010, and 0.66 μg/L in NHANES 2011-2012. In U.S. children, geometric mean 3-PBA levels were 0.40 μg/L in NHANES 2007-2008, 0.46 μg/L in NHANES 2009-2010, and 0.70 μg/L in NHANES 2011-2012. These results demonstrate that pyrethroid exposures remain a current environmental health concern and lay the foundation for further preclinical and epidemiological studies assessing human health risks associated with pyrethroids.

Capsule:

Exposure to pyrethroid pesticides was widespread in a nationally representative sample of U.S. adults and children and varied by demographic, socioeconomic, and lifestyle factors.

Graphical Abstract

1. Introduction

Pyrethroids are the second most-used group of insecticides in the world and account for more than 30% of insecticide use worldwide (Bhardwaj et al., 2020). Large quantities of these insecticides are applied every year in agricultural and residential settings. This class of insecticides is based on pyrethrins, a class of naturally occurring insecticides found in the flowers of Chrysanthemum cineraraefolum. Synthetic pyrethroids have increased insecticidal activity and are less light-sensitive than pyrethrins (Bradberry et al., 2005). Pyrethroids are often found in consumer products, including garden insecticides, pet shampoos, lice treatments as well as mosquito repellents that are applied to clothing. They are also used to manage pests, for example, of cotton, corn, soybeans, vegetables, and fruits (USGS, n.d. [not dated]). Pyrethroids have, to some extent, replaced organophosphate insecticides that exhibit toxicity at relatively low levels and pose a human health concern (Schleier and Peterson, 2011). Approximately sixteen pyrethroid pesticides are on the market in the United States (U.S.), with cypermethrin, cyfluthrins, esfenvalerate, fenpropathrin, and permethrin being major pyrethroids used in agriculture, based on usage estimates by the United States Geological Survey (USGS) (USGS, n.d.).

Humans are exposed to pyrethroid pesticides through the diet and by residential and occupational pyrethroid pesticide use (Lee et al., 2020; Rauch et al., 2018; Riederer et al., 2008). Several pyrethroid pesticides are quickly absorbed and distributed to tissues, where they are metabolized by esterases or cytochrome P450 enzymes to form metabolites, such as 3-PBA (3-phenoxybenzoic acid). Depending on the specific pyrethroid, other metabolites include DCCA (trans-dichlorovinyl-dimethylcyclopropane carboxylic acid), F-PBA (4-fluoro-3-phenoxybenzoic acid), and DBCA (cis-dibromovinyl-dimethylcyclopropane carboxylic acid) (Leng et al., 1997; Oulhote and Bouchard, 2013). Structures of relevant pyrethroids and the corresponding metabolites used in human biomonitoring studies are shown in Figure 1. These general metabolites are rapidly excreted with the urine and are used by human biomonitoring studies as biomarkers of exposure (Barr et al., 2010). For example, the half-life of the pyrethroids cyfluthrin and cypermethrin were approximately 5.52 and 8.09 hours, respectively (Kuhn et al., 1999). Humans may also be exposed to 3-PBA, formed by environmental degradation of pyrethroids, via the diet or by ingestion of household dust (Chen et al., 2012; Starr et al., 2008).

Figure 1.

Names and structure of pyrethroid pesticides used in U.S. agriculture and the corresponding metabolites analyzed by NHANES 2007-2012. In addition, d-phenothrin (not shown), a pyrethroid pesticide used in residential applications, forms the nonspecific pyrethroid metabolite 3-PBA.

The insecticidal activity of pyrethroids involves their interaction with voltage-gated sodium channels (VGSCs), causing hyper-depolarization of the nerve membrane and neurotransmitter release, ultimately resulting in death (Field et al., 2017). Pyrethroids have a much lower affinity for mammalian VGSCs and, thus, have low acute toxicity in mammals. However, exposure to pyrethroids may cause toxicity in mammals by mechanisms involving, for example, oxidative stress, inflammation, and mitochondrial dysfunction (ATSDR, 2003). Pyrethroids are only weakly estrogenic and have no androgenic potential based on in vitro screening assays (reviewed in Saillenfait et al., 2016). Exposure of pregnant mice to α-cypermethrin and stress affects embryonic neurodevelopment (Elser et al., 2020). These effects are caused by indirect mechanisms involving the placenta and maternal liver. In humans, pyrethroid exposure is linked to adverse effects on thyroid hormones (Hu et al., 2019; Hwang et al., 2019). There is also evidence that adverse outcomes associated with pyrethroid exposure may be mediated by the gut microbiome (Nasuti et al., 2016) by altering the toxicokinetics of pyrethroids or by affecting the composition and function of the microbiome (Chiu et al., 2020; Nichols et al., 2019).

An analysis of NHANES 1999-2002 data demonstrated widespread exposure of the general U.S. population to pyrethroids, such as permethrin and cypermethrin (Barr et al., 2010). Other biomonitoring studies from across the world also document near-ubiquitous environmental exposure to pyrethroids in children (Ding et al., 2012; Roca et al., 2014; Rodriguez et al., 2012; Trunnelle et al., 2014a; Trunnelle et al., 2014b; van Wendel de Joode et al., 2016) and adults (Jurewicz et al., 2016; Jurewicz et al., 2015; Motsoeneng and Dalvie, 2015; Mwanga et al., 2016; Radwan et al., 2015; Radwan et al., 2014; Wielgomas et al., 2013; Yoo et al., 2016). Epidemiological studies implicate exposure to pyrethroids in impaired neurodevelopment (Wagner-Schuman et al., 2015), Parkinson disease (Baltazar et al., 2014), altered semen quality (Jurewicz et al., 2016; Jurewicz et al., 2015; Radwan et al., 2014), and an increased risk of cardiovascular disease (Han et al., 2017) and diabetes (Park et al., 2019). We recently reported a positive association between pyrethroid exposure and an increased risk of all-cause and cardiovascular disease mortality in the NHANES 1999-2002 population (Bao et al., 2020). However, more preclinical and epidemiological studies are needed to characterize adverse outcomes following pyrethroid exposure across the lifetime and to assess interactions of pyrethroid exposure with other factors, such as stress and the microbiome.

Robust preclinical studies investigating pyrethroid × environment interactions need to be based on current, real-world exposures of the general U.S. population. Therefore, the present study investigates urinary levels of 3-PBA and other pyrethroid metabolites in recent data from the National Health and Nutrition Examination Survey (NHANES) and compares these data to other human biomonitoring studies from the same period. Besides, demographic and lifestyle factors that may be associated with urinary levels of 3-PBA, as well as temporal trends, were analyzed in U.S. adults and children. How these factors affect pyrethroid exposures in U.S. children remains mostly unknown.

2. Materials and Methods

2.1. NHANES survey

NHANES is a nationally representative survey of the non-institutionalized U.S. population. This survey is administered by the National Center for Health Statistics at the Centers for Disease Control and Prevention (CDC). Information about the demographics, socioeconomic status, diet, lifestyle, and medical conditions of study participants is collected as part of NHANES. In addition, extensive health examinations are performed, and specimens for laboratory tests are collected from study participants. NHANES data are released in 2-year cycles. NHANES has been approved by the National Center for Health Statistics Ethics Review Board, and written informed consent is obtained from all participants. More detailed information about NHANES can be obtained elsewhere (CDC, n.d.-a). For this study, we analyzed data from the NHANES 2007-2012 cycles, the most recent NHANES cycles for which urinary levels of pyrethroid metabolites are available. The final study population consisted of 2,295 U.S. children and 5,233 U.S. adults who had data available on urinary concentrations of pyrethroid metabolites. A summary of the characteristics of the NHANES 2007-2012 study population is provided in Table 1.

Table 1.

Subject demographics and characteristics.

Characteristics	Age, years
	6-11	12-19	20-39	40-59	≥ 60
Number of participants	1152	1143	1735	1772	1726
Gender, % (SE)
Male	50.8 (1.9)	51.4 (1.9)	49.4 (1.5)	48.4 (1.5)	44.8 (1.2)
Female	49.2 (1.9)	48.6 (1.9)	50.6 (1.5)	51.6 (1.5)	55.2 (1.2)
Race/ethnicity,a % (SE)
Non-Hispanic white	55.6 (2.6)	57.7 (3.3)	59.6 (2.6)	69.3 (2.6)	77.9 (1.8)
Hispanic	22.2 (2.2)	20.0 (2.2)	19.4 (2.0)	12.3 (1.4)	7.4 (1.1)
Non-Hispanic black	14.2 (1.4)	14.7 (1.7)	12.9 (1.2)	11.3 (1.2)	8.7 (1.0)
Other	8.0 (1.0)	7.6 (1.0)	8.2 (1.0)	7.1 (0.9)	6.1 (1.0)
Family income to poverty ratio (PIR),b % (SE)
≤ 1.30	32.5 (2.1)	30.0 (2.5)	26.4 (1.6)	17.9 (1.2)	17.7 (1.7)
1.31-3.50	36.3 (1.8)	30.4 (2.6)	33.2 (1.7)	27.7 (1.5)	36.2 (1.7)
> 3.50	24.2 (2.0)	31.7 (2.9)	33.2 (2.1)	47.2 (2.0)	37.9 (2.1)
Missing	6.9 (1.2)	7.9 (1.2)	7.1 (0.7)	7.3 (0.6)	8.2 (1.0)

All the variables were adjusted using population weights for the sample in which 3-PBA concentrations were measured except the number of participants.

^aRace/ethnicity was categorized based on self-reported data into Hispanic (including Mexican and non-Mexican Hispanic), non-Hispanic white, non-Hispanic black, and other race/ethnicity.

^bPIRs were categorized as ≤ 1.30, 1.30-3.50, and > 3.50.

SE, standard error.

2.2. Analysis of pyrethroid metabolites in urine samples

Spot urine samples were collected throughout each 2-year cycle, shipped to the National Center for Environmental Health at the CDC, and stored at −20°C until analysis. Urinary concentrations of 3-PBA, F-PBA, and DBCA were quantified with solid-phase extraction-high-performance liquid chromatography-heated electrospray ionization tandem mass spectrometry after deconjugation β-glucuronidase type H-1 from Helix pomatia. A detailed description of the analysis, including relevant quality assurance and quality control information, has been reported previously by the CDC (CDC, n.d.-c; Davis et al., 2013). The detection limits were 0.10 ng/mL, 0.10 ng/mL, and 0.50 ng/mL for 3-PBA, F-PBA, and DBCA, respectively. Because of its high prevalence and past use as a biomarker (Wagner-Schuman et al., 2015), we focused on 3-PBA for this analysis. A short overview of urinary F-PBA and DBCA levels is included to provide a more complete insight into pyrethroid exposures in the general U.S. population. An imputed fill value was assigned by CDC staff as the lower limit of detection (LLOD) divided by the square root of 2 for urinary 3-PBA levels below the LLOD.

2.3. Potential factors influencing urinary 3-PBA levels

NHANES used standardized questionnaires to collect information of age, gender, race/ethnicity, education, PIR, smoking, physical activity, alcohol intake, and BMI from adult survey participants, and age, gender, race/ethnicity, PIR, and BMI from participating children. This information was used to assess an association with pyrethroid exposures using urinary levels of 3-PBA, F-PBA, and DBCA as biomarkers of exposure. For the purpose of this study, race/ethnicity was categorized as non-Hispanic white, Hispanic (Mexican and non-Mexican Hispanic), non-Hispanic black, and other race/ethnicity. Education was grouped as less than high school, high school, and college or higher. The PIR was categorized as ≤ 1.30, 1.31-3.50, and > 3.50 (Johnson et al., 2013). Education was grouped as less than high school, high school, and college or higher. Adults were classified as never smokers (smoked < 100 cigarettes in their lifetime), current smokers (had smoked > 100 cigarettes and smoked cigarettes at the time of the survey), and ever smokers (had smoked > 100 cigarettes but did not smoke at the time of the survey). Alcohol intake was categorized as non-drinker (0 g/d for both male and female), moderate drinker (0.1-27.9 g/d for male, 0.1-13.9 g/d for female), and heavy drinker (≥ 28 g/d for male, ≥ 14 g/d for female). Physical activity was assessed by the Global Physical Activity Questionnaire, and metabolic equivalents of task (MET) minutes per week were derived to take both the duration and intensity of different activities into account (WHO, n.d.). Weight and height were measured by trained health technicians, and the body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. The BMI was classified as normal weight (adults: BMI < 25 kg/m²; children: BMI < 85^th percentile), overweight (adults: 25 kg/m² < BMI ≤ 29.9 kg/m²; children: 85^th percentile ≤ BMI < 95^th percentile), and obesity (adults: BMI ≥ 30 kg/m²; children: BMI ≥ 95^th percentile) based on the CDC BMI-for-age growth charts (Kuczmarski et al., 2002).

2.4. Statistical analysis

NHANES uses a complex, multistage probability sampling design to represent the national civilian non-institutionalized population of the United States. In this study, appropriate published sampling weights from the CDC were applied to account for the differential probability of selection, non-response adjustment, and adjustment to independent population controls. The Taylor series linearization method was used for variance estimation to account for stratification and clustering, following the NHANES Analytic Guidelines (CDC, n.d.-b). We calculated both the initial concentrations and the creatinine adjusted concentration of 3-PBA in this study (Barr et al., 2005). The median and geometric mean of urinary concentrations of 3-PBA according to population characteristics were computed for children and adults, respectively. Analysis of variance (ANOVA) was used to compare differences of log-transformed urinary concentrations of 3-PBA among various categorical variables (e.g., age, gender, race/ethnicity, and PIR). In addition, linear regression analyses were conducted by including all the variables simultaneously in the model to detect the independent effects of each variable. We also calculated the changes in urinary concentrations of 3-PBA from 2007-2010 to 2011-2012. The temporal trend was assessed by including the survey year as a continuous variable in linear regression analyses. Important demographic factors, i.e., age, sex, and race/ethnicity, were adjusted in the models. All statistical analyses were performed with SAS software (version 9.4; SAS Institute). P < 0.05 was considered statistically significant. Data discussed below were based on the analysis of volume adjusted data as recommended by the CDC. We also evaluated creatinine-adjusted data as supplementary material (Tables S1–S6).

3. Results and Discussion

3.1. Urinary levels of 3-PBA in U.S. adults

3-PBA was detected in most urine samples from this adult population. The overall detection rate was 78.1% (N = 5,233; Table 2). Geometric means and median levels of 3-PBA in these spot urine samples were 0.48 μg/L (standard error [SE], 0.01 μg/L) and 0.47 μg/L (interquartile range [IQR], 0.14-1.22 μg/L). Urinary 3-PBA concentrations were over an order of magnitude higher in the 95^th percentile of the NHANES 2007-2012 population. The geometric means and median levels of 3-PBA in this more highly exposed subpopulation (N = 262) were 12.60 μg/L (SE, 0.54 μg/L) and 10.57 μg/L (IQR, 7.33-18.71 μg/L).

Table 2.

Urinary concentration of 3-PBA (μg/L) in U.S. adults in the overall NHANES 2007-2012 population and the 95^th percentile.

Variable	NHANES 2007-2012					95th percentile
	N	Detection rate, % (SE)	Geometric means (SE)	Median (P25-P75)	P	N	Geometric means (SE)	Median (P25-P75)	Min-Max	P
All	5233	78.1 (0.9)	0.48 (0.01)	0.47 (0.14-1.22)		262	12.60 (0.54)	10.57 (7.33-18.71)	5.88-91.40
Age
20-39 years	1735	79.5 (1.3)	0.48 (0.02)	0.50 (0.16-1.09)	0.48	77	12.54 (1.15)	11.42 (6.80-18.79)	5.91-83.10	0.67
40-59 years	1772	77.6 (1.3)	0.50 (0.02)	0.45 (0.14-1.29)		108	12.27 (0.81)	10.26 (7.83-17.15)	5.90-58.70
≥ 60 years	1726	76.6 (1.8)	0.45 (0.03)	0.45 (0.13-1.23)		77	13.46 (1.52)	10.67 (7.08-25.18)	5.88-91.40
Gender
Male	2521	79.4 (0.9)	0.50 (0.02)	0.48 (0.17-1.19)	0.27	125	12.95 (0.78)	11.49 (7.41-17.70)	5.90-91.40	0.53
Female	2712	76.8 (1.3)	0.47 (0.02)	0.46 (0.13-1.23)		137	12.29 (0.67)	9.68 (7.32-19.28)	5.88-83.10
Race/ethnicitya
Non-Hispanic white	2325	76.3 (1.2)	0.47 (0.02)	0.44 (0.12-1.21)	0.34	140	12.63 (0.65)	10.20 (7.31-18.87)	5.88-91.40	0.79
Hispanic	1357	79.7 (1.4)	0.45 (0.02)	0.46 (0.16-1.04)		40	13.34 (1.28)	12.73 (7.66-18.58)	5.94-60.07
Non-Hispanic black	1102	82.8 (1.9)	0.57 (0.03)	0.58 (0.22-1.38)		59	11.91 (0.93)	11.62 (7.45-16.91)	5.97-71.86
Other	449	84.5 (2.1)	0.59 (0.06)	0.59 (0.21-1.68)		23	12.57 (2.26)	10.08 (7.24-21.04)	5.90-43.90
Educationb
Less than high school	1431	79.5 (1.8)	0.49 (0.02)	0.48 (0.17-1.14)	0.82	60	12.88 (1.17)	13.06 (8.08-17.78)	6.16-43.90	0.84
High school	1178	78.4 (1.3)	0.49 (0.03)	0.49 (0.13-1.34)		51	13.24 (1.14)	10.99 (6.75-21.42)	5.88-83.10
College or higher	2624	77.5 (1.1)	0.48 (0.02)	0.46 (0.14-1.20)		151	12.40 (0.63)	10.18 (7.33-18.46)	5.90-91.40
Family income to poverty ratio (PIR)c
≤ 1.30	1579	82.7 (1.3)	0.55 (0.02)	0.57 (0.21-1.35)	0.03	72	13.34 (1.16)	10.88 (7.37-18.81)	5.94-83.10	0.69
1.31-3.50	1727	75.7 (1.4)	0.45 (0.02)	0.45 (0.11-1.06)		76	12.19 (0.94)	11.26 (6.96-17.81)	5.88-71.86
>3.50	1442	77.9 (1.7)	0.48 (0.03)	0.44 (0.13-1.23)		92	12.77 (0.78)	10.16 (7.43-19.34)	5.90-91.40
Missing	485	76.2 (3.0)	0.44 (0.05)	0.42 (0.13-1.28)		22	11.56 (0.84)	10.10 (7.43-15.78)	6.23-51.27
Smokingd
Never smoker	2866	77.5 (1.2)	0.47 (0.02)	0.46 (0.13-1.21)	0.55	131	12.39 (0.65)	9.96 (7.18-18.04)	5.88-91.40	0.92
Current smoker	1122	81.2 (1.4)	0.51 (0.03)	0.51 (0.18-1.29)		56	12.57 (1.54)	11.41 (6.63-24.07)	5.91-58.70
Ever smoker	1245	76.6 (1.5)	0.47 (0.03)	0.48 (0.12-1.16)		75	13.04 (0.95)	10.79 (7.88-16.77)	5.90-73.10
Physical activity, metabolic equivalent of task (MET)-min/weeke
<600	2180	76.2 (1.3)	0.46 (0.02)	0.46 (0.12-1.23)	0.24	95	12.28 (1.07)	10.18 (7.25-20.46)	5.88-71.86	0.13
600-1200	562	77.2 (2.2)	0.46 (0.04)	0.46 (0.14-1.16)		30	15.39 (2.17)	14.54 (9.01-17.67)	6.03-91.40
> 1200	2491	79.6 (1.1)	0.50 (0.02)	0.49 (0.16-1.20)		137	12.33 (0.68)	10.18 (7.29-17.79)	5.93-83.1
Alcohol intakef
Non-drinker	3755	76.9 (1.2)	0.46 (0.02)	0.44 (0.13-1.15)	0.22	181	12.70 (0.63)	10.89 (7.16-18.81)	5.88-83.10	0.83
Moderate drinker	429	74.5 (3.1)	0.50 (0.06)	0.49 (0.07-1.32)		30	11.46 (0.97)	9.13 (7.46-12.42)	5.94-91.40
Heavy drinker	772	84.6 (1.5)	0.58 (0.03)	0.59 (0.22-1.36)		41	13.92 (1.54)	12.14 (9.08-23.46)	5.91-42.23
Missing	277	79.5 (3.0)	0.44 (0.04)	0.47 (0.13-1.10)		10	9.39 (1.04)	8.03 (6.92-10.32)	5.97-22.53
BMI, kg/m2 g
<25	1510	76.5 (1.4)	0.46 (0.03)	0.45 (0.12-1.24)	<0.0001	86	12.08 (0.91)	10.28 (7.06-19.37)	5.90-71.86	0.73
25-29.9	1773	77.7 (1.5)	0.46 (0.03)	0.44 (0.14-1.14)		80	12.67 (1.10)	9.97 (7.25-18.64)	5.88-91.40
≥30	396	79.8 (1.1)	0.52 (0.02)	0.53 (0.18-1.25)		96	13.15 (1.05)	11.40 (7.78-17.78)	6.02-83.10

All the variables were adjusted using population weights for the sample in which 3-PBA concentration was measured except N (unweighted sample size). Analysis of variance (ANOVA) was used to compare differences of urinary levels of 3-PBA among various categorical variables. The lower limit of detection of the urinary 3-PBA analysis was 0.1 μg/L. See Table S1 for a summary of analogous creatinine-adjusted data.

^aRace/ethnicity was categorized based on self-reported data into Hispanic (including Mexican and non-Mexican Hispanic), non-Hispanic white, non-Hispanic black, and other race/ethnicity.

^bSelf-reported education was grouped as less than high school, high school, and college or higher.

^cPIRs were categorized as ≤ 1.30, 1.30-3.50, and > 3.50.

^dSelf-reported smoking was classified as never smokers (smoked < 100 cigarettes in their lifetime), current smokers (had smoked >100 cigarettes and smoked cigarettes at the time of the survey), and ever smokers (had smoked > 100 cigarettes but did not smoke at the time of the survey).

^eSelf-reported physical activity was used to derive the metabolic equivalent of task (MET) minutes per week according to the Global Physical Activity Questionnaire Analysis Guide and categorized as <600, 600-1,200, and >1,200 MET-min/week.

^fAlcohol intake was categorized as non-drinker (0 g/d for both male and female), moderate drinker (0.1-27.9 g/d for male, 0.1-13.9 g/d for female), and heavy drinker (≥ 28 g/d for male, ≥ 14 g/d for female).

^gBMI was calculated as weight in kilograms divided by the square of height in meters and classified as normal weight (BMI <25), overweight (25-29.9), and obese (≥30). Weight and height were determined by trained health technicians according to the NHANES Anthropometry Procedures Manual.

3-PBA, 3-phenoxybenzoic acid; BMI, body mass index; SE, standard error.

Creatinine adjusted geometric mean and median levels of 3-PBA in U.S. adults were 0.51 μg/g creatinine (S.E., 0.2 μg/L) and 0.48 μg/g creatinine (IQR, 0.21-1.11 μg/g creatinine) (Table S1). The creatinine adjusted geometric means and median levels of 3-PBA in individuals in the 95^th percentile (N = 262) were over 20-fold higher compared to the overall study population, with a geometric mean of 10.85 μg/g creatinine (SE, 0.41 μg/L) and a median of 9.41 μg/g creatinine (IQR, 6.71-13.83 μg/g creatinine).

Urinary 3-PBA levels for the entire U.S. population ≥ 6 years of age have been reported previously for the NHANES 1999-2002 cycles (Barr et al., 2010). Weighted detection rates in NHANES 1999-2000 and 2001-2002 for individuals ≥ 6 years of age were 66.5% and 75.4%, respectively. Geometric means and median levels of 3-PBA, both adjusted for volume and creatinine, were lower compared to the NHANES 2007-2012 cycles. For example, geometric means in the NHANES 1999-2000 and 2001-2002 populations were 0.292 μg/L and 0.318 μg/L, respectively. Median levels of 3-PBA were 0.25 μg/L and 0.27 μg/L in NHANES 1999-2000 and 2001-20002, respectively. This comparison indicates that the exposure of the general U.S. population to pyrethroid pesticides has increased with time. A few smaller human biomonitoring studies from North Carolina and California (N = 50 to 105) similarly report 3-PBA levels in samples collected from 2007-2011 that are higher than the levels observed in earlier NHANES cycles, with median levels ranging from 0.82 to 1.63 μg/L (0.61 to 1.46 μg/g creatinine) (Morgan et al., 2016; Trunnelle et al., 2014a; Trunnelle et al., 2014b). However, the higher exposures in these studies may, in part, be a result of residential pesticide use, occupational exposures, or because the studies were performed in an agricultural community.

3.2. Urinary levels of 3-PBA in other relevant adult study populations

Several other studies reported 3-PBA levels in urine samples collected from other study populations that overlapped with the 2007-2012 timeframe, based on a scoping review of the literature (Table S7). The Korean National Environmental Health Survey (KNEHS) 2009-2011 reported geometric mean 3-PBA levels of 1.47 μg/L (95% CI, 1.39–1.57 μg/L; N = 6232) (Park et al., 2016) and 1.83 μg/g creatinine (geometric standard deviation 2.71 μg/g creatinine, N = 3,671) in South Korean adults 19 years or older (Yoo et al., 2016). These numbers are higher compared to the NHANES data; however, exposures in the 95^th percentile were higher in the U.S. vs. the South Korean population (e.g., median 3-PBA levels of 12.60 μg/L vs. 8.86 μg/L, respectively) (Park et al., 2016).

A human biomonitoring study reported higher urinary levels of 3-PBA in urine samples collected in 2009 from women (~37 years old) in South Africa (median of 3.4 μg/g creatinine, IQR 2.21-6.00 μg/g creatinine, N = 182) (Motsoeneng and Dalvie, 2015; Mwanga et al., 2016). Compared the NHANES 2007-2012, studies from Poland reported lower urinary 3-PBA levels in men collected from 2008 to 2011, with median levels of 0.16 μg/L (0.13 μg/g creatinine, N = 334) (Jurewicz et al., 2016; Jurewicz et al., 2015; Radwan et al., 2015; Radwan et al., 2014) or 0.255 μg/L (0.19 μg/g creatinine, N = 132) (Wielgomas et al., 2013). These limited data demonstrate that contemporary (i.e., 2007 to 2012) exposures to pyrethroid pesticides are near-ubiquitous and display variability across the world.

3.3. Factors affecting urinary levels of 3-PBA in U.S. adults across all three NHANES cycles

Urinary 3-PBA levels, expressed as μg/L, were not significantly different between age groups (P = 0.48), male and female U.S. adults (P = 0.27), or race/ethnicity (P = 0.34) based on the ANOVA (Table 2). However, according to the linear regression analysis, urinary 3-PBA levels in the 40-59 (β coefficient 0.23, P < 0.0001) and ≥ 60 year age group (β coefficient 0.22, P = 0.01) were significantly different compared to the 20-39 year age group (Table 3). Urinary 3-PBA levels were significantly different between men and women (β coefficient 0.26, P < 0.0001). No sex difference in 3-PBA levels was reported in an analysis of NHANES 1999-2002 data (Barr et al., 2010). The linear regression analysis revealed no significant differences in 3-PBA concentrations by race/ethnicity. In contrast, the analysis of NHANES 2001-2002 data, but not NHANES 1999-2000 data, revealed significant differences by race/ethnicity. In the NHANES 2001-2002, non-Hispanic blacks had higher least-squares geometric mean concentrations in their urine compared to non-Hispanic whites and Mexican Americans (Barr et al., 2010).

Table 3.

Association of demographic and lifestyle factors in adults from NHANES 2007-2012 with urinary 3-PBA concentrations (μg/L).

Variable	β coefficient	P
Age
20-39	[reference group]
40-59	0.23 (0.05)	<0.0001
≥60	0.22 (0.08)	0.01
Gender
male	[reference group]
female	0.26 (0.05)	<0.0001
Race/ethnicitya
Non-Hispanic white	[reference group]
Hispanic	−0.11 (0.07)	0.11
Non-Hispanic black	−0.11 (0.08)	0.17
Other	0.33 (0.10)	0.001
Educationb
Less than high school	[reference group]
High school	−0.01 (0.07)	0.86
College or higher	−0.04 (0.07)	0.61
Family income to poverty ratio (PIR)c
≤ 1.30	[reference group]
1.31-3.50	−0.23 (0.05)	<0.0001
>3.50	−0.14 (0.10)	0.14
Missing	−0.20 (0.12)	0.11
Smokingd
Never smoker	[reference group]
Current smoker	0.02 (0.06)	0.83
Ever smoker	−0.02 (0.06)	0.67
Physical activity, MET-min/weeke
<600	[reference group]
600-1200	0.003 (0.08)	0.97
>1200	0.09 (0.05)	0.046
Alcohol intakef
Non-drinker	[reference group]
Moderate drinker	0.13 (0.10)	0.19
Heavy drinker	0.26 (0.07)	0.0003
Missing	−0.07 (0.10)	0.51
BMI, kg/m2 g
<25	[reference group]
25-29.9	−0.05 (0.07)	0.44
≥30	−0.06 (0.07)	0.50

Urinary creatinine was adjusted in the model. See Table S2 for an analogous analysis of creatinine-adjusted data.

^aRace/ethnicity was categorized based on self-reported data into Hispanic (including Mexican and non-Mexican Hispanic), non-Hispanic white, non-Hispanic black, and other race/ethnicity.

^bSelf-reported education was grouped as less than high school, high school, and college or higher.

^cPIRs were categorized as ≤ 1.30, 1.30-3.50, and > 3.50.

^dSelf-reported smoking was classified as never smokers (smoked < 100 cigarettes in their lifetime), current smokers (had smoked > 100 cigarettes and smoked cigarettes at the time of the survey), and ever smokers (had smoked > 100 cigarettes but did not smoke at the time of the survey).

^eSelf-reported physical activity was used to derive the metabolic equivalent of task (MET) minutes per week according to the Global Physical Activity Questionnaire Analysis Guide and categorized as <600, 600-1,200, and >1,200 MET-min/week.

^fAlcohol intake was categorized as non-drinker (0 g/d for both male and female), moderate drinker (0.1-27.9 g/d for male, 0.1-13.9 g/d for female), and heavy drinker (≥ 28 g/d for male, ≥ 14 g/d for female).

^gBMI was calculated as weight in kilograms divided by the square of height in meters and classified as normal weight (BMI < 25), overweight (25-29.9), and obese (≥ 30). Weight and height were determined by trained health technicians according to the NHANES Anthropometry Procedures Manual.

3-PBA, 3-phenoxybenzoic acid; BMI, body mass index.

In the NHANES 2007-2012 population, significant differences in the urinary 3-PBA levels were also observed depending on the PIR (P = 0.03) and the BMI (P < 0.0001), but not for other physiological or lifestyle factors investigated, including physical activity and alcohol consumption (Table 2). According to the linear regression analysis (Table 3), urinary 3-PBA levels in the group with an intermediate but not a high PIR were significantly different compared to the group in the lowest PIR (β coefficient −0.23, P < 0.0001). These differences suggest that socioeconomic factors, housing status, and diet may affect pyrethroid exposures. An analysis of NHANES 2001-2002 data speculated that similar factors contributed to significant differences in 3-PBA levels by race/ethnicity (Barr et al., 2010).

No statistically significant differences were observed for BMI in the linear regression analysis. 3-PBA levels in U.S. adults with high physical activity (MET > 1,200) were significantly different from individuals with low physical activity (MET < 600; β coefficient 0.09, P = 0.046). Moreover, urinary 3-PBA levels in heavy drinkers were significantly different from those in non-drinkers (β coefficient 0.26, P = 0.0003).

The KNEHS 2009-2011 also identified several factors that were associated with differences in 3-PBA levels in urine, for example, age, alcohol intake, and BMI (Yoo et al., 2016). For example, higher alcohol intake was associated with higher urinary 3-PBA levels. The analysis of the KNEHS 2009-2011 data revealed a correlation between 3-PBA and BMI that was positive at lower but negative at higher 3-PBA concentrations. Importantly, the BMI range was lower in the KNEHS 2009-2011 than the NHANES 2007-2012 populations, with the highest BMI category in KNEHS 2009-2011 corresponding to the lowest BMI category in NHANES 2007-2012. A direct comparison of 3-PBA trends in relation to BMI across both studies is, therefore, not possible. However, these findings, while not necessarily consistent across these two large study populations, raise the question if other factors alter the elimination of pyrethroids by altering the expression of xenobiotic processing genes (XPGs) in the liver. Indeed, alcohol intake, BMI, and other socioeconomic and lifestyle factors can alter the toxicokinetic of xenobiotics, such as pesticides (Tardif et al., 2002).

3.4. Urinary levels of 3-PBA in U.S. children

3-PBA was detected with an overall detection rate of 79.3% in U.S. children aged 6-19 years (N = 2,295; Table 4). Geometric mean and median levels in urine samples were 0.50 μg/L (SE, 0.03 μg/L) and 0.49 μg/L (IQR, 0.17-1.29 μg/L), respectively. Analogous to U.S. adults, 3-PBA levels were over an order of magnitude higher in children in the 95^th percentile. Geometric mean and median levels in U.S. children in the 95^th percentile were 12.97 μg/L (SE, 0.86 μg/L) and 11.87 μg/L (IQR, 7.68-15.85 μg/L). These values are comparable to the geometric mean and median levels of 3-PBA observed in U.S. adults (Table 3).

Table 4.

Urinary concentration of 3-PBA (μg/L) in U.S. children in the overall NHANES 2007-2012 population and the 95^th percentile.

Variable	NHANES 2007-2012					95th percentile
	N	Detection rate, % (SE)	Geometric means (SE)	Median (P25-P75)	P	N	Geometric means(SE)	Median (P25-P75)	Min-Max	P
All	2295	79.3 (1.5)	0.50 (0.03)	0.49 (0.17-1.29)		115	12.97 (0.86)	11.87 (7.68-15.85)	6.10-286.53
Age
6-11 years	1152	80.5 (1.5)	0.56 (0.04)	0.49 (0.18-1.52)	0.08	65	14.15 (1.36)	11.44 (7.79-19.53)	5.99-122.00	0.08
12-19 years	1143	78.4 (2.0)	0.47 (0.03)	0.49 (0.16-1.13)		51	11.67 (0.52)	11.82 (7.61-15.07)	6.20-286.53
Gender
Male	1213	79.3 (1.4)	0.49 (0.03)	0.48 (0.17-1.17)	0.38	52	12.61 (0.83)	11.84 (7.42-15.53)	6.10-286.53	0.67
Female	1082	79.3 (2.6)	0.53 (0.05)	0.53 (0.16-1.42)		64	13.25 (1.18)	11.44 (7.78-16.48)	6.20-122.00
Race/ethnicitya
Non-Hispanic white	663	81.5 (2.2)	0.56 (0.05)	0.53 (0.19-1.54)	0.03	53	12.82 (0.82)	11.95 (7.78-15.70)	6.10-245.00	0.85
Hispanic	832	76.5 (1.8)	0.44 (0.03)	0.45 (0.13-1.08)		30	12.92 (1.43)	10.90 (7.78-17.1)	6.29-286.53
Non-Hispanic black	574	74.7 (2.7)	0.44 (0.04)	0.48 (0.07-1.11)		22	12.73 (1.31)	9.99 (7.41-20.15)	6.20-47.10
Other	226	78.5 (4.7)	0.45 (0.07)	0.45 (0.14-0.98)		11	14.59 (6.04)	9.08 (6.91-10.56)	6.11-122.00
Family income to poverty ratio (PIR)b
≤ 1.30	952	81.2 (1.7)	0.59 (0.05)	0.62 (0.21-1.58)	0.33	50	12.64 (0.90)	12.52 (7.35-16.93)	6.10-245.00	0.61
1.31-3.50	724	80.5 (2.1)	0.50 (0.04)	0.46 (0.19-1.21)		33	16.06 (2.4)	11.34 (7.83-29.27)	6.20-286.53
> 3.50	437	76.4 (3.6)	0.46 (0.05)	0.44 (0.12-1.07)		26	10.71 (0.61)	9.66 (7.50-14.99)	6.29-49.40
Missing	182	76.7 (3.3)	0.41 (0.05)	0.42 (0.12-1.00)		7	15.82 (2.64)	14.74 (9.43-16.23)	7.88-80.98
BMI categoryc
Normal weight	1365	79.6 (1.7)	0.49 (0.03)	0.48 (0.17-1.23)	0.73	62	13.24 (1.14)	12.06 (7.57-15.64)	6.11-286.53	0.88
Overweight	367	72.7 (3.3)	0.45 (0.05)	0.41 (0.07-1.27)		24	12.15 (1.21)	10.31 (7.99-16.70)	6.80-47.10
Obese	465	84.0 (2.3)	0.59 (0.04)	0.59 (0.26-1.32)		25	14.24 (1.69)	11.08 (6.89-23.26)	6.10-245.00
Missing	98	77.1 (5.9)	0.53 (0.12)	0.50 (0.13-1.59)		5	10.17 (1.03)	8.75 (7.28-10.62)	6.73-17.30

All the variables were adjusted using population weights for the sample in which 3-PBA concentration was measured except N (unweighted sample size). Analysis of variance (ANOVA) was used to compare differences of urinary concentrations of 3-PBA among various categorical variables. The lower limit of detection of the urinary 3-PBA analysis was 0.1 μg/L. See Table S6 for a summary of analogous creatinine-adjusted data.

^aRace/ethnicity was categorized based on self-reported data into Hispanic (including Mexican and non-Mexican Hispanic), non-Hispanic white, non-Hispanic black, and other race/ethnicity.

^bPIRs were categorized as ≤ 1.30, 1.30-3.50, and > 3.50.

^cBMI was calculated as weight in kilograms divided by the square of height in meters and classified as normal weight, overweight, and obese based on the CDC’s sex-specific 2000 BMI-for-age growth charts for the U.S. Weight and height were determined by trained health technicians according to the NHANES Anthropometry Procedures Manual.

3-PBA, 3-phenoxybenzoic acid; CDC, Center for Disease Control and Prevention; BMI, body mass index; SE, standard error.

Creatinine adjusted geometric mean and median levels of 3-PBA in U.S. children were 0.53 μg/g creatinine (SE, 0.03 μg/L) and 0.47 μg/g creatinine (IQR, 0.22-1.23 μg/g creatinine), respectively (Table S4). The creatinine adjusted geometric means and median levels in children in the 95^th percentile were 12.80 μg/g creatinine (SE, 0.91 μg/L) and 10.81 μg/g creatinine (IQR, 7.48-15.92 μg/g creatinine). These levels are over an order of magnitude higher than the geometric mean and median levels observed in the entire population.

3.5. Factors affecting urinary levels of 3-PBA in U.S. children

Factors affecting pyrethroid exposures in U.S. children have received limited attention to date. Levels of 3-PBA in 12-19-year-old children were not significantly different from 6-11-year-old children (Table 4). However, urinary 3-PBA levels were significantly different between both age groups in the linear regression analysis, both for unadjusted (β coefficient −0.50, P < 0.0001; Table 4) and creatinine-adjusted levels (β coefficient −0.61, P < 0.0001; Table S5). Significant differences were observed based on race/ethnicity for unadjusted (P = 0.03; Table 6) and creatinine-adjusted (P = 0.004; Table S4) urinary 3-PBA levels, but not for other physiological or lifestyle factors investigated.

Table 6.

Urinary concentration of F-PBA (μg/L) in the NHANES 2007-2012 population and DBCA (μg/L) in the NHANES 2007-2010 population ^a

Total population	F-PBA (μg/L)			DBCA (μg/L)
	Total	Adults	Children	Total	Adults	Children
Results including samples <LODb
N	7516	5225	2291	5326	3768	1558
Detection rate (%)	8.61	9.26	7.11	1.27	1.08	2.0
Median (P25-P75)	0.07 (0.07-0.07)	0.07 (0.07-0.07)	0.07 (0.07-0.07)	0.35 (0.35-0.35)	0.35 (0.35-0.35)	0.35 (0.35-0.35)
Geometric mean (SE)	0.08 (0.001)	0.08 (0.001)	0.08 (0.001)	0.36 (0.001)	0.35 (0.001)	0.36 (0.003)
Results excluding samples <LODb
N	659	492	167	77	44	33
Median (P25-P75)	0.22 (0.15-0.45)	0.23 (0.15-0.46)	0.20 (0.14-0.37)	1.22 (0.78-1.77)	1.09 (0.74-1.80)	1.37 (0.87-1.61)
Geometric mean (SE)	0.30 (0.02)	0.30 (0.02)	0.28 (0.02)	1.23 (0.06)	1.20 (0.08)	1.30 (0.06)

^aUrinary levels of DBCA were not measured in NHANES 2011-2012.

^bThe limit of detections (LODs) were 0.10 and 0.50 ng/mL for 3-PBA and DBCA, respectively.

F-PBA, fluoro-3-phenoxybenzoic acid; DBCA, cis-3-(2,2-dibromoethenyl)-2,2-dimethylcyclopropanecarboxylic acid.

Based on the linear regression analysis, urinary 3-PBA levels in non-Hispanic white children were significantly higher compared to Hispanic (β coefficient −0.33, P = 0.0008) and non-Hispanic black children (β coefficient −0.49, P < 0.0001) (Table 5). Also, significantly lower 3-PBA levels were observed for male vs. female children (β coefficient 0.26, P < 0.004). For example, geometric mean levels were 0.49 (SE, 0.03 μg/L) for male and 0.53 (SE, 0.05 μg/L) for female children (Table 4). According to the linear regression analysis, urinary 3-PBA levels also depend on the PIR. 3-PBA levels in U.S. children in the group with lowest household income (PIR < 1.30) were significantly higher compared to children in both the middle (PIR 1.30-3.50; β coefficient −0.20, P = 0.04) and high income group (PIR > 3.50; β coefficient −0.31, P = 0.03; Table 5). Similar statistically significant differences were observed for gender, race/ethnicity, and PIRs when creatinine-adjusted 3-PBA levels were analyzed (Table S5).

Table 5.

Association of demographic factors in children from NHANES 2007-2012 with urinary 3-PBA concentrations (μg/L).

Variable	β coefficient	P
Age
6-11	[reference group]
12-19	−0.50 (0.10)	<0.0001
Gender
Male	[reference group]
Female	0.26 (0.09)	0.004
Race/ethnicitya
Non-Hispanic white	[reference group]
Hispanic	−0.33 (0.09)	0.0008
Non-Hispanic black	−0.49 (0.12)	<0.0001
Other	−0.25 (0.17)	0.16
Family income to poverty ratio (PIR)b
≤ 1.30	[reference group]
1.31-3.50	−0.20 (0.10)	0.04
>3.50	−0.31 (0.14)	0.03
Missing	−0.33 (0.14)	0.02
BMI categoryc
Normal weight	[reference group]
Overweight	−0.13 (0.12)	0.28
Obese	0.05 (0.08)	0.56
Missing	−0.08 (0.24)	0.74

Urinary creatinine was adjusted in the model. See Table S5 for an analogous analysis of creatinine-adjusted data.

^aRace/ethnicity was categorized based on self-reported data into Hispanic (including Mexican and non-Mexican Hispanic), non-Hispanic white, non-Hispanic black, and other race/ethnicity.

^bPIRs were categorized as ≤ 1.30, 1.30-3.50, and > 3.50.

^cBMI was calculated as weight in kilograms divided by the square of height in meters and classified as normal weight, overweight, and obese based on the CDC’s sex-specific 2000 BMI-for-age growth charts for the U.S. Weight and height were determined by trained health technicians according to the NHANES Anthropometry Procedures Manual.

3-PBA, 3-phenoxybenzoic acid; CDC, Center for Disease Control and Prevention; BMI, body mass index; SE, standard error.

3.6. Urinary levels of 3-PBA in children from other relevant study populations

Several small studies (N ≤ 180) reported 3-PBA levels in urine samples collected from children that overlapped with the 2007-2012 timeframe, based on a scoping review of the literature (Table S7). Two small studies from California, USA reported median 3-PBA levels of 0.75 μg/L (0.80 μg/g creatinine; N = 83) (Trunnelle et al., 2014b) and 1.93 μg/L (2.56 μg/g creatinine; N = 103) (Trunnelle et al., 2014a) in approximately 2-8 year old children. These studies were performed in or near agricultural communities, or in households with possible residential pesticide use, which likely explains why 3-PBA levels were higher compared to children in the NHANES 2007-2012 cycles. The levels of 3-PBA in the second study population, which included Hispanic children living in an agricultural community in Mendota, California, were 4-5-fold higher compared to median 3-PBA levels in NHANES children (Trunnelle et al., 2014a). In comparison, Hispanic children had significantly lower 3-PBA levels than non-Hispanic white children in NHANES 2007-2012 (Table 4). The higher levels of 3-PBA levels in the smaller cohort from California are indicative of higher pesticide exposure in immigrant farmworker families living close to agricultural fields under poor housing conditions (Trunnelle et al., 2014a).

A study from Spain reported urinary 3-PBA levels below the limit of quantification (L.O. Q) in children aged 6-11 years old (N = 125) (Roca et al., 2014). It is unclear if pyrethroid exposures in Spain are lower compared to other countries or if the low detection frequency is due to a lack of sensitivity of the analytical method (limit of quantification = 0.8 μg/g creatinine). Studies from Central America reported higher 3-PBA levels compared to NHANES 2007-2012 (Rodriguez et al., 2012; van Wendel de Joode et al., 2016). Children 6-9 years old, living near banana plantations and plantain farms in Costa Rica, had geometric mean levels of 0.8 μg/L (N = 140) (van Wendel de Joode et al., 2016). Even higher geometric mean levels of 1.5 μg/L (3.2 μg/g creatinine; N = 77) were reported in children aged 7-9 years from Nicaragua (Rodriguez et al., 2012).

A study from China observed 3-PBA levels in healthy < 14-year-old children that were comparable to the NHANES data (median 0.47 μg/L or 0.68 μg/g creatinine; N = 180) (Ding et al., 2012). The same study reported a geometric mean of 0.54 μg/L (0.81 μg/g creatinine) and median of 0.65 μg/L (0.93 μg/g creatinine) for 3-PBA levels in children (N = 176) newly diagnosed with childhood acute lymphocytic leukemia (Table 4) (Ding et al., 2012).

3.7. Temporal trends in urinary 3-PBA levels in NHANES 2007-2008 to 2011-2012

3-PBA levels increased significantly with time in both U.S. adults and children (Figures 2A and 2B). For example, geometric mean 3-PBA levels in U.S. adults were 0.41 μg/L (SE, 0.03 μg/L) in NHANES 2007-2008, 0.41 μg/L (SE, 0.01 μg/L) in NHANES 2009-2010, and 0.66 μg/L (SE, 0.04 μg/L) in NHANES 2011-2012. In U.S. children, geometric mean 3-PBA levels were 0.40 μg/L (SE, 0.04 μg/L) in NHANES 2007-2008, 0.46 μg/L (SE, 0.04 μg/L) in NHANES 2009-2010, and 0.70 μg/L (SE, 0.05 μg/L) in NHANES 2011-2012. Similarly, 3-PBA levels were significantly different across NHANES cycles when creatinine-adjusted 3-PBA levels were compared (P <0.0001 for both U.S. adults and children). A recent study investigating temporal trends in pesticide biomarkers in the urine of Swedish adolescents also reported an increase in 3-PBA levels from 2000-2017 (Norén et al., 2020).

Figure 2.

Geometric mean levels of 3-PBA differ significantly across the three NHANES cycles investigated for (A) adults and (B) children in the NHANES 2011-2012 population. This difference was statistically significant for 3-PBA levels expressed as μg/L (P <0.0001 for both U.S. adults and children; Tables S8 and S9) and μg/g creatinine (P <0.0001 for both U.S. adults and children; Tables S3 and S6). (C) In contrast, the estimated use of relevant pyrethroid pesticides did not show a clear increase in the period from 2007 to 2012. The low and high use estimates are the sum of use data of γ- and λ-cyhalothrin, cypermethrin, deltamethrin, esfenvalerate, fenpropathrin, τ-fluvalinate, permethrin, and tralomethrin calculated on a molar basis (Σpyrethroids). These pyrethroids are metabolized to 3-PBA, and their use data are available from the USGS (USGS). (D) Estimated agricultural uses of cyfluthrin and deltamethrin are lower compared to the total estimated uses of pyrethroids that are metabolized to 3-PBA. These estimates are consistent with the low detection rate of F-PBA, a specific metabolite of cyfluthrin, and DBCA, a specific metabolite of deltamethrin (see text for details).

It is unclear why the 3-PBA levels in NHANES 2011-2012 cycle are higher compared to the earlier NHANES cycles. Changes in the residential, agricultural, and other commercial uses of pyrethroids over time are a likely explanation for the observed temporal trends that warrants further attention. However, the estimated agricultural use of relevant pyrethroid pesticides, based on USGS data (USGS, n.d.), did not show a clear increase in the period from 2007 to 2012 (Figure 2C). In contrast, residential uses of pyrethroids have likely increased over time, in part due to the phaseout of organophosphate pesticides for indoor uses (Power and Sudakin, 2007; Williams et al., 2008), which has resulted in a decrease in human exposures to these pesticides over time (Gillezeau et al., 2019; Jain, 2017). Moreover, pyrethroid use has increased globally for vector control (van den Berg et al., 2012). Consistent with this interpretation of our findings, a study from Japan reported an increase in the exposure of children to some pyrethroid insecticides used from 2006 to 2015 to prevent the spread of diseases in public spaces (Hamada et al., 2020).

3.8. Contemporary urinary levels of F-PBA and DBCA in U.S. adults and children

Urinary concentration of F-PBA in the 2007-2012 NHANES population and DBCA in the 2007-2010 NHANES populations are summarized in Table 6 to provide an overview of the current levels of these metabolites in U.S. adults and children. Other pyrethroid metabolites were not reported in the NHANES 2007-2012 cycles. Detection rates of F-PBA, a specific metabolite of cyfluthrin (Angerer and Ritter, 1997; Barr et al., 2010), were much lower compared to 3-PBA, with a detection rate of 8.61 % in the NHANES 2007-2012 population (N = 7,516; Table 6). The detection rate in U.S. adults and children was 9.26 % (N = 5,225) and 7.11 % (N = 2,291), respectively. For comparison, the detection frequencies of F-PBA in the NHANES 1999-2000 and NHANES 2001-2002 populations were only 3.2% and 0.6%, respectively (Barr et al., 2010). Geometric mean and median levels of F-PBA in the NHANES 2007-2012 population, calculated after excluding individuals with urinary concentrations below the detection limit (N = 659), were 0.30 μg/L (SE, 0.02 μg/L) and 0.22 μg/L (IQR, 0.15-0.45 μg/L), respectively.

These observations are consistent with estimated use numbers from the USGS (USGS, n.d.). On a molar basis, the low and high estimates of cyfluthrin use in the U.S. in 2007-2012 are 10-16% of the total use estimates of other pyrethroids that are metabolized to 3-PBA, including γ- and λ-cyhalothrin, cypermethrin, deltamethrin, esfenvalerate, fenpropathrin, τ-fluvalinate, permethrin, and tralomethrin (Figure 2). While the low use estimates of cyfluthrin remained constant from 1999 to 2012 at 147 Kmol/year, the high use estimates increased from 196 Kmol/year in 1999-2002 to 288 Kmol/year in 2007-2012. This increase could explain the higher detection frequency observed in the 2007-2012 compared to the 1999-2002 NHANES cycles.

An even lower detection rate of 1.27 % (1.08 % and 2.0% in U.S. adults and children from NHANES 2007-2010, respectively) was observed for DBCA (Table 6), a selective metabolite of deltamethrin (Angerer and Ritter, 1997; Barr et al., 2010). Similar detection limits of 1.3% and 0.5% of DBCA were reported for the NHANES 1999-2000 and NHANES 2001-2002 cycles, respectively (Barr et al., 2010). After excluding individuals with urinary concentrations below the detection limit (N = 77), geometric mean and median levels of this metabolite were 1.23 μg/L (SE, 0.06 μg/L) and 1.22 μg/L (IQR, 0.78-1.77 μg/L) in the NHANES 2007-2010 population (Table 6). The low detection frequency of DBCA in the general U.S. population is consistent with the estimated use data of deltamethrin, with is approximately 100-times lower compared to the total use estimates for pyrethroids that are metabolized to 3-PBA (USGS, n.d.).

4. Conclusions

This analysis of data from the National Health and Nutrition Examination Survey 2007-2012 demonstrates that the exposure of the general U.S. population to pyrethroid pesticides is near-ubiquitous and varies in population subgroups categorized by demographic, socioeconomic and lifestyle factors. Exposures of adults and children to pyrethroids are also prevalent across the globe based on our scoping review of human biomonitoring studies performed in the same timeframe. The NHANES data also show an increase in pyrethroid exposures in the general U.S. population over time. The number of human biomonitoring studies sampling between 2007 to 2012 is small, which makes it challenging to assess global trends in pyrethroid exposures. Global pyrethroid use and, as a consequence, human exposures have likely increased over the last two decades due to the phaseout of the more toxic organophosphate pesticides. However, a systematic review of human biomonitoring studies is needed to characterize trends in human exposures over time. Overall, there is a need to study the health outcomes and the underlying mechanisms associated with environmental exposures to pyrethroid pesticides across the lifetime.

Highlights.

• Pyrethroid exposure was assessed in the U.S. population using urinary biomarkers

• 3-PBA is near-ubiquitous in urine samples from U.S. adults and children

• Exposure to pyrethroids, measured with urinary 3-PBA levels, increased over time

• 3-PBA levels are associated with demographic, socioeconomic and lifestyle factors

• Detection rates were low for urinary biomarkers of cyfluthrin and deltamethrin