Abstract
Pyrethroids are synthetic pesticides associated with adverse health effects in humans, but there is much we do not know about the exposures and absorption routes among the general population. This study examines if there is an association between reported household insecticide use and pyrethroid levels (urinary concentrations of 3-phenoxybenzoic acid (3-PBA)) among nationally representative samples of U.S. adults aged 18+ (N = 7448) and children aged 6–17 (N = 2772) from four cross-sectional waves (2007–2014) of the National Health and Nutrition Examination Survey (NHANES). The results indicate significant and positive relationships between self-reported indoor insecticide use in the prior 7 days and pyrethroid levels among adults and children. Indoor insecticide use was associated with about twice the probability of being at the 95th percentile of pyrethroid levels for adults and children in the United States. Direct exposures to household insecticides may be a meaningful and significant mechanism for pyrethroid levels among the U.S. population.
Introduction
Pyrethroids are synthetic pesticides derived from pyrethrins, a chemical naturally occurring in Chrysanthemum cinerariaefolium flowers. Pyrethroids are commonly used in plant protection products (PPPs) to manage pests in agriculture and in biocides, also called insecticides, to control insect vectors and nuisance pests. The use of pyrethroids in PPPs and insecticides has increased as organophosphate use has been found to harm the environment and human health and thus decreased in prevalence over the past decades. 1 Pyrethroids are widely used in insecticide formulations globally, and are recognized by the World Health Organization for applications including insecticidal-treated nets. Household insect control products in the United States such as aerosol insect sprays, lice treatments, and pet products commonly use pyrethroids. 2 Research has found increasing levels of pyrethroids in soil and water and in humans.2–6
Pyrethroids are neurotoxicants, targeting the nervous systems of insects through sodium channels.1,2 Mammals have higher levels of enzymes that metabolize pyrethroids, and thus levels of toxicity are lower for humans. However, there is accumulating evidence that pyrethroid exposure is associated with adverse health effects for humans. Research has shown health consequences for individuals working at or living near agricultural settings. 7 More recently, studies examining residential exposures report associations between pyrethroids and health outcomes, including adverse neurodevelopment, rheumatoid arthritis, diabetes, and all-cause and cardiovascular mortality.8–12
Given the wide use of pyrethroids in household products, prior research documenting increasing levels, and the potential for health impacts, there is a crucial need for observational studies examining exposure mechanisms and who is most at risk of exposure. In addition to agricultural pesticide applications, individuals may be exposed via food residue or residential insecticide use. Residential use of insecticides is positively associated with pyrethroid levels, although this evidence is based on non-probability samples.13,14 One study draws on the nationally representative dataset National Health and Nutrition Examination Study (NHANES), and using a restricted sample from the 1999 to 2002 cycles, it found no statistically significant relationship between insecticide use and pyrethroids. 15 An analysis with more recent waves of NHANES data and a nationally representative sample will provide a new assessment of this relationship at the population level. The aim of this study is therefore to identify if there is an association between reports of indoor insecticide use and detected levels of pyrethroids among representative samples of U.S. adults and children.
Subjects and methods
Data
The data for the analysis were four waves of NHANES data with pyrethroid levels publicly available (2007–2008, 2009–2010, 2011–2012, 2013–2014). The samples for this study included all adults (aged 18+; N = 7448) and children (aged 6–17; N = 2772) who were in the random subsample of participants completing urine collection and with valid laboratory results. Children under the age of 6 were not eligible for the survey question on insecticide use or for the laboratory analysis and were thus excluded from our study.
Measures
The dependent variable was pyrethroid level measured through 3-phenoxybenzoic acid (3-PBA), a metabolite commonly used to measure short-term pyrethroid exposure levels and that is not specific to any particular type of pyrethroid. 6 3-PBA was measured in spot urine samples using solid-phase extraction high-performance liquid chromatography with tandem mass spectrometry, and metabolite concentrations were provided in µg/L. Samples below the detection limit of 0.10 µg/L were imputed with this lower limit divided by the square root of 2. For further information, see the NHANES laboratory manual. 16 This study uses the terms pyrethroid level and 3-PBA interchangeably.
The independent variable of interest was indoor use of insecticides. The question asked respondents “In the past 7 days, were any chemical products used in your home to control fleas, roaches, ants, termites, or other insects?” For children, a proxy provided the answer to this question.
Models controlled for survey year and social and demographic variables that have previously been found to correlate with pyrethroid exposure. Some variables only apply to adults and not children. Demographic controls included age, a measure which combined race/ethnicity and gender into 10 categories, and marital status (adults only). Models also included three measures for socioeconomic status: annual household income-to-needs ratio, home ownership status, and educational attainment (adults only). Two other health measures served as potential confounders: smoking status (adults only) and body-mass index (BMI) constructed from measured height and weight. Lastly, the analysis controlled for urinary creatinine, a measure of globular filtration that is recommended as a covariate in diverse samples to adjust for urine dilution. 17
Statistical analyses
First, descriptive statistics compared the means and medians of pyrethroid levels across insecticide use and potential confounders. Adjusted Wald Tests assessed bivariate significant differences for independent variables across pyrethroid levels and indoor insecticide use. Next, ordinary least squares (OLS) models used a logged value for 3-PBA to account for the non-normal distribution. To ease interpretation and further examine individuals with high concentrations, average marginal effects (AME) models estimated predicted probabilities for high pyrethroid levels, defined as at or above the 95th percentile, across in-home insecticide use. Lastly, we examined whether the relationship between insecticide use and pyrethroid levels differed by race/ethnicity and gender groups, age, survey year, and marital status. Results indicated no significant interaction terms.
All analyses used Stata 19.0. 18 Analyses adjusted for complex sampling design with the svyset command, using NHANES subsample weights, strata, and to ensure the estimates reflected the nationally representative population. To retain the full samples of individuals with valid pyrethroid levels and sampling information, multiple imputation addressed item missingness for both adult and child samples, employing the command “mi impute chained” for a chained equations approach with all independent and dependent variables informing the imputation models. Missingness ranged from 0 to 9.9% across independent variables.
Results
Table 1 presents descriptive statistics, including both mean and median of pyrethroid levels. Among adults and children reporting indoor insecticide use, pyrethroid level means are significantly higher compared to those reporting no such use. Survey year and racial/ethnic and gender groups also indicate significantly different mean pyrethroid levels for adults and children. Significant differences also emerged for age and homeownership among children and marital status and income-to-needs among adults.
Descriptive statistics, United States 2007–2014.
Source: NHANES 2007–2014.
NH = non-Hispanic.
Note. Adjusted for complex sampling design.
NS p ≥ 0.10, +p < 0.10. *p < 0.05. **p < 0.01. ***p < 0.001.
There is also variation of indoor insecticide use across the other independent variables. Those with lower socioeconomic status demonstrate higher indoor use, with the results indicating significant differences across income-to-needs for both children and adults and across education and homeownership for adults. Adults further show differences across age, race/ethnicity, and smoking, whereas children differ in insecticide use across BMI. These differences further motivate the importance of multivariate results that control for confounding.
Table 2 displays coefficients and 95% confidence intervals for pyrethroid levels (measured as logged 3-PBA) among adults and children, with the first model using a base set of controls and the second model using the complete set of confounders. In both models for adults and children, those reporting indoor insecticide use in the prior 7 days have higher pyrethroid levels, with a fairly strong effect size across both models (Adult Model 2: b = 0.44, 95% CI = 0.28, 0.60; Child Model 2: b = 0.51, 95% CI = 0.25, 0.77).
Coefficients and 95% confidence intervals (CIs) from OLS regression for pyrethroid levels (logged urinary concentrations of 3-PBA in μg/L), United States 2007–2014.
Source: NHANES 2007–2014.
NH = non-Hispanic.
Note. Referent in parentheses. Adjusted for complex sampling design. Bold indicates significance p < 0.05.
Other notable results confirm prior research and demonstrate social patterning of pyrethroid exposures. Pyrethroid levels are increasing over time,5,6 and Table 2 shows a modest positive coefficient for survey year among both adults and children. Adults aged 45–64 have slightly higher pyrethroid levels (significant in Model 2 only) compared to those in the oldest group, and younger children have significantly higher levels than older children. Compared to non-Hispanic White men/boys, most other subgroups show lower levels of pyrethroids. Men, women, and boys reporting other race did not differ from non-Hispanic White men/boys. Divorced and cohabiting adults show higher pyrethroid levels compared to their married peers. Net of all other factors, those less educated have lower pyrethroid levels compared to those with college degrees. Adults and children living in households with incomes below the poverty threshold show higher levels of pyrethroids than those in the highest income category. Adults with BMI below 25 demonstrate increased pyrethroid levels compared to with BMI of 30 or greater.
Figure 1 shows predicted probabilities for being in the 95th percentile of 3-PBA levels. The figure illustrates that those who reported indoor use have twice the probability of high pyrethroid levels compared to those who did not. The probability of high pyrethroid levels is about 5% (Pr = 0.05, 95% CI = 0.04, 0.06) for adults and children who reported not using any insecticides indoors, compared to 11% (Pr = 0.11, 95% CI = 0.08, 0.13) for adults and 14% for children (Pr = 0.14, 95% CI = 0.07, 0.21) not reporting such use.
Predicted probabilities for high levels of pyrethroids,a United States 2007–2014.
Discussion
The aim of this study was to assess whether insecticide use was associated with pyrethroid levels among adults and children in the United States. Prior studies have shown associations between insecticide use and pyrethroid levels among convenience samples,13,14 and one analysis 15 using a limited sample from the 1999 to 2002 NHANES found no correlation. The results of this study confirm prior research that indicates pyrethroid levels are rising over time and further show that there is a robust and fairly strong relationship between reporting indoor insecticide use and urinary metabolites of pyrethroids. The similarity of results across adults and children and robustness to confounders bolster this result. A number of factors were related to pyrethroid levels, including age, race/ethnicity and gender, marital status, and income. No significant interactions emerged; the insecticide-pyrethroid relationship does not appear to differ by social characteristics in these data. Overall, these findings suggest that pyrethroid levels in urine may reflect direct exposures to household products and that residential insecticide use may be an important exposure for the general U.S. population.
Future research should consider how exposures from household chemical products can be minimized and how consequences can be mitigated. Technological or other innovations can improve the products themselves. An additional consideration is how the products are used, as proper or improper use of insecticides may affect pyrethroid exposures. Social and environmental factors may shape usage patterns, such as storage, ventilation, frequency, and intensity. For example, one study found that low-income residents are more likely to store insecticides in the kitchen, whereas high-income households store them in the garage. 19 Understanding how individuals use insecticides could inform policies or regulations seeking to reduce exposure through changing how or when products are applied.
High quality housing and proper indoor environments are important for human health. Housing that provides adequate protection from household pests will preclude the need for chemical products. Further research distinguishing the types of pests and insecticides used may shed light on ways to mitigate pyrethroid exposure.
There are important limitations that should be considered. First, we use available data on 3-PBA from 2007 to 2014 that may not reflect today’s exposures. 3-PBA has a short half-life and therefore can only indicate short-, but not long-term exposure. This nonspecific metabolite also cannot differentiate specific pyrethroids. We further cannot distinguish between the insecticide used or the different types of pests targeted. Second, this observational study examining cross-sectional associations is not able to establish causality. To mitigate concerns of confounding and bolster substantive conclusions, the models control for multiple factors, including socioeconomic status, that might affect both insecticide use and pyrethroid levels. However, there may be other unobserved variables that were not included. Lastly, there are other important pyrethroid exposure routes beyond the scope of this study that warrant examination, including food residue and outdoor biocide use. Examining specific pyrethroids or household products will provide useful information. For example, research could examine pet products such as ectoparasiticides which are used to control fleas and ticks. Pyrethroid levels have been detected among dogs and cats, and humans are also exposed to the products in indoor environments. 20
This study demonstrates a relationship between indoor insecticide use and pyrethroid levels among the general U.S. population. Further research is needed on biological and social vulnerabilities to pyrethroids as well as the dose-response relationship. Regulations and recommendations for safe insecticide use can reduce or potentially eliminate human exposure to pyrethroids which may have wide-ranging benefits for the U.S. population.
Footnotes
Acknowledgements
The authors thank Cassidy LeFevre for her assistance with reviewing the literature.
Ethical considerations
The study uses publicly available and deidentified data, so human subjects does not apply.
Author contributions
Elizabeth Lawrence: Conceptualization, Methodology, Formal Analysis, Writing – Original Draft, Writing – Review and Editing, Supervision. Erika Marquez: Conceptualization, Writing – Original Draft, Review and Editing. Amanda Haboush-Deloye: Conceptualization, Methodology, Writing – Original Draft, Review and Editing. Courtney Coughenour: Conceptualization, Methodology, Writing – Original Draft, Review and Editing. Selam Ayele: Conceptualization, Writing – Original Draft.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.