We used seven populations enrolled in five independent studies, all of which took place in Washington State between 1994 and 1999, in this analysis. Our previous reports included detailed descriptions of population recruitment, sample collection, and sample analysis (Curl et al. 2002; Koch et al. 2002; Loewenherz et al. 1997; Lu et al. 2000, 2001; Simcox et al. 1999; Thompson et al. 2003). Tables 1 and 2 provide geographic location, number and age of participants, and number of samples collected in these studies and describe relevant occupational or para-occupational factors, as well as relationships between sampling time and active crop spraying with OP pesticides.

The five studies examined in this analysis included 437 children and 233 adults, who provided > 2,000 urine samples. Maximum values and selected percentiles for DMTP levels in the urine of the adult farmworkers, applicator children, farmworker children, and Seattle children in the cross-sectional studies are presented in Table 3. At the median (50th percentile), DMTP concentrations were highest in children of applicators (25 μg/L), followed by adult farmworkers (10 μg/L), and then children living in the Seattle area and children of farmworkers (6.1 and 5.8 μg/L, respectively). This pattern continued at the 75th percentile, with applicator children higher than adult farmworkers (44 vs. 32 μg/L), and Seattle children higher than farmworker children (17 vs. 13 μg/L). Even at the 90th percentile, the DMTP concentrations for applicator children were greater than for adult farmworkers (110 vs. 99 μg/L) and the other two groups. At the 95th percentile, however, this trend changed, with adult farmworkers having the highest concentrations (180 μg/L), followed by applicator children (130 μg/L), farmworker children (50 μg/L), and finally Seattle children (39 μg/L). The cumulative distributions for the top 50th percentile of these four populations are presented in Figure 2.

Metabolite concentrations in applicator children were higher than those in adult farmworkers (p = 0.02), although this difference was not statistically significant when correction for multiple comparisons was included in the analysis. DMTP concentrations in the urine of the applicator children were higher than those of either the farmworker children (Mann-Whitney U-test, p < 0.0001) or the Seattle children (Mann-Whitney U-test, p < 0.0001). Adult farmworker concentrations were higher than those of either the farmworker children (Wilcoxon matched pairs, p < 0.0001) or the Seattle children (Mann-Whitney U-test, p = 0.002). DMTP concentrations in farmworker children were not significantly different from those in Seattle children (p = 0.73). This information is summarized in Table 4.

We compared DMTP levels in these four populations qualitatively with DMTP levels in the NHANES III population survey (CDC 2003) for children 6–11 years of age (the youngest population sampled) and Mexican Americans (most of the adult farmworkers in the 1999 cross-sectional study were Hispanic). Table 3 presents DMTP levels for the 25th, 50th, 75th, 90th, and 95th percentiles. Applicator children had higher levels than did the NHANES III children at all percentiles. Farmworker children and Seattle children had higher DMTP concentrations at the 50th percentile than did the NHANES III children, but not at the higher percentiles. DMTP concentrations for the adult farmworkers were consistently higher than the NHANES III Mexican-American subgroup.

Table 5 presents the distributions for composite DMAP concentrations from the Seattle children and the farmworker children. DMAP concentrations were significantly higher for the Seattle children compared with the farmworker children (median values of 117 nmol/L and 87 nmol/L, respectively; p = 0.007). At the 50th percentile, DMAP concentrations were similar for the NHANES III children (6–11 years of age) and the farmworker children. At the 75th and 90th percentiles, the NHANES III values were similar to the Seattle children values and substantially higher than the farmworker children values. At the 95th percentile, the NHANES III values were substantially higher than either the Seattle children or farmworker children values.

Distributions of DMTP levels in the repeated-measures studies are presented in Table 6, which provides the 25th, 50th, 75th and 95th percentiles of the arithmetic mean urinary DMTP concentrations for the apple thinners, the farm community children (spray season), and the farm community children (nonspray season). DMTP concentrations among the apple thinners, who worked in fields soon after crop spraying, were one to two orders of magnitude greater than those of the farm community children. The 50th percentile DMTP concentration for the apple thinners (530 μg/L; Table 5) was > 50 times higher than that of the adult farmworkers (10 μg/L; Table 2). As reported by Koch et al. (2002), metabolite concentrations were significantly higher for the farm community children during the spray season than during the nonspray season.

Figure 2 includes data points describing the 75th and 95th percentiles of the arithmetic mean urinary DMTP concentrations for the farm community children during both the spray and nonspray seasons. Metabolite levels for the apple thinners were beyond the scale of this figure. Figure 2 demonstrates that, at the 75th percentile, all children except for the applicator children had roughly similar metabolite levels. At the 95th percentile, the farm community children (spray season) demonstrated higher levels than did the Seattle children, the farmworker children, and the farm community children (nonspray season). The maximum value for the farm community children (spray season) also exceeded the maximum value for the farmworker children [180 μg/L (Table 6) and 140 μg/L (Table 3), respectively].

The two most striking findings from this analysis were the relatively high levels of metabolites among applicator children compared with the other study groups and the similarity in metabolite levels between farmworker children sampled in 1999 and Seattle metropolitan area children sampled in 1998.

DMTP levels in children of pesticide applicators sampled in 1995 were significantly higher than those in the farmworker children in 1999, and were also higher than those in farmworkers in 1999 up to the 90th percentile. Direct comparison of these populations is problematic because pest control strategies may have changed over the 4 years that separated these studies. Pesticide use practices in this region are geared to many factors, including crop type, weather, pest infestation levels, and adoption of less chemical-intensive integrated pest management techniques. Additionally, the 1995 study sampled children during the active crop spraying season, whereas the 1999 sampling occurred for the most part after the peak spraying period.

One way to gauge changes in pesticide use that is less affected by the time of sampling is to examine pesticide concentrations in household dust in these studies. Such an approach is based on the observation that pesticide concentrations in house dust are less susceptible to short-term fluctuations than are urinary metabolite measurements, and on the assumption that changes in pesticide levels in dust reflect changes in pesticide use practices and therefore differences in exposure opportunity. In a previous study, we noted that ethyl parathion concentrations in house dust decreased dramatically after this compound was withdrawn from agricultural use (Fenske et al. 2002). Figure 3 provides the median house dust concentrations of azinphos-methyl and phosmet—the primary dimethyl OP pesticides used in regional tree fruit production—from studies conducted in 1992 (Simcox et al. 1995), 1995 (Lu et al. 2000), and 1999 (Curl et al. 2002). These concentrations decreased over time: Azinphosmethyl concentrations in 1999 were about half those measured in 1995, and phosmet concentrations decreased even more dramatically. Other factors that might explain this difference include relatively high parental exposures during pesticide handling with consequent para-occupational exposure for the children, and the close proximity of the homes of many applicators to pesticide-treated farmland.

A second striking finding from this analysis was the lack of a significant difference between DMTP levels in farmworker children and Seattle children (Table 3) and the significantly higher DMAP concentrations among Seattle children compared with those in the farmworker children (Table 5). We had anticipated that the farmworker children would exhibit higher metabolite concentrations, given the widespread use of dimethyl OP pesticides in the Yakima Valley where they resided. The Seattle population consisted of 2- to 5-year-old healthy children living in the Seattle metropolitan area who had no known risk factors for pesticide exposure; less than half of the parents of these children reported use of any pesticides on lawns, gardens, indoors, or pets (Lu et al. 2001). The farmworker children were of a similar age, resided in the lower Yakima Valley about 150 miles east of Seattle, and were also considered a healthy population; all farmworker children lived with at least one adult actively engaged in farm labor (Thompson et al. 2003). It appears that most of these farmworker children had dimethyl OP pesticide metabolite concentrations lower than or indistinguishable from those of children living in urban and suburban environments whose parents did not work with pesticides. We have also examined the DMTP and DMAP metabolite concentrations of child subgroups in the Yakima Valley study based on the agricultural task of the adult farmworker (e.g., pesticide application, crop thinning) and did not find differences across these subgroups (Fenske et al. 2004). That the farmworker children did not have unusually high metabolite levels is further confirmed by comparison of these data with the 1999–2000 NHANES data for children 6–11 years of age (Tables 3, 5). Data comparisons across laboratories can be problematic, but in this case the CDC laboratory that analyzed the NHANES samples and the University of Washington laboratory that analyzed the present results had participated in a round-robin test for the DAP compounds, and these labs were found to produce comparable results (James et al. 2003).

DAP metabolite measurements probably reflect exposures that occurred in the previous 3–5 days. Most of the farmworker children’s samples were collected well after dimethyl OP pesticide crop spraying in the region, so the measured urinary metabolite concentrations did not necessarily capture peak exposures for this population that may have occurred in the late spring and early summer. The longitudinal study of children in an agricultural community reviewed here (Koch et al. 2002) provides persuasive evidence that DAP metabolite levels can increase during such spraying periods, and the 1995 study of children of pesticide applicators sampled during the active crop spraying period had higher metabolite levels than did other children in the same community (Loewenherz et al. 1997; Lu et al. 2000).

Our comparison of metabolite levels in farmworker children and Seattle children led us to conduct a dietary exposure study in the Seattle metropolitan area. We examined OP pesticide exposures in preschool children who consumed primarily organic juice and produce and those of children who consumed conventional (nonorganic) foods (Curl et al. 2003). We found that diet appeared to be the primary contributor to OP pesticide exposure among these children: The median DMAP concentration for children consuming organic juice and produce was about five times lower than for children with conventional diets (30 vs. 170 nmol/L, respectively). The median DMAP concentration for the farmworker children in the Yakima Valley study was 87 nmol/L. We speculate that a conventional diet rich with juices and fresh produce may have been more common among the Seattle children compared with the farmworker children. Supporting evidence for the importance of dietary differences was reported recently for a family in Germany (Heudorf et al. 2004). DAP concentrations were relatively low in a father and son compared with the mother’s levels. It was then learned that the mother had a special diet with a very high intake of fresh fruit. Substitution of supermarket fruit with organic (pesticide-free) fruit reduced the mother’s DAP concentration to those of her family members.

A third finding of this analysis was the 50-fold difference in DMTP concentrations between apple thinners and adult farmworkers. Although the apple thinner study was conducted in 1994, the metabolite levels measured at the time are considered representative of 1999 exposures, based on dislodgeable foliar residues measured at the time, allowable application rates, and U.S. Environmental Protection Agency estimates (Fenske et al. 2003). Biologic measurements of farmwork-ers’ pesticide exposure need to be collected during active work periods to capture peak exposures. It is interesting to note, however, that the adult farmworkers in the 1999 study had higher DMTP levels than did the NHANES III Mexican-American population, despite the fact that most farmworker samples were collected after peak exposures. This comparison supports the view that farmwork-ers represent an important subpopulation of Mexican Americans with respect to pesticide exposure.

Several other studies have employed DAP metabolites to estimate OP pesticide exposures among children. DAP concentration data from most of these studies have been compared in a recent publication (Barr et al. 2004a). A 1995 study in central Italy included collection of spot urine samples from 195 school children 6–7 years of age (Aprea et al. 2000). The primary findings of the study were that DAP concentrations were higher if pest control operations had been performed inside or outside the house in the preceding month, and that higher DAP concentrations were found for children than for a comparable adult population. Results were expressed as nanomole per gram of creatinine and so could not be compared directly with the present data. A 1998 study in Frankfurt am Main, Germany, involved collection of spot urine samples from residents in former U.S. Forces housing, including 309 children ≤5 years of age (Heudorf and Angerer 2001; Heudorf et al. 2004). These children had higher DAP concentrations than did adults (Heudorf and Angerer 2001). The median DMTP concentration among these children was 18.8 μg/L (Heudorf et al. 2004), exceeding the median concentrations of all of our study groups except applicator children. The authors concluded that the primary source of OP pesticide exposure in this population was from diet, and that these exposures in children may reach and even exceed the World Health Organization’s acceptable daily intake values (Heudorf et al. 2004).

A 1997 study of farmworker families near Fresno, California, included collection of spot urine samples from 9 children and 18 adults (Mills and Zahm 2001). Most samples did not have detectable levels of the six DAP metabolites. Frequency of detection was higher among children than among adults. The mean DMTP concentration (13 μg/L) for children was similar to the mean value (14 μg/L) found in the study of farmworker children in Washington (Curl et al. 2002).

A 2000 study in an agricultural community near the U.S.–Mexico border in southern Texas included collection of spot urine samples from 41 children (Shalat et al. 2003). Only eight of these samples (19.6%) had detectable levels of DMTP. The authors concluded that wipe samples of children’s hands may serve as a better exposure metric for epidemiologic studies than do house dust samples. Comparison with the Washington State studies was not possible because DMTP detection frequency was low and DMAP concentrations were not reported.

A 2000 study in Imperial County, California, focused on 20 children 11–17 months of age. The study examined the relationship between proximity of homes to treated farmland and DAP concentrations in the urine of the young children living in these homes (Royster et al. 2002). The median DMTP concentration reported was 3.2 μg/L, or about one-half that observed in our studies of farmworker and Seattle children. No significant difference was found between DAP concentrations of children living within 400 m (one-quarter mile) and those living more than 400 m from an agricultural field. It is not clear that the statistical analysis (nonparametric Mann-Whitney U-test) had sufficient power to detect a difference, given the sample sizes of five and nine, respectively. The authors did not compare DAP concentrations in the children living nearest to (< 400 m) and farthest from (> 800 m) farmland. It is interesting to note that the mean DAP concentration in the first group was 4.4 times higher than the mean concentration for the second group (123 vs. 28 μg/g creatinine), indicating that the most highly exposed children lived nearer to farmland. The 1995 study of applicator children in Washington State found that DAP concentrations were higher in children who lived closer to agricultural fields (Loewenherz et al. 1997; Lu et al. 2000), using several distance categories that were < 400 m, and a larger sample size. This study also focused on children more likely to spend time outdoors (2- to 5-year-olds) than did the Imperial County study (1- to 2-year-olds).

This analysis makes evident that measurements of short-lived metabolites in urine show marked variability both within and across different studies. Children of pesticide applicators sampled during the active spraying season exhibited high DAP concentrations relative to other child populations. Children in an agricultural community without pesticide applicators in their households exhibited higher DAP concentrations during the active crop spraying season than during the rest of the year. Children of farmworkers sampled largely outside of the peak spraying season had DAP concentrations similar to or lower than those of children in the Seattle metropolitan area. This analysis highlights the importance of sample timing in biomarker studies of pesticide exposure and suggests that identification of high-exposure subpopulations in urban and rural communities can be challenging. Future studies should be designed as longitudinal investigations with frequent repeated measurements to capture peak exposures and characterize intrapersonal and interpersonal variability. This analysis also highlights the difficulty of designing epidemiologic studies to evaluate potential health effects of pesticide exposure in children. It cannot be assumed that children in agricultural communities have higher exposures than children in other environments, without taking into account crop spraying patterns and parental contact with pesticides. Studies that seek to categorize children’s exposure in these communities will need to sample both peak and nonpeak exposure periods and will need to evaluate multiple exposure pathways.