Pesticides refer to a broad class of chemicals that prevent, destroy, repel, or mitigate any pests [1], such as unwanted insects (insecticides, insect repellents), weeds (herbicides), microbes (fungicides, disinfectants), and mice and rats (rodenticides) [2]. Due to their widespread use, exposures to pesticides have been documented in humans globally. Pesticides such as the insect repellent N-N-diethyl-meta-toluamide (DEET), the herbicides 2,4-dichlorophenoxyacetic acid (2,4-D) and 2,4,5-trichlorophenoxyacetic acid (2,4,5-T), and the metabolites of various pyrethroid insecticides (e.g., permethrin) have been measured in the urine of children [3–8] and adults [9, 10], including pregnant women [11–13]. These pesticides have also been measured in umbilical cord blood and sera [14, 15] and meconium [16] of newborn humans, and the breast milk of lactating humans [17, 18] and other mammals [19, 20], which suggests that fetal and early post-natal exposure to these pesticides is possible.

In the general population, skin contact is the primary route of exposure to DEET (e.g., applied directly to skin or clothing or bedding that makes contact with skin) [21], whereas exposure to both pyrethroids and chlorophenoxy acids occurs mainly through ingestion of tainted crops [20, 22]. However, depending on the scenario and population (e.g., occupational settings, improper product use), exposure to all of these pesticides may occur via several routes from a number of sources [20–22]. Epidemiology studies have linked exposures to pyrethroids or chlorophenoxy acids with lymphatic or blood cancers [6, 23, 24], and several adverse endocrine-related conditions, such as poor semen quality [25–28] and altered serum hormone levels [29]. In comparison, evidence of the impact of DEET on human health is limited and mostly derived from illness surveys and case reports [21]. Animal studies involving inhalation or oral exposures to DEET have demonstrated neurotoxicity of varying severity depending on the circumstances surrounding the exposures [21].

Despite the potential adverse health risks to both the mother and developing fetus, exposure studies in pregnant women have been limited for DEET, pyrethroids, and chlorophenoxy acids [11–13, 30–32]. Thus, additional exposure studies are necessary, which will in turn assist with understanding the potential burden of exposure globally, identifying sources of exposure, and designing epidemiology studies. Pregnant women living in Puerto Rico are one at-risk and important population to study as Puerto Rico has an unexplained increase in adverse pregnancy outcomes such as preterm birth over the past several decades [33] and the island has a history of pesticide drift associated with agricultural operations [34], pesticide contaminated hazardous waste sites [35], use of illegal pesticides [36], and use of pesticides in illegal ways (e.g., applying non-approved pesticides to crops) [37]. The primary aims of this study were to: 1) describe distributions, 2) assess within-subject temporal variability, and 3) identify predictors of urinary concentrations of DEET and two of its metabolites, four metabolites of synthetic pyrethroids, and two chlorophenoxy acids in a cohort of pregnant women from Puerto Rico. To our knowledge, this is the first biomarker study to report on exposures to select pesticides among pregnant women in Puerto Rico.

Table 1 shows the distributions of the urinary biomarkers relative to those in U.S. women ages 18–40 from NHANES. In this sample of pregnant Puerto Rican women, 152 urine samples were available for analysis for most analytes, except for 4-F-3-PBA (N = 116) and 3-PBA (N = 141) due to quality control issues and the presence of interfering compound(s) in the urine, respectively. All analytes were detected in fewer than 50% of the samples, except DCBA, which was detected in 79%. Ninety-fifth percentile urinary concentrations of DEET (<LOD = 0.1 ng/ml vs. 0.2 ng/ml), 3-PBA (2.3 ng/ml vs. 6.5 ng/ml), trans-DCCA (2.4 ng/ml vs. 3.3 ng/ml), and 2,4-D (0.6 ng/ml vs. 1.1 ng/ml) were lower than the U.S. population-based sample, whereas 95th percentile urinary concentrations of 4-F-3-PBA (both < LOD = 0.1 ng/ml), cis-DBCA (both < LOD = 0.5 ng/ml), and 2,4,5-T (both < LOD = 0.1 ng/ml) were comparable in both cohorts.The ICCs for urinary concentrations of DCBA uncorrected and corrected for SG were 0.60 (95% CI: 0.45, 0.74) and 0.60 (95% CI: 0.44, 0.73), respectively, demonstrating fair to good reproducibility. As shown in Figure 1, the distributions of urinary levels of DCBA were also not significantly different between the three time points measured. No ICCs or Friedman tests were calculated for the other analytes because detection frequencies were <50%.

Table 2 shows the associations between time of urine collection or demographic characteristics and urinary concentrations of pesticide biomarkers. There were no statistically significant associations between urinary concentrations of the analytes and time of urine collection (AM vs. PM) or woman’s age (<24 vs. ≥24 years). However, less educated women (≤12 years of education) were more likely to have greater urinary concentrations of DCBA (β: 0.9, 95% CI: 0.1, 1.7; geometric means of less educated vs. more educated women: 8.0 ng/ml vs. 3.4 ng/ml) and trans-DCCA (OR: 4.4, 95% CI: 1.3, 15.6) compared with more educated women. Married women were more likely to have detectable urinary concentrations of 2,4-D (OR: 3.5, 95% CI: 1.5, 8.4) relative to unmarried women. Unemployed women were also more likely to have detectable urinary concentrations of DHMB (OR: 2.7, 95% CI: 1.1, 6.3) compared to employed women.

Shown in Table 3 are the associations between the consumption of select fruits, vegetables, or legumes in the past 48-hr and urinary concentrations of pesticide biomarkers. We found no statistically significant associations between urinary concentrations of the biomarkers and consumption of nearly all fruits, vegetables, and legumes. However, women that consumed collards (OR: 5.9, 95% CI: 1.3, 26.7) or spinach (OR: 4.4, 95% CI: 1.1, 17.9) in the past 48-hr were more likely to have detectable urinary concentrations of 2,4-D relative to women that did not consume those foods in the past 48-hr. Associations between urinary concentrations of the pesticides biomarkers and the participants’ consumption of other foods (Brussels sprouts, celery, or wine) in the past 48-hr were not assessed due to the low number of women who reported consuming those items.

Table 4 shows the associations between pest-related issues and urinary concentrations of pesticide biomarkers. There were no statistically significant associations if insects were a common nuisance in the home, pesticides had been applied outside the home by the participant or inside the home by a professional exterminator (regardless of time since last application), or pesticides were currently stored in the home. However, if the participant applied pesticides inside the home (regardless of time since last application), there was significant increase in the odds of having a detectable urinary concentration of trans-DCCA (OR: 4.9, 95% CI: 1.1, 22.1) compared to other women. In addition, insect repellent use since becoming pregnant was positively associated with urinary concentrations of DCBA (β: 1.2, 95% CI: 0.4, 2.0; geometric means of users vs. non-users: 8.3 ng/ml vs. 2.7 ng/ml). Associations between urinary concentrations of the pesticides biomarkers and the participants’ use of pet grooming products, pet flea/tick prevention applications, or pet flea/tick spray in the past 48-hr were not assessed due to the low number of women who reported use of those items.

Finally, the predictors that were statistically significant were simultaneously included in multivariate models for each biomarker that had more than one statistically significant predictor (e.g., 2,4-D in relation to both marital status and spinach or collards consumption). While this served to increase p-values somewhat due to reduced statistical power, effect estimates were similar (results not shown).

To our knowledge, this is the first biomarker study to report on exposures to select pesticides among pregnant women in Puerto Rico, and also the first to report on temporal variability of DCBA, an oxidative metabolite of DEET. We found that 95th percentile urinary concentrations of DEET, 3-PBA, trans-DCCA, and 2,4-D were lower than urinary concentrations in women of reproductive age on the mainland U.S., whereas 95th percentile urinary concentrations of 4-F-3-PBA, cis-DBCA, and 2,4,5-T were similar in both populations. We also found evidence that demographic factors, consumption of leafy greens, and pesticide use are potentially important determinants of exposure to certain pesticides among this group of pregnant women.

Aside from NHANES, there have been only a few studies in other pregnancy cohorts from around the world that have reported on urinary concentrations for some of these analytes [11–13, 30, 32, 44]. Perhaps most relevant to our findings, in a study of pyrethroid exposures among pregnant women from 10 Caribbean countries (the authors did not include the Commonwealth of Puerto Rico), urinary detection frequencies for data across all countries were between 77-100% for cis-DBCA, trans-DCCA, 3-PBA, and 4-F-3-PBA [12]. In our study, detection frequencies were between 0–46.1% for the same group of metabolites. However, the LODs in the Caribbean study were about 1–2 orders of magnitude lower than those in our study, and, as a result, the greater detection frequency in the Caribbean study may not necessarily be equivalent to greater exposure. For example, it is notable that the 95th percentile urinary concentration of 3-PBA was 2.3 ng/ml in our study, compared to 0.54 ng/ml in the Caribbean study, which suggests that exposure to some pyrethroids may have been greater in our population.

The temporal reliability analysis for DCBA suggests that more than one spot sample may be needed to characterize exposure to DEET in pregnant women over the course of pregnancy. Thus, epidemiology studies conducted in pregnant women where the exposure assessment strategy relies on DCBA concentrations from a single spot sample will likely result in a moderate level of exposure misclassification, which, if non-differential, would underestimate (bias toward the null) true associations. In other published studies, temporal variability of urinary concentrations of the pesticides and pesticide metabolites assessed in our study has largely been unexplored, especially over a time period of several months which would be most relevant to pregnancy. ICCs for creatinine-corrected 3-PBA (0.85) and SG-corrected 2,4-D (0.57) associated with spot urine samples collected in Polish adults over seven consecutive days [45] and up to six spot urine samples collected in U.S. adults over a 48-hr period [46], respectively, have been reported.

We found that pregnant women were more likely to have detectable urinary concentrations of DHMB if they were unemployed. One hypothesis explaining this finding is that unemployed women may spend more time outdoors (e.g., gardening, exercising), thus necessitating increased use of insect repellents containing DEET, relative to working women. We had questionnaire information on weekly exercise and chore frequency and duration, which may correlate with outdoor activities, but further investigation of these variables with respect to employment status did not reveal any notable relationships (data not shown).

Participants in our study were also more likely to have detectable urinary concentrations of the herbicide 2,4-D if they were married. One plausible explanation is that married women might be more likely to be home owners relative to unmarried women and, as a result, might be more likely to be involved with home upkeep activities, such as controlling weed growth with herbicides containing 2,4-D. Unfortunately, information on home ownership was not collected so this hypothesis could not be explored further. Positive relationships between being married or living in an owned residence and urinary concentrations of 3-PBA, a non-specific metabolite of several pyrethroid insecticides, among U.S. pregnant women have been reported [30].

In addition, less educated women in our study were more likely to have greater urinary concentrations DCBA and trans-DCCA than other women, a finding supported by other studies for trans-DCCA [13], but not 3-PBA [30]. One hypothesis is that compared with more educated women, less educated women had lower household incomes, which resulted in lifestyle patterns (e.g., insufficient income to purchase organic produce) and/or conditions (e.g., living in poorer communities with decreased access to organic produce) with increased risk of exposure. For example, in a study of U.S. adults [47], consumption of organic produce was shown to increase with both educational attainment and household income levels, which supports our hypothesis.

Furthermore, we found a positive association between participants’ use of pesticides at home and urinary concentration of trans-DCCA. Similar findings have been observed with pyrethroid metabolites in pregnant Chinese women [13], but not in Italian adults [31]. Use of insect repellents since becoming pregnant was also a positive predictor of urinary concentrations of DCBA, which is a novel finding that may validate DCBA as a biomarker of DEET exposure.

Participants in our study were also more likely to have detectable urinary concentrations of 2,4-D if they consumed collards or spinach in the past 48-hr. Consumption of unspecified “leafy greens” [31] or unspecified “dark green vegetables” [10] has also been reported in adults to be positively associated with urinary concentrations of 3-PBA. Consistent with our analysis, no associations between age and urinary concentrations of 3-PBA and trans-DCCA have been reported by others [10, 13, 30, 31].

Several strengths of our study include focusing on an understudied and potentially at-risk population, evaluating biomarkers of understudied exposures, and the collection of repeated data, which allowed for a powerful analysis where each participant served as her own reference in longitudinal models. One primary limitation was the lack of detailed questionnaire information, especially information on how demographic characteristics and pesticide use may lead to increased exposures. Although information of this kind would have assisted with understanding our findings, the increase in detail on the questionnaires would have increased participant burden and may have resulted in reduced participation and study compliance or potentially introduced added recall error. The interpretation of our findings was also further complicated by limited publicly-available information on pesticide use patterns in Puerto Rico. Additional limitations which precluded further quantitative analyses included the modest sample size (N = 54), the small number of participants reporting consumption of certain food items or performing certain pest-related activities, and the relatively low detection frequencies (<50%) of most urinary biomarkers. Finally, it should be noted that the results of our study may not be generalizable to other populations, especially of young children, because interaction with sources of pesticides and/or their metabolism of pesticides may be different from those of pregnant women or other adults.

We showed that pesticide use in a group of Puerto Rican pregnant women is associated with exposure to several pesticides. We also showed that when using DCBA as a specific urinary biomarker of DEET, if possible, more than one sample should be collected over the course of the pregnancy to minimize exposure measurement error. We demonstrated that among these Puerto Rican pregnant women being married, not employed, or less educated was associated with certain biomarkers of exposure to pesticides and should be further investigated to understand what aspects of these demographic characteristics lead to greater exposures. Consumption of collards, spinach, and potentially other leafy greens may also be important determinants of exposure to certain pesticides. Because human exposure to pesticides may be associated with adverse health effects, further research will increase our understanding of the drivers of pesticide exposure among pregnant women.