# Exposures of preschool children to chlorpyrifos, diazinon, pentachlorophenol, and 2,4-dichlorophenoxyacetic acid over 3 years from 2003 to 2005: A longitudinal model

## Abstract

The impact of the US EPA-required phase-outs starting in 2000–2001 of residential uses of the organophosphate (OP) pesticides chlorpyrifos (CPF) and diazinon (DZN) on preschool children's pesticide exposures was investigated over 2003–2005, in the Raleigh-Durham-Chapel Hill area of North Carolina. Data were collected from 50 homes, each with a child initially of age 3 years (OCh) and a younger child (YCh). Environmental samples (indoor and outdoor air, dust, soil) and child-specific samples (hand surface residue, urine, diet) were collected annually over 24-h periods at each home. Child time-activity diaries and household pesticide use information were also collected. Analytes included CPF and DZN; pentachlorophenol (PCP); 2,4-dichlorophenoxyacetic acid (2,4-D); the CPF metabolite 3,5,6-trichloro-2-pyridinol (TCP); and the DZN metabolite 2-isopropyl-6-methyl-4-pyrimidinol (IMP). Exposures (ng/day) through the inhalation, dietary ingestion, and indirect ingestion were calculated. Aggregate potential doses in ng/kg body weight per day (ng/kg/day) were obtained by summing the potential doses through the three routes of exposure. Geometric mean aggregate potential doses decreased from 2003 to 2005 for both OCh and YCh, with the exception of 2,4-D. Child-specific longitudinal modeling indicated significant declines across time of the potential doses of CPF, DZN, and PCP for both children; declines of IMP for both children, significant only for OCh; a decline of TCP for OCh but an increase of TCP for YCh; and no significant change of 2,4-D for either child. Age-adjusted modeling indicated significant effects of the child's age for all except CPF, and of time for all except PCP and 2,4-D. Within-home variability was small compared with that between homes; variability was smallest for 2,4-D, both within and between homes. The aggregate potential doses of CPF and DZN were well below published reference dose values. These findings show the success of the US EPA restrictions in reducing young children's pesticide exposures.

## Introduction

Children can be exposed to environmental pollutants from multiple sources and through multiple routes (inhalation, dietary ingestion, non-dietary/indirect ingestion, and dermal absorption). Very young children learn about their environment by exploring not only the appearance and texture of objects, but also their taste and smell. Thus, their exposures to environmental pollutants may be greater than those of adults or older children. Because of children's smaller body masses, immature body systems, and rapid physical development, the impact of these exposures may contribute to adverse health effects to a greater extent than on adults (Perera, 1977; Schettler, 2001; Mendola et al., 2002).

Pesticide exposures and exposures to other xenobiotics have been implicated as compromising children's health (Goldman, 1995; Landrigan et al., 1999; Perera et al., 2005). To understand these exposures better, many studies have estimated the exposures of children and others to specific pesticides through measurement of environmental levels and urinary biomarkers of exposure (Bradman et al., 1997; Loewenherz et al., 1997; Chuang et al., 1998; Adgate et al., 2001; Wilson et al., 2001, 2003, 2004, 2007; Lu et al., 2001, 2004; Fenske et al., 2002, 2005; Koch et al., 2002; Curl et al., 2003; Wessels et al., 2003; Heudorf et al., 2004; Kissel et al., 2005; Lambert et al., 2005; Morgan et al., 2005, 2008; Arcury et al., 2006, 2007; Valcke et al., 2006; Alexander et al., 2007; Curwin et al., 2007).

Concerns about the possible adverse effects of exposures to the organophosphate (OP) pesticides chlorpyrifos (CPF) and diazinon (DZN) on children, and the requirements of the US Food Quality Protection Act of 1996 (FQPA, 1996), led the US Environmental Protection Agency (USEPA) to require the phase-out of the sale of these pesticides for use in and about residences, schools, or other locations where children are found. Lower tolerances were also required on crops regularly eaten by children: CPF starting in 2000 and DZN in 2001 (Federal Register, 2000, 2001; USEPA, 2000a, 2000b). These actions were undertaken to decrease the exposures of US residents, especially young children, to these pesticides as their introduction into the environment was phased out.

The main objective of this research was to estimate the longitudinal changes in aggregate exposures to CPF and DZN for selected preschool children in the same age group over 3 years. A second objective was to estimate the interpersonal variability of these exposures between preschool children of the same age, but living in different homes, and between preschool children of different ages living within the same homes.

## Methods

The design strategy, survey sampling, recruiting and retention, field sampling, and analytical methods for this study are summarized briefly below.

### Recruiting and Participants

The Pesticide Exposures of Preschool Children Over Time (PEPCOT) study was conducted in the Raleigh-Durham-Chapel Hill, North Carolina (NC) area, to compare the aggregate potential doses of young children before and after the US EPA restrictions on sales of CPF and DZN. Information on preschool children's pesticide exposure levels before and in the initial year of the CPF use reduction was available from our previous work (Wilson et al., 2001, 2003, 2004; USEPA, 2004). To deal with the difficult interpretation of the exposures of a given child over time as she or he progresses through different developmental stages, the PEPCOT study included siblings in the same households. Households with two or more children, one being 3 years of age and a second child being younger than 3 years at enrollment, were recruited. Age-eligible twins and a third age-eligible child were enrolled when possible.

Selecting households with a child in the 3-year age group and a younger sibling facilitated comparison of the exposures of children in the same household at the same developmental stage/age, but in different years. Changes in the children's exposures over the 3 years could be measured, and the interpersonal variabilities of these exposures could be estimated, not only within households for children at different ages, but also between households for children at the same ages.

Families with a head of household between ages 18 and 35 years were contacted by telephone. Of 3011 cases in the telephone screening lists, 1860 were ineligible, 409 refused screening, 664 were non-working numbers, and 78 were eligible and agreed to be visited by the project staff. Informed consent was obtained from 51 potential participating households, one of whom withdrew.

### Field Sampling

Fifty selected families were followed over 3 years. The families were randomly assigned into three seasonal subsamples. Each family was sampled annually in the same season—Spring, Summer, or Fall. To minimize the effect of seasonality in pesticide levels on the within-household (or child) trends, the second and third year sampling was scheduled ±2 weeks of the corresponding date of the first year sampling. The numbers of households that were sampled in each season in years 1, 2, and 3, respectively, are as follows: Spring, 17, 16, 16; Summer, 14, 12, 12; and Fall, 19, 18, 18.

During the annual 24-h household sampling visit, indoor and outdoor air, indoor floor (carpet) dust, and outdoor soil samples were collected. In households with little or no carpet, floor surface wipe samples were collected. In those households that had pesticide applications during the monitoring period or in the previous 7 days, floor and food preparation surface wipe samples were collected. Child-specific personal samples included duplicates of the children's liquid and solid food eaten in the 24-h period, hand surface wipes, and first morning void urine. Drinking water samples were not collected, as previous work (USEPA, 2004; Wilson et al., 2004) showed that the target pesticides were unlikely to be detectable in drinking water samples. If the children were not yet toilet-trained, overnight diapers were collected. If the children were being breast-fed, samples of breast milk were requested. Ancillary questionnaire and survey data included child food/time-activity diaries, household characteristics, and other interpretive information.

The indoor and outdoor air samplers, operating at 4 l/min for 24 h, were set up and maintained by study staff. Each sampler had a 10 μm impactor-equipped inlet, followed by a glass cartridge containing a quartz fiber filter followed in series by XAD-2 sorbent resin, to catch both vapor and particulate phases of the targeted compounds. Staff collected floor dust samples with an HVS3 vacuum sampler using an ASTM standard procedure (ASTM, 1997) as well as soil samples from the top 1 cm of the outdoor play area, over an area of 0.1 m2, and food preparation surface and hard floor surface wipe samples. Staff also traced the children's hands on pre-calibrated sheets of paper for hand surface area calculations. Each child's four daily hand wipe samples, from the entire surface area of all fingers and the front and back of both hands, were obtained before meals (breakfast, lunch, or dinner), before the child's hands were washed. All wipe samples were collected with cotton gauze, moistened with 50% isopropanol/water.

Refrigerated samples at the end of the 24-h collection period were packed in dry ice and shipped to Battelle's Columbus Laboratories (BCL) for analysis. Upon receipt of multimedia samples, the solid food samples were weighed and homogenized. The volumes of the liquid food samples were measured, the gels from the diaper samples were removed for storage, and the dust samples were sieved to collect the fine (<150 μm) fractions. Collected samples were frozen and stored at <−10°C, pending chemical analysis.

### Chemical Analysis

The target analytes for PEPCOT were divided for analysis into two groups, neutral and acidic, on the basis of their chemical properties. The neutral group included CPF, DZN, and pyrethroids; the acidic group included 2,4-dichlorophenoxyacetic acid (2,4-D), pentachlorophenol (PCP), the CPF and DZN metabolites 3,5,6-trichloro-2-pyridinol (TCP) and 2-isopropyl-6-methyl-4-pyrimidinol (IMP), respectively, and the pyrethroid metabolite 3-phenoxybenzoic acid (PBA).

The multimedia samples were solvent-extracted using Soxhlet, sonication, or accelerated solvent extraction techniques. Dust sample extracts were processed through Florisil SPE columns, and food sample extracts were fractionated by gel permeation column followed by ENVI-Carb SPE columns. Acidic compounds were derivatized by silylation or methylation. Concentrated extracts of all samples were analyzed by gas chromatography/mass spectrometry (GC/MS) in the selected ion-monitoring mode. Quality control samples analyzed included the following: field and/or laboratory duplicates, matrix spike samples, and field and laboratory blanks. Surrogate recovery standards (IMP-13C, 2,4-D-13C, PBA-13C, cis-permethrin-13C, trans-permethrin-13C) and internal standards (DZN-d10, TCP-13C15N, and dicamba-d3) were used to assess recovery and quantification.

Neutral and acidic compounds were measured in the environmental and personal samples, excluding urine. Acidic pollutants and/or metabolites were measured in urine/diaper samples, using published procedures (Wilson et al., 2004; Nishioka et al., 2006). Creatinine was not measured in the urine samples, because of the small individual sample volumes and the uncertain value of creatinine correction for small children (O’Rourke et al., 2000). Hand areas were calculated from the hand traces on precalibrated sheets of paper using IMAGE J software (http://rsb.info.nih.gov/ij/).

### Calculation of Daily Exposure and Dose

In PEPCOT, multimedia environmental and child-specific food and hand surface residue measurements were used to calculate maximum potential external/intake dose. Urinary metabolites of the target pesticides and unchanged pesticides in the urine were measured to estimate the actual absorbed pesticide dose.

Individual laboratory measurements below the method detection limits for the sample type and medium were replaced by the detection limit divided by the square root of two.

In a few cases, planned samples were not available for analysis either because they could not be obtained in the field, they were lost during processing, or they could not be analyzed. Two methods were used to account for the missing samples. In the first method, missing samples were simply excluded from all analyses. The downside to this approach is that the power to detect differences in the data is reduced because less data are available for analysis. The second method involves imputing values for missing samples based on results obtained for all samples that were successfully collected and analyzed. Comparisons of the statistical methodology for using imputed results to those that relied only on non-missing data suggested that no significant biases would be introduced, and the power to detect trends over time would be maximized. This approach takes advantage of additional information and was therefore used in the data analyses.

The imputation procedure involved using simple regression and analysis of variance techniques to estimate the missing values based on the distribution of observed concentrations of the target chemical in the same media over time, or in some cases in a related medium. The largest number of missing data was for soil samples: measurements were obtained for all 46 households in Year 2, but for only 12 of 50 households in Year 1, and 12 of 46 households in Year 3. A simple linear regression model was developed that assessed the trend over time in a natural log-transformed soil concentrations (e.g., ln(CSoil)=β0+β1·Year+ɛ), separately for each target pesticide or metabolite, with predicted values from the model used to impute soil concentrations for years in which these data were missing. A similar technique was used to develop imputed values for missing dust concentration data, where measurements were obtained for 29 of 50, 26 of 46, and 28 of 46 households in Years 1, 2, and 3, respectively. Both soil and dust measurements were log-normally distributed.

The initial calculations of exposures and potential doses were based on EPA-published equations and exposure factors (USEPA, 1997, 2008). The exposure values (ng/day) for inhalation and ingestion (dietary and non-dietary/indirect) were converted to units of maximum potential dose by assuming 100% availability for absorption in the body and normalizing for body weight. This conversion gives upper limits on the amount of a pollutant available for intake and delivery to target organs. Dermal absorption was assumed negligible at the low levels encountered here (Griffin et al., 1999; Garfitt et al., 2002; Meuling et al., 2005; Morgan et al., 2005). Preliminary modeling indicated that the inclusion of the hand loadings in the aggregate dose estimates did not affect the estimates significantly.

Three routes of exposure were investigated: dietary ingestion, which was limited to solid food, as the liquid food samples generally did not contain measurable levels of the target compounds; non-dietary/indirect ingestion through soil and house dust; and inhalation through indoor and outdoor air. This allowed analysis of three types of measurements: concentrations in the sampled media, daily exposures, and daily doses.

To translate the measurements from the exposure pathways into potential doses, the data were converted by the series of equations detailed below. The daily aggregate potential dose measurement in ng/kg body weight per day (ng/kg/day) was calculated by summing the potential doses for the dietary ingestion, non-dietary ingestion, and inhalation routes.

The following equations were used to convert daily dietary exposure to daily potential dietary dose:

where Cs is the concentration in solid food (ng/g), and Ms is the total weight of solid food (g), sample collected over 24 h,

where R is the absorption fraction, set to 1, and W is the child's weight (kg).

The following equations were used to convert daily inhalation exposure to daily inhalation dose:

where Cind is the concentration in indoor air (ng/m3), Cout is the concentration in the outdoor air (ng/m3), tind is time spent indoors (h), tout is time spent outdoors (h), and V is the ventilation rate (m3/day; set to 8.53–12.65 m3/day, depending on the child's age and sex, from the US EPA Child-Specific Exposure Factors Handbook (USEPA, 2008)).

where R is the absorption fraction, set to 1, and W is the child's weight (kg).

The following equations were used to convert daily non-dietary (indirect) ingestion exposure to daily non-dietary dose:

where Cd is the concentration in dust indoors (ng/g), Cs is the concentration in soil outdoors (ng/g), tind is time spent indoors (h), tout is time spent outdoors (h), Md is the child's estimated daily ingestion rate for dust (g/day), and Ms is the child's estimated daily ingestion rate for soil (g/day).

where R is the absorption fraction, set to 1, and W is the child's weight (kg).

The EPA Child-Specific Exposure Factor Handbook (USEPA, 2008) recommendations for soil and dust ingestion together are 60 mg/day for children of ages of 6 to <12 months; and 100 mg/day for children of age of 1 to <6 years, but have no further recommendation for the specific distribution between dust and soil as to age. Children in the PEPCOT study were of ages 1 month to 5 years overall (median 3.0 years), and from 1 month to 3 years (median 1.0 years) initially. Therefore, the non-dietary ingestion assumptions Md and Ms for the analysis of the PEPCOT data were adjusted, using the age-stratified defaults suggested by the EPA Integrated Exposure Uptake Biokinetic (IEUBK) model (USEPA, 1999).

The IEUBK model suggests that children between ages 1 and 4 years ingest more dust and soil than either younger or older children. Therefore, the IEUBK defaults for soil and dust ingestion by children of ages 0–6 years were used. These defaults are as follows: age 0–1 year, 85 mg/day; ages 1–4 years, 135 mg/day; ages 4–5 years, 100 mg/day; and ages 5–6 years, 90 mg/day. The values of Md and Ms in the model equation shown above were thus set to 0.085, 0.135, 0.100, or 0.090 g/day, depending on the child's age at the time of sample collection, then weighted between soil and dust depending on the time spent outdoors and indoors, as shown in the equation.

Urinary concentrations (ng/ml) were converted to daily doses (ng/kg/day) by multiplying by a conversion factor of 22.4 ml/kg/day for children's daily urine output obtained from the medical literature (Ballauf et al., 1988; Miller and Stapleton, 1989; Szabo and Fegyverneki, 1995).

### Data Analysis

On completion of the sampling and analytical data, we characterized seasonal differences in pesticide levels and within-household and within-child variance components for the child-specific outcomes (urine, hand-wipe, and food). We developed some descriptive statistics and simple models to characterize the observed trends over time within residential units and within siblings in the same residential unit.

To meet the major study objective, the year-to-year trends in the calculated exposures were explored. Study results were analyzed separately for each combination of target pesticide (or metabolite) and the other environmental or child-specific measurements (air, floor surface dust or wipe, urine, hand surface wipe, or food) using the following statistical model:

where Yij is the natural log-transformed pesticide level for the ith child during the jth sampling campaign (j=1, 2, or 3); β0 and β1 are the fixed-effect intercept and slope associated with the trend in pesticides over time across the entire study sample; Timeij is measured in years, with a value of zero corresponding to 1 July 2004 (the approximate temporal mid-point of PEPCOT). The β3 and τ parameters describe the seasonally adjusted trend over time. Seasonality is fit using a two-stage process that allows identification of the day of the year at which peak exposure occurs and the magnitude of the peak exposure assuming a yearly sinusoidal seasonality pattern among the data.

Random effects were added to the above model to adjust for the anticipated positive correlation among repeated measurements on the same study subject. The childhood pesticide metabolite models allow each child to have their own intercept and slope (exposure curve over time), with an assumption that δ0i and δ1i jointly follow a multivariate normal distribution with mean zero and covariance matrix

. The residual error (ɛij) takes on the interpretation of within-subject variability, and followed a normal distribution with mean zero and variance σerror2.

The parameter of interest in the above model is β1, which characterizes (on the natural-log scale) the yearly reduction in pesticide levels from the first through the third sampling campaign of the study. This model can be applied to different subsets of the study sample, for example, separate analyses for the younger child (YCh) and the older child (OCh) within each household, to allow for additional clarity in childhood exposure patterns, and was fit using the PROC MIXED procedure within SAS statistical software.

A second longitudinal model (age-adjusted) was used as described below:

This model is fit using a linear mixed effects model (Fitzmaurice et al., 2004). Two new explanatory variables are introduced in the model—BaselineAgei, which measures the age (in months) of the ith study subject when first observed (Year 1); and TSBij represents the time-since-baseline measured in years (where the value is equal to zero for the first measure, and then increases over time as appropriate). Two functions (f1 and f2) were introduced to be applied to these two variables, to indicate that their effect on pesticide concentrations may not be linear. These functions were explored as polynomial terms, for example, adding quadratic and cubic terms of TSB, which expanded the β1 and/or β2 parameters from a scalar to a vector in some of the models. This model also adjusts for the effects of seasonality using a similar methodology, as used in the model described earlier. The advantage of the second model is that it adjusts for the age of the child upon baseline, and can therefore be used to explore trends in pesticide exposure over time across the entire study population in a manner that appropriately adjusts for the different ages of the children enrolled in the study, while also properly accounting for broad-based changes in environmental pesticide concentrations that were experienced over time within the study population (for example, the phase-out of CPF).

To meet the secondary objective of the study, the variance components associated with child-specific exposures (aggregate potential dose and pesticide or metabolite concentrations in urine) were estimated, on the basis of a post-modeling analysis of the random effects from the above models. In this analysis, the random effects from children within each household were estimated to create household-level random effects. These household-level random effects (slopes and intercepts) are used to characterize between-household variability in pesticide exposures. The difference between child-specific random effects and household-level random effects was used to assess within-household variability in pesticide exposures. This partitioning of the variance components was pursued only in cases where the entire study population was modeled simultaneously.

This second model was used to describe observed urinary metabolite/pesticide levels as a function of predicted aggregate exposure to both pesticide and metabolite.

## Results and discussion

In the first year, the PEPCOT team personally visited 78 households and obtained informed consent from 51 potential participants, one of whom withdrew. Thus, the study successfully met the recruitment goal of 50 households, with 101 participating children. Participant retention was also successful, with 46 households completing 2 years and 44 households (88%) completing all 3 years of the study. The children in families that completed the study comprised 50 boys and 47 girls, including five sets of twins and one set of triplets. The race and ethnicity of participants were self-identified as 85% white, 4% Black, 2% Hispanic, and 9% Other.

Parents were asked to follow their usual pesticide use habits during the 3 years. Pesticides were applied in the 7 days preceding or during the monitoring period by 26%, 30%, and 28% of households in Years 1, 2, and 3, respectively. Of the target pesticides discussed in this paper, CPF was applied by one family in Years 1 and 3, and two families in Year 2; 2,4-D was applied by one family in Years 1 and 2, and by two families in Year 3.

### Aggregate Potential Dose Levels

Summary statistics for the aggregate potential doses of CPF, DZN, 2,4-D, and PCP; and their metabolites/degradation products TCP and IMP, are presented in Table 1. These aggregate potential dose levels, sometimes described as the intake or applied dose levels, for the 3 years of the field sampling separately and all years together, represent the maximum potentially available dose for intake into the body and possible absorption, but not the estimated absorbed dose. The older of the two siblings in a given household, who was approximately 3 years old (median 2.9 years) at the time of sampling in the first year (Year 1) is indicated by OCh, whereas his or her younger sibling is indicated by YCh throughout this paper. Comparing the aggregate potential doses in Table 1 for OCh and YCh over all 3 years, one sees that the overall geometric mean (GM) potential doses (ng/kg/day) of TCP, IMP, and 2,4-D over the 3 years were higher for OCh than those for YCh, but in contrast, the overall GM potential doses of CPF, DZN, and PCP were higher for YCh than for OCh.

For both the OP pesticides, CPF and DZN, and for PCP (the use and sale of which was restricted several years earlier (USEPA, 1984), the aggregate dose levels for YCh were generally higher, reflecting both the more likely frequent exposures of YCh to these substances through behavior and activities, such as crawling on the floor, hand-to-mouth activities, and more time spent indoors, as well as the smaller body weight of YCh, which increases the potential dose. However, for the degradation products TCP and IMP, and the outdoor herbicide 2,4-D, the potential doses were higher for OCh, who is likely to come in contact with a greater variety of sources, eats a greater variety of foods, and spends more time outdoors.

In each case, the environmental degradation products of the OP pesticides contribute substantially to the overall intakes and hence are likely to contribute to a large portion of the excreted dose of the parent pesticide that might be inferred from urinary measurements. The aggregate potential doses (intake or applied doses) are compared graphically in Figure 1 with the excreted doses calculated from the urine measurements for all children in this study over the 3 sampling years. The Pearson correlation coefficients with their parent compounds were 0.1936 for excreted TCP and the potential intake of its parent CPF, and 0.4662 for excreted TCP and the potential intake of TCP itself; both were significant at the <0.0005 level. For all children in the study over 3 years, the ratio of the GMs in nmol/kg body weight per day (nmol/kg/day) for (CPF+TCP)in/TCPout was 0.857. Thus for CPF, the aggregate potential doses of CPF and TCP taken together are reasonably close to the urinary excreted doses of TCP. This apparent relationship was probably aided by the fact that CPF was phased out in 2001–2002 for most uses that might lead to children's exposures.

For PCP and 2,4-D, which are not metabolically degraded before excretion, the urinary excreted doses were consistent with their aggregate potential doses. The GM ratios of the aggregate potential doses to the urinary excreted doses were 0.60 for PCPin/PCPout and 0.88 for 2,4-Din/2,4-Dout; again accounting for most of the estimated exposure.

However, for DZN and IMP, the GM excreted dose of IMP (0.422 nmol/kg/day) was about 10 times higher than the maximum aggregate potential dose of IMP+DZN (0.043 nmol/kg/day). Although the Pearson correlation coefficient for excreted and intake IMP was 0.1221, significant at the <0.025 level, the correlation coefficient for excreted IMP and intake of its parent DZN was 0.0279, which was not significant. A partial explanation lies in the difference in recoveries for the intake and excreted doses of IMP. Quantitative recoveries (>70%) were obtained for the environmental samples, with the exception of a few food samples, which had recoveries 50–70%, and the results are uncorrected for any losses during analysis. However, the isotope dilution method was used for IMP in urine; this method automatically corrects for sample losses. Because of the low and highly variable absolute recoveries of IMP from the urine samples (typically about 10–30%), this correction can result in an overestimate of the excreted IMP relative to the intake of IMP and DZN. An additional possibility is that another degradation product and metabolic intermediate of DZN—diazoxon—could have been present in the environmental samples (Zhang and Pehkonen, 1999; Larkin and Tjeerdema, 2000; Shemer and Linden, 2006), but was not measured. This degradation product if present could be metabolized to IMP and contribute to the urinary IMP levels. Additional investigation is needed.

Many factors can influence the urinary excretion of the pesticide metabolites and their degradation products, including the behavior of the individuals, the degree of absorption by the body, the metabolic transformation and route of elimination, and the possible existence of intake routes that may not have been considered. If no significant environmental degradation of the parent pesticide to a chemically identical metabolite occurs, then the urinary product may serve as a good biomarker of exposure, especially in the case of high exposures. However, these findings suggest again that measurements of certain pesticide metabolites in the urine of young children, without consideration of the environmental exposures to the metabolites themselves, may not always be sufficiently good indicators of exposures to the parent pesticides.

Changes in the aggregate potential dose levels over time are also evident. For OCh, the GM and median aggregate potential doses (ng/kg/day) of all analytes decreased from Year 1 to Year 3, with the exception of 2,4-D. For YCh, decreasing trends are apparent, with the exception of 2,4-D and the CPF degradation product TCP, which showed a slight increase with time.

The aggregate potential doses of CPF and DZN are well below the established reference doses (RfD), 0.001 and 0.0007 mg/kg/day (1000 and 700 ng/kg/day) for CPF and DZN, respectively (ATSDR, 1997, 2006; IRIS), by factors of roughly 3–100. With additional safety factors of 10 for children's exposures, the highest observed aggregate doses of CPF and DZN were at or near RfD established in EPA's Integrated Risk Information System (IRIS). For PCP, RfD is 0.03 mg/kg/day for acute and 0.001 mg/kg/day (103 ng/kg/day) for chronic exposure. This contrasts to the much lower GM PCP aggregate dose found in this study of approximately 7–9 ng/kg/day for both OCh and YCh. For 2,4-D, RfD is 0.01 mg/kg/day (104 ng/kg/day), which is in contrast to the much lower GM 2,4-D aggregate potential dose found in this study of approximately 8–10 ng/kg/day.

### Potential Dose Levels Through Different Routes of Exposure

Potential doses were estimated for the dietary ingestion, inhalation, and non-dietary/indirect ingestion routes of exposure. These are shown as percentages of the aggregate potential dose in Table 2, presented as the geometric means over all 3 years. All the data appear to follow a log-normal distribution, as expected for most environmental data. The median dose in these data is generally very close to the GM.

For both OCh and YCh, the dietary ingestion and inhalation routes of exposure contributed approximately equally to the aggregate doses of the semivolatile pesticides CPF, DZN, and PCP. In contrast, for the relatively non-volatile degradation products TCP and IMP, and the herbicide 2,4-D, dietary ingestion contributed most (88–97%) to the aggregate doses. This dependence on the volatility of the pesticides and their degradation products makes sense, as the compounds that are more volatile tend to equilibrate between the air and other environmental media with increased concentrations in air depending on temperature and vapor pressure, or be adsorbed and desorbed from the surfaces and dust, which may increase their concentrations in air, especially in the warm indoor environment. Non-dietary/indirect ingestion accounted for <10% of the aggregate doses of CPF, DZN, TCP, 2,4-D, and PCP for both OCh and YCh, on the basis of the measured concentrations of the target compounds in dust and soil and the indirect ingestion assumptions cited earlier (USEPA, 1999)

In comparison with the relative routes of exposure found earlier (2000–2001) in the CTEPP study for preschool children in the Durham County, NC area, the non-dietary contributions were similarly low (2–6%; USEPA, 2004). The dietary contributions to the dose for CPF shown in Table 2 were lower in PEPCOT than that found in CTEPP (CTEPP CPF dietary 54%, inhalation 39%) and also slightly lower for DZN (CTEPP DZN dietary 56%, inhalation 40%). This decreased contribution of the dietary route to overall exposure likely reflects the continuing reduction in their presence in the food supply as a result of the US EPA phase-outs of these two pesticides.

Analysis of the data including the hand surface residue data indicated that the measured hand loadings did not affect the aggregate doses significantly. However, significant Pearson's correlations at <0.0005 level were obtained for urinary levels of IMP with hand loadings of both DZN and of IMP, and urinary levels of PCP with hand loadings of PCP. Urinary levels of 2,4-D were correlated with hand loadings of 2,4-D at <0.005 level. Dermal contact with contaminated surfaces other than floor surfaces and subsequent ingestion through hand-to-mouth and object-to-mouth actions may have contributed to some extent to the aggregate potential dose. This contribution is assumed to be included in the soil/dust ingestion values (USEPA, 1999) used in the calculation of the aggregate dose. As indicated earlier, dermal absorption was assumed to be negligible at the low concentrations measured.

### Longitudinal Modeling

Within this section are shown the results of longitudinal modeling of the data for both Child-Specific Models run separately for OCh and YCh, and an Age-Adjusted Model that combines data across the multiple children observed within each household. The modeling results presented here are for aggregate potential dose, and use imputed data to fill in missing values. Models were explored that adjusted for race/ethnicity and gender; however, these covariates were not statistically significant predictors of pesticide exposures in the clear majority of the analyses. Therefore, the results herein were not adjusted for these demographic characteristics.

### Child-Specific Modeling

Table 3 contains the child-specific modeling results for each analyte. Most effects were significant, as indicated by the bold-faced values of P≤0.05. The table contains the parameter estimate, expressed on a natural log scale, for each effect considered: Intercept—the intercept of the trend line as of 1 July 2004, the mid-period of the PEPCOT study; Time—the slope of the trend line, which indicates the change in the aggregate potential dose with time predicted by the model; Seasonality—the ratio of the seasonal peak to the trend; and Max Day—the day of the year at which the model predicts peak levels for each child. Accompanying the parameter estimates are the associated standard errors and P-values indicating the statistical significance of the effect. Seasonality was assessed using an annual sinusoidal curve, with the seasonality term corresponding to the magnitude of the seasonal peak.

The child-specific modeling results presented here, as well as the longitudinal age-adjusted modeling results that follow, are in a natural logarithm space. For example, the effect “Time” in Table 3, for the OCh, for CPF of −0.276, the slope of a trend line plotting aggregate dose vs time, indicates a decrease in the aggregate potential dose of CPF of e−0.276 or 0.759 ng/kg/day per year over the 3 study years. Similar relationships hold for the “Time since baseline,” the slope of the plots of potential dose vs time, and the change in the potential dose with time in the longitudinal modeling results. The seasonality effect indicates the ratios of the maximum seasonality to the trend line. Comparing the seasonality of CPF with that of DZN for example, the seasonality of CPF, 0.463, indicates a value of e0.463 or 1.59 for this ratio, whereas that for DZN, 0.280, indicates a value of e0.280 or 1.32. This suggests that the seasonal fluctuations for CPF may be larger than those for DZN.

As indicated by the descriptive analysis results presented earlier, the levels measured for each analyte are not always significantly different over time, as detailed below:

• For CPF, DZN, and PCP, the time effect is significant for both OCh and YCh, with significant declines across time.

• For TCP, the time effect is significant for both OCh and YCh, but in the opposite directions, with a decline for OCh and an increase for YCh.

• For IMP, the time effect is significant for the OCh, but is not significant for the YCh, both with declines across time. For 2,4-D, the time effect is not significant.

These results reflect the success of the phase-outs of CPF and DZN with the consequent continued decreases in the exposures of young children to these pesticides. In addition, the influences of the children's developmental stages and behavior are reflected in the different directions of change observed for some analytes.

The seasonality parameter, corresponding to the magnitude of the seasonal peak for a natural log-transformed pesticide or metabolite concentrations, was not significant in all cases, as detailed below:

• For CPF, DZN, and PCP, the seasonality effect is significant for both OCh and YCh, with the maxima in the late fall or winter seasons.

• For IMP, the seasonality effect is significant for OCh, with the maximum in the fall season. However, seasonality is not significant for YCh.

• For TCP and 2,4-D, the seasonality effects associated with either child are not significant.

Because the dietary route accounts for most of the intake of TCP, IMP, and 2,4-D, it is reasonable that seasonality would be minimally or not significant for these compounds. In contrast, the inhalation route accounts for approximately half the intake of CPF, CZN, and PCP. These seasonality results are consistent with the relative volatilities of these more volatile pesticides, which show greater seasonal dependences. The maxima in the late fall or early winter seasons also suggest the influence of weather, as children spend more time indoors, where the relative concentrations of the targeted pesticides are generally higher, in the cooler seasons. The lack of seasonality for IMP for YCh is also consistent with YCh spending more time indoors, where this degradation product of the outdoor pesticide DZN is less likely to be found.

Table 4 contains the detailed modeling output from the longitudinal model that analyzed all the data together. This model did not separately analyze data for OCh and YCh, but rather included age variables (Age at Baseline and Age at Baseline Squared) that adjusted for a child's age at each data collection point. Similarly, the model did not include a Year variable, but included variables that adjusted for time elapsed since the start of the study (Time Since Baseline and Time Since Baseline Squared). If the squared variables were not significant, they were dropped from the model.

In addition to the information included in Table 3 (parameter estimates, SE, P-values, and dates of maximum value), Table 4 includes values for the covariance parameters included to account for random intercept and slope effects. These covariance terms were partitioned into parameters representing home-to-home variability in the random intercept and slope effects, and parameters that represent within-home variability in these random effects. The within-home variance components capture the variability in pesticide exposure among siblings living within the same household. There are three parameters each for within- and between-home variability: $σ β 0 2$, which captures variability in the random intercept effects, $σ β 1 2$, which captures variability in the random slope effects, and $σ β 0 , β 1 2$, which provides the covariance between random intercept and slope effects. The residual error term is listed in the within-home variability column in the table, and provides the variance attributable to error not explained by the model.

Some results of interest are listed below:

• Age at Baseline was significant (at the P≤0.05 level) for all but one of the analytes considered in the modeling. It was not significant for CPF.

• Age at Baseline Squared was significant for TCP, IMP, and 2,4-D, but not significant for CPF, DZN, or PCP. The effect of this parameter was small.

• Time Since Baseline was significant for TCP, DZN, and IMP at the P≤0.05 level, and for CPF at P≤0.06. It was not significant for PCP or 2,4-D.

• Time Since Baseline Squared was significant for DZN only.

• Seasonality was significant for CPF, DZN, IMP, and PCP. It was not significant for TCP and 2,4-D.

In general, the results of the longitudinal modeling are consistent with those observed in the simpler exploratory data analyses. For example, in Table 4, the effect “Time Since Baseline” of −0.1773 for CPF indicates that the model estimates a decrease of e−0.1773=0.8375 ng/kg/day per year in the aggregate potential dose over the course of the study. Similarly, the exploratory data analysis results summarized in Table 1 show a decrease of 1.5–2.3 ng/kg/day per year in the GM aggregate potential dose of CPF, for OCh and YCh, respectively.

As apparent in the last two columns of Table 4, the within-home variability, which reflects the differences in exposures between siblings living in the same home, but at different ages, is relatively small for all six analytes. It is smallest for 2,4-D, which is encountered mainly outdoors rather than within the home. The between-home variability, which reflects the differences in exposures between children of the same age but living in different homes, is much larger. Again, the between-home variability is smallest for 2,4-D.

In conclusion, this study shows that the children's exposures to and potential doses of the OP pesticides CPF and DZN are decreasing with time, which confirms the success of the US EPA restrictions on these pesticides in reducing children's exposures and possible associated health risks. These exposures are also expected to be lower because of the natural maturation and behavioral changes of the children as they grow older and approach school age. Also as expected, dietary ingestion and inhalation remain the primary routes of exposure to these OP pesticides. All of the aggregate potential doses of the target pesticides are well below levels considered to pose health risks. The children's potential doses of the OP metabolites/degradation products TCP and IMP are also continuing to decrease with time, as their parent pesticides are gradually disappearing from the residential environment. Dietary ingestion is still an important route of exposure to the parent pesticides and to their metabolites/degradation products. The relatively high levels of TCP and IMP in environmental media including food, compared with those of CPF and DZN, can confound estimates of aggregate exposure based solely on the metabolite excretion in urine.

Interestingly, the children are still exposed to some extent to low levels of PCP, despite the severe reductions in its environmental uses since 1984. Fortunately, these exposures and potential doses are also well below levels considered to be of risk, and they are decreasing with time.

## References

1. Adgate J.L., Barr D.B., Clayton C.A., Eberly L.E., Freeman N.C., Lioy P.J., Needham L.L., Pellizzari E.D., Quackenboss J.J., Roy A., and Sexton K. Measurement of children's exposure to pesticides: analysis of urinary metabolite levels in a probability-based sample. Environ Health Perspect 2001: 109: 583–590.

2. Alexander B.H., Mandel J.S., Baker B.A., Burns C.J., Bartels M.J., Acquavella J.F., and Gustin C. Biomonitoring of 2,4-dichlorophenoxyacetic acid exposure and dose in farm families. Environ Health Perspect 2007: 115: 370–376.

3. Arcury T.A., Grzywacz J.G., Barr D.B., Tapia J., Chen H., and Quandt S.A. Pesticide urinary metabolite levels of children in eastern North Carolina farmworker households. Environ Health Perspect 2007: 115: 1254–1260.

4. Arcury T.A., Grzywacz J.G., Davis S.W., Barr D.B., and Quandt S.A. Organophosphorus pesticide urinary metabolite levels of children in farmworker households in eastern North Carolina. Am J Ind Med 2006: 49: 751–760.

5. ASTM. Standard practice for collection of floor dust for chemical analysis D5438-94. Annual Book of ASTM Standards. American Society for Testing and Materials, West Conshohoken, PA,, 1997, Vol. 11.03, pp. 517–523.

6. ATSDR. Toxicological Profile for Chlorpyrifos. U.S. Department of Health and Human Services, Agency for Toxic Substances and Disease Registry, Atlanta, GA, 1997.

7. ATSDR. Draft Toxicological Profile for Diazinon. U.S. Department of Health and Human Services, Agency for Toxic Substances and Disease Registry, Atlanta, GA, 2006.

8. Ballauf A., Kersting M., and Manz F. Do children have an adequate fluid intake? Water balance studies carried out at home. Ann Nutr Metab 1988: 32: 332–339.

9. Bradman M.A., Harnly M.E., Draper W., Seidel S., Teran S., Wakeham D., and Neutra R. Pesticide exposures to children from California's central valley—Results of a pilot study. J Expo Anal Environ Epidemiol 1997: 7: 217–234.

10. Chuang J.C., Lyu C., Chou Y.-L., Callahan P.J., Nishioka M., Andrews K., Pollard M.A., Brackney L., Hines C., Davis D.B., and Menton R. Evaluation and application of methods for estimating children's exposure to persistent organic pollutants in multiple media. Report, Contract 68-D4-0023, EPA 600/R-98/164a, 164b, and 164c 1998.

11. Curl C., Fenske R.A., and Elgethun K. Organophosphorus exposure of urban and suburban preschool children with organic and conventional diets. Environ Health Perspect 2003: 111: 377–382.

12. Curwin B.D., Hein M.J., Sanderson W.T., Striley C., Heederik D., Kromhout H., Reynolds S.J., and Alavanja M.C. Pesticide dose estimates for children of Iowa farmers and non-farmers. Environ Res 2007: 105: 307–315.

13. Federal Register. Volume 65, Number 235, pp 76233–76240 (6 December 2000).

14. Federal Register Volume 66, Number 139, pp 37673–37678 (July 19, 2001), and Volume 68, Number 170, pp 52394–52397 (September 3 2003).

15. Fenske R.A., Lu C., Barr D., and Needham L. Children's exposure to chlorpyrifos and parathion in an agricultural community in central Washington State. Environ Health Perspect 2002: 110: 549–553.

16. Fenske R.A., Lu C., Curl C.L., Shirai J.H., and Kissel J.C. Biologic monitoring to characterize organophosphorus pesticide exposure among children and workers: an analysis of recent studies in Washington State. Environ Health Perspect 2005: 113: 1651–1657.

17. Fitzmaurice G.M., Laird N.M., and Ware J.H. Applied Longitudinal Analysis. John Wiley & Sons, Inc., Hoboken, New Jersey, 2004.

18. FQPA Food Quality Protection Act of 1996 http://www.epa.gov/pesticides/regulating/laws/fqpa/.

19. Garfitt S.J., Jones K., Mason H.J., and Cocker J. Exposure to the organophosphate pesticide diazinon: data from a human volunteer study with oral and dermal doses. Toxicol Lett 2002: 134: 105–113.

20. Goldman L. Case studies of environmental risks to children. Future Child 1995: 5: 27–33.

21. Griffin P., Mason H., Heywood K., and Cocker J. Oral and dermal absorption of chlorpyrifos: a human volunteer study. Occup Environ Med 1999: 56: 10–13.

22. Heudorf U., Angerer J., and Drexler H. Current internal exposure to pesticides in children and adolescents in Germany: urinary levels of metabolites of pyrethroid and organophosphorus insecticides. Int Arch Occup Environ Health 2004: 77: 67–72.

23. IRIS EPA Integrated Risk Information System http://www.epa.gov/ngispgm3/iris/.

24. Kissel J.C., Curl C.L., Kedan G., Lu C., Griffith W., Barr D.B., Needham L.L., and Fenske R.A. Comparison of organophosphorus pesticide metabolite levels in single and multiple daily urine samples collected from preschool children in Washington State. J Expo Anal Environ Epidemiol 2005: 15: 164–171.

25. Koch D., Lu C., Fisker-Andersen J., Jolley L., and Fenske R.A. Temporal association of children's pesticide exposure and agricultural spraying: report of a longitudinal biological monitoring study. Environ Health Perspect 2002: 110: 829–833.

26. Lambert W.E., Lasarev M., Muniz J., Sherer J., Rothlein J., Santana J., and McCauley L. Variation in organophosphate pesticide metabolism in urine of children living in agricultural communities. Environ Health Perspect 2005: 113: 504–508.

27. Landrigan P.J., Claudio L., Markowitz S.B., Berkowitz G.S., Brenner B.L., Romero H., Wetmur J.G., Matte T.D., Gore A.C., Godbold J.H., and Wolff M.S. Pesticides and inner-city children: exposures, risks, and prevention. Environ Health Perspect 1999: 107: 431–437.

28. Larkin D.J., and Tjeerdema R.S. Fate and effects of diazinon. Rev Environ Contam Toxicol 2000: 166: 49–82.

29. Loewenherz C., Fenske R.A., Simcox N.J., Bellamy G., and Kalman D. Biological monitoring of organophosphorus pesticide exposure among children of agricultural workers in Washington State. Environ Health Perspect 1997: 105: 1344–1353.

30. Lu C., Kedan G., Fisker-Andersen J., Kissel J.C., and Fenske R.A. Multipathway organophosphorus pesticide exposures of preschool children living in agricultural and nonagricultural communities. Environ Res 2004: 96: 283–289.

31. Lu C., Knutson D.E., Fisker-Andersen J., and Fenske R.A. Biological monitoring of organophosphorus pesticide exposure among pre-school children in the Seattle metropolitan area. Environ Health Perspect 2001: 109: 299–303.

32. Mendola P., Selevan S.G., Gutter S., and Rice D. Environmental factors associated with a spectrum of neurodevelopmental deficits. Ment Retard Dev Disabil Res Rev 2002: 8: 188–197.

33. Meuling W.J., Ravensberg L.C., Roza L., and van Hemmen J.J. Dermal absorption of chlorpyrifos in human volunteers. Int Arch Occup Environ Health 2005: 78: 44–50.

34. Miller L.A., and Stapleton F.B. Urinary volume in children with urolithiasis. J Urol 1989: 141: 918–920.

35. Morgan M.K., Sheldon L.S., Croghan C.W., Jones P.A., Robertson G.L., Chuang J.C., Wilson N.K., and Lyu C.W. Exposures of preschool children to chlorpyrifos and its degradation product 3,5,6-trichloro-2-pyridinol in their everyday environments. J Expo Anal Environ Epidemiol 2005: 15: 297–309.

36. Morgan M.K., Sheldon L.S., Thomas K.W., Egeghy P.P., Croghan C.W., Jones P.A., Chuang J.C., and Wilson N.K. Adult and children's exposure to 2,4-D from multiple sources and pathways. J Expo Sci Environ Epidemiol 2008: 18: 486–494.

37. Nishioka M., McCauley M., Andrews K., Lukuch C., Wilkins III J., and Fortmann R. Development, validation, and field use of a novel method for extracting and analyzing organophosphate (OP) and pyrethroid pesticide metabolites and creatinine from commercial disposable diapers. Epidemiol 2006: 17: S176.

38. O’Rourke M.K., Lizardi P.S., Rogan S.P., Freeman N.C., Aguirre A., and Saint C.G. Pesticide exposure and creatinine variation among young children. J Expo Anal, Environ Epidemiol 2000: 10: 672–681.

39. Perera F.D. Environment and cancer: who are susceptible? Science 1977: 278: 1068–1073.

40. Perera F.D., Raub V., Whyatt R.M., Tang D., Tsai W.Y., Bernert J.T., Tu Y.H., Andrews H., Barr D.B., Camann D.E., Diaz D., Dietrich J., Reyes A., and Kinney P.L. A summary of recent findings on birth outcomes and developmental effects of prenatal ETS PAH, and pesticide exposures. Neurotoxicology 2005: 26: 573–587.

41. Schettler T. Toxic threats to neurologic development of children. Environ Health Perspect 2001: 109 (Suppl. 6): 813–816.

42. Shemer H., and Linden A.G. Degradation and by-product formation of diazinon in water during UV and UV/H2O2 treatment. J Hazard Mat 2006: 136: 553–559.

43. Szabo L., and Fegyverneki S. Maximum and average urine flow rates in normal children—the Miskole nomograms. Brit J Urol 1995: 76: 16–20.

44. USEPA. Position Document 4 for Wood Preservative Pesticides: Creosote, Pentachlorophenol, and Organic Arsenicals 1984 Federal Register 70, 16278 (2005).

45. USEPA. Exposure Factors Handbook. EPA/600/P-95/002Fa-c 1997: http://www.epa.gov/ncea/pdfs/efh/front.pdf.

46. USEPA. Short Sheet: IEUBK Model Soil/Dust Ingestion Rates, U.S., Environmental Protection Agency, Washington, DC. EPA #540-F-00-007, OSWER #9285.7-33 1999: http://www.epa.gov/superfund/lead/products/ssircolo.pdf.

47. USEPA Chlorpyrifos Revised Risk Assessment and Risk Mitigation Measures. 2000a: http://www.epa.gov/pesticides/op/chlorpyrifos/consumerqs.htm.

48. USEPA Diazinon: Phase Out of all Residential Uses of the Insecticide. 2000b: http://www.pa.gov/pesticides/factsheets/chemicals/diazinon-factsheet.htm.

49. USEPA. A pilot study of children's total exposure to persistent pesticides and other persistent organic pollutants (CTEPP), EPA/600/R-041/193 2004: http://www.epa.gov/heasd/ctepp/ctepp_report.pdf.

50. USEPA. Child-Specific Exposure Factors Handbook. EPA/600/R-06/096F 2008: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=199243.

51. Valcke M., Samuel O., Bouchard M., Dumas P., Belleville D., and Tremblay C. Biological monitoring of exposure to organophosphate pesticides in children living in peri-urban areas of the Province of Quebec, Canada. Int Arch Occup Environ Health 2006: 79: 568–577.

52. Wessels D., Barr D.B., and Mendola P. Use of biomarkers to indicate exposure of children to organophosphate pesticides: implications for a longitudinal study of children's environmental exposures. Environ Health Perspect 2003: 111: 1939–1946.

53. Wilson N.K., Chuang J.C., Iachan R., Lyu C., Gordon S.M., Morgan M.K., Őzkaynak J., and Sheldon L.S. Design and sampling methodology for a large study of preschool children's aggregate exposures to persistent organic pollutants in their everyday environments. J Expo Anal Environ Epidemiol 2004: 14: 260–274.

54. Wilson N.K., Chuang J.C., and Lyu C. Levels of persistent organic pollutants in several child day care centers. J Expo Anal Environ Epidemiol 2001: 11: 449–458.

55. Wilson N.K., Chuang J.C., Lyu C., Menton R., and Morgan M.K. Aggregate exposures of nine preschool children to persistent organic pollutants at day care and at home. J Expo Anal Environ Epidemiol 2003: 13: 187–202.

56. Wilson N.K., Chuang J.C., Morgan M.K., Lordo R.A., and Sheldon L.S. An observational study of the potential exposures of preschool children to pentachlorophenol, bisphenol-A, and nonyphenol at home and daycare. Environ Res 2007: 103: 9–20.

57. Zhang Q., and Pehkonen S.O. Oxidation of diazinon by aqueous chlorine: kinetics, mechanisms, and product studies. J Agric Food Chem 1999: 47: 1760–1766.

## Acknowledgements

This research was funded in part by the USEPA National Center for Environmental Research (NCER) through STAR Grant R829363 to Battelle Memorial Institute. We also thank Dr. Chris Saint, the Project Officer at NCER; Christopher Lyu and the field staff at Battelle's Center for Public Health Research in Durham, NC; and Marielle C. Brinkman, the database staff, Kelley M Hand and the laboratory staff at Battelle's Columbus, OH laboratories.

## Author information

Authors

### Corresponding author

Correspondence to Nancy K Wilson.

## Ethics declarations

### Competing interests

The authors declare no conflict of interest.

## Rights and permissions

Reprints and Permissions

Wilson, N., Strauss, W., Iroz-Elardo, N. et al. Exposures of preschool children to chlorpyrifos, diazinon, pentachlorophenol, and 2,4-dichlorophenoxyacetic acid over 3 years from 2003 to 2005: A longitudinal model. J Expo Sci Environ Epidemiol 20, 546–558 (2010). https://doi.org/10.1038/jes.2009.45

• Accepted:

• Published:

• Issue Date:

### Keywords

• exposure
• potential dose
• pesticides
• children
• modeling
• longitudinal
• variability

• ### Development of a questionnaire-based insecticide exposure assessment method and comparison with urinary insecticide biomarkers in young Australian children

• Karin English
• , Yan Li
• , Paul Jagals
• , Robert S. Ware
• , Xianyu Wang
• , Chang He
• , Jochen F. Mueller
•  & Peter D. Sly

Environmental Research (2019)

• ### Assessment of residential environmental exposure to pesticides from agricultural fields in the Netherlands

• Maartje Brouwer
• , Hans Kromhout
• , Roel Vermeulen
• , Jan Duyzer
• , Henk Kramer
• , Gerard Hazeu
• , Geert de Snoo
•  & Anke Huss

Journal of Exposure Science & Environmental Epidemiology (2018)

• ### Occurrence of imidacloprid, carbendazim, and other biocides in Italian house dust: Potential relevance for intakes in children and pets

• Severyn Salis
• , Cecilia Testa
• , Sara Armorini
• , Nicola Rubattu
• , Angelo Ferrari
• , Roberto Miniero
•  & Gianfranco Brambilla

Journal of Environmental Science and Health, Part B (2017)

• ### Factors that contribute to insecticide poisoning among immigrant agricultural workers in Thailand

• Anamai Thetkathuek
•  & Wanlop Jaidee

International Journal of Occupational and Environmental Health (2017)

• ### Adult and child urinary 2,4-D in cities with and without cosmetic pesticide bylaws: a population-based cross-sectional pilot study

• Scott A Venners
• , Neda Khoshnood
• , Matthew Jeronimo
• , Aaron Sobkowicz
• , Philip Provencher
• , Guanting Tang
• , Winnie Chu
•  & Ray Copes

Journal of Exposure Science & Environmental Epidemiology (2017)