1. Introduction
Respiratory diseases represent a major global health burden, encompassing various types such as infectious, chronic, occupational, and environmental conditions, exerting profound impacts on population health, healthcare systems, and socioeconomics.[1] According to the Global Burden of Disease Study, respiratory diseases account for 12.5% of total global mortality, ranking as the third leading cause of death after cardiovascular diseases and cancer, resulting in approximately 7 million deaths annually.[2] Among these, chronic obstructive pulmonary disease, lower respiratory tract infections, asthma, tuberculosis, and lung cancer constitute the primary contributors to the burden of respiratory diseases, accounting for over 80% of respiratory-related deaths.[3,4] Therefore, this situation has prompted calls from stakeholders to strengthen prevention and control of major risk factors for respiratory diseases.[5] Recent evidence indicates that respiratory pathologies exhibit pan-airway involvement, with disease manifestations spanning the entire lower respiratory tract structure, from the upper respiratory tract (including the nasal cavity and oropharynx) down to the alveolar level.[6] The pathogenesis typically involves persistent inflammatory responses triggered or exacerbated by various exogenous factors, including microbial pathogens (viruses, bacteria, and fungi) and environmental exposures.[7] Notably, airborne particulate matter, aeroallergens, toxic chemicals, tobacco smoke, and various chemical agents (especially agricultural pesticides) have been identified as significant respiratory hazards.[8–11]
Particular attention should be paid to household pesticide use, which has become a significant risk factor due to its widespread application in pest management and gardening. Evidence from the United States indicates that approximately 75% of households use at least one household pesticide.[12] N, N-diethyl-m-toluamide (DEET), a commonly used active ingredient in household insect repellents, is often detected in human urine, plasma samples, and drinking water.[13–15] Similarly, pyrethroid-related compounds are globally used in household insecticide products,[16,17] which raises concerns about potential adverse health effects at low levels of exposure. Although multiple cross-sectional studies have explored the association between pesticide exposure in occupational or agricultural settings and respiratory diseases, research on the health effects of pesticide exposure in residents’ daily lives is relatively scarce.[18] A recent nationwide cohort study conducted in the Netherlands found that populations living in corn-growing areas are more likely to experience deaths related to chronic lower respiratory diseases.[19] To date, no prospective cohort study has directly assessed the relationship between household pesticide exposure and mortality from respiratory diseases in adults.
Therefore, this research aims to explore the link between household pesticide exposure and respiratory disease mortality in a large, forward-looking cohort of adults to address this critical gap. This study employs comprehensive exposure assessments and longitudinal follow-up to investigate potential associations between chronic low-level household pesticide exposure and elevated risks of respiratory disease-related mortality. Utilizing data from the National Health and Nutrition Examination Survey (NHANES), we conducted a rigorous epidemiological analysis to evaluate the relationship between domestic pesticide exposure and respiratory mortality outcomes in adult populations.
2. Methods
2.1. Study design
The NHANES,[20] established in the 1960s, represents a nationally representative surveillance program that systematically collects biological specimens (venous blood and urine) from about 5000 individuals each year through standardized protocols. This comprehensive survey captures a wide range of data encompassing sociodemographic characteristics, clinical parameters, and laboratory test results. The research protocol received thorough ethical evaluation and official approval from the National Center for Health Statistics Institutional Review Board.
Before specimen collection, all participants provided written informed consent following institutional guidelines to maintain strict adherence to ethical standards in human subject research.
Our investigation employed NHANES datasets spanning 2007 to 2014 and then initially screened 23,482 adult subjects. Following the exclusion of 3115 participants lacking complete pesticide exposure records and 28 individuals with incomplete mortality data (with follow-up continuing through December 31, 2019), the final study population included 20,339 subjects for examining the relationship between household pesticide exposure and respiratory-related mortality. Furthermore, we performed specialized analyses on subgroups with available urinary biomarkers: 6565 participants for insect repellent metabolite assessment and 6563 for insecticide metabolite evaluation, allowing for more accurate quantification of pesticide exposure effects.
The complete participant selection methodology is illustrated in Supplementary Figure S1, Supplemental Digital Content 1. All analyses were conducted using identified, publicly accessible data that had undergone proper ethical review and clearance.
2.2. Covariates
The study employed a comprehensive approach to assess participant characteristics across multiple domains, which encompasses demographic factors, socioeconomic indicators, self-reported questionnaire data, and clinical laboratory measurements. The cohort was stratified based on age groups (20–39 years, 40–59 years, and ≥60 years), sex, racial background (African American and Caucasian), nativity status (native-born or immigrant), marital status, educational attainment, and annual household income. Income levels were classified into 3 categories: low (below the 40th percentile), intermediate (40th–80th percentile), and high (above the 80th percentile).
Anthropometric measurements and lifestyle factors were collected through standardized questionnaires, including body mass index (BMI), smoking status (never, former, or current smoker), alcohol intake, exercise habits, employment details, and preexisting medical conditions (such as hypertension, renal failure, and diabetes mellitus). Alcohol consumption was assessed based on self-reported questionnaire data and quantified as the number of drinks per day. BMI was computed using the conventional equation: body weight in kilograms divided by height in meters squared, with classification following World Health Organization criteria: underweight (<18.5 kg/m2), normal range (18.5–24.9 kg/m2), and overweight/obese (≥25 kg/m2).
Physical activity was quantified as weekly minutes of moderate-to-vigorous exercise and grouped into: inadequate (<150 minutes), sufficient (150–300 minutes), and optimal (>300 minutes). Occupational classification utilized self-reported occupational categories provided in NHANES to assess occupational pesticide exposure. Participants were divided into 3 groups for covariate adjustment: Agriculture, Fishing, Forestry: considered as occupationally exposed to pesticides; Mining: considered as potentially exposed to pesticides or with low pesticide exposure, as mining environments may require the use of pesticides for pest and vegetation control; and Other: all other occupations with no known pesticide exposure.
This classification was used as a covariate in adjusted models to account for potential occupation-related factors. Exposure to environmental chemicals was evaluated through quantitative analysis of pesticide and insect repellent metabolites in urine specimens.
2.3. Measurements of exposure and outcomes
2.3.1. Measurements of household pesticide exposure in NHANES
The NHANES employed standardized questionnaires to evaluate residential pesticide exposure among study participants. Specifically, individuals were queried regarding the use of chemical products for pest control (including fleas, cockroaches, ants, termites, and other household pests) during the preceding 7-day period. Based on questionnaire responses, participants were stratified into 2 cohorts: those reporting pesticide exposure (exposed group) and those reporting no exposure (control group).
Biological monitoring was conducted through quantitative analysis of urinary biomarkers. This included measurement of DEET and its primary metabolic derivatives: N, N-diethyl-3-hydroxymethylbenzamide (DHMB) and 3-(diethylcarbamoyl)benzoic acid (DCBA). Furthermore, the study assessed urinary metabolites of pyrethroid insecticides, specifically analyzing 4-fluoro-3-phenoxybenzoic acid, cis-3-(2,2-dibromovinyl)-2,2-dimethylcyclopropanecarboxylic acid, and 3-phenoxybenzoic acid (3-PBA).
These tests were conducted on urine samples collected at baseline from a nationally representative sample of participants aged 6 years and older, as part of the NHANES biomonitoring program by the CDC’s National Center for Environmental Health.
2.3.2. The respiratory disease outcome measurements
In this population-based investigation utilizing NHANES data, mortality outcomes (comprising both survival duration and etiology-specific mortality categories) were derived from the National Death Index (NDI) database, with follow-up records extending through December 2019. The observation period was calculated from the initial assessment date until either the study endpoint (December 31, 2019 for living participants) or the documented mortality date. The NDI system implemented sophisticated record-matching algorithms to ascertain participant survival status.
Meanwhile, it employed standardized International Classification of Diseases, 10th Revision protocols for mortality cause determination. Following the International Classification of Diseases, 10th Revision classification framework established by NDI documentation, respiratory-related fatalities in this analysis encompassed codes J10 - J18 (representing influenza and pneumonia cases) and J40-J47 (denoting chronic obstructive pulmonary conditions).
2.4. Statistical analysis
This study accounted for the complex sampling design of the NHANES database by employing weighted Cox proportional hazards regression models to analyze the association between household pesticide exposure and respiratory disease mortality. The weighting scheme incorporated both primary sampling units (SDMVPSU) and pseudo-strata (SDMVSTRA) to adjust for the stratified cluster sampling design and targeted oversampling of specific populations.
First, descriptive statistical analyses were conducted for demographic, socioeconomic, and behavioral factors, including frequency distributions and proportional comparisons; Fisher’s exact tests and χ2 tests were used to determine the statistical significance of these analyses. Next, a primary model was constructed, followed by covariate selection based on directed acyclic graphs and a threshold criterion of effect estimate changes >1%. By gradually adding covariates based on the results of the directed acyclic graphs and the threshold criterion, the model was transitioned stepwise to a fully adjusted model.
These models evaluated the relationship between self-reported pesticide exposure, log-transformed urinary metabolite levels, and respiratory disease mortality. Exposure-time interaction terms were included, and proportional hazards assumptions were verified via standardized residual analysis (Supplementary Figure S2, Supplemental Digital Content 2 and Table S1, Supplemental Digital Content 5).
Mediation analysis using Cox models assessed whether in vivo pesticide metabolite concentrations mediated the pesticide exposure–respiratory disease mortality relationship. Subgroup analyses were performed to identify potential effect modifications across population strata with different characteristics. Sensitivity analyses evaluated residual confounding by additionally adjusting for comorbidities, occupational history, and education level.
All statistical tests used a 2-sided α level of 0.05. Analyses were conducted using STATA 18.0, SPSS 27, and R 4.4.2. Full methodological details are provided in the Supplementary Methods (Sections 1–3), available in Supplementary Tables S5, S7, and S9, respectively.
3. Results
3.1. Summary for the baseline differences across status of household pesticide exposures
Table 1 displays the demographic, socioeconomic, and behavioral factors of the adults in the study who were exposed to household pesticides. A total of 2276 (11.19%) adults reported exposure to household pesticides. Adults aged 60 or older, those who are Black, have low educational attainment, come from low-income families, are overweight or obese, engage in less than 150 minutes of physical activity per week, or have diabetes and hypertension were exposed to household pesticide.
The baseline characteristic distributions of the total NHANES adults who participated in the 2007 to 2014 waves, and remained in the analyses of self-reported household pesticide exposure with respiratory disease death risks and urinary metabolites with respiratory disease death risks, are summarized in Supplementary Table S2, Supplemental Digital Content 6. Overall, both the self-reported exposure group and the urinary metabolites detection group show strong consistency with the overall population in demographic, socioeconomic, and behavioral factors.
3.2. The results of examining the associations of self-reported household pesticide exposures with respiratory disease mortality risks
In the NHANES participants, exposure to household pesticide was linked to an increased risk of incident respiratory disease death (Supplementary Fig. S3, Supplemental Digital Content 3 and Table 2). In comparison to non-exposed participants, the unadjusted hazard ratio (95% confidence interval [CI]) of incident respiratory disease death was 1.933 (1.923–1.943) for those who reported exposure to household pesticide.
After adjusting for the potential confounders, the analysis showed a hazard ratio (HR) of 1.318 (95% CI: 1.311–1.325).
3.3. The results of examining the associations of urinary metabolites related to household pesticide with respiratory disease mortality risks
The incidence of respiratory disease mortality increased with higher urinary pesticide metabolites. Specifically, the rate increased from 54 cases per 53,857.1 person-years in participants with DEET levels below the limit of detection to just 3 cases per 929.1 person-years in those with DEET levels above the median. For DHMB, the incidence increased from 51 cases per 48,901.7 person-years (below limit of detection) to 5 cases per 3251.1 person-years (above the median) (Supplementary Table S3, Supplemental Digital Content 7).
Substantial evidence suggests that each log10-unit increase in urinary DEET is associated with a higher risk of respiratory disease mortality (unadjusted HR 1.651, 95% CI: 1.631–1.671). Each log10-unit increase in urinary DHMB is also associated with an enhanced risk (unadjusted hazard ratio 1.382, 95% CI 1.372–1.393). After further adjusting for socioeconomic, demographic, and behavioral confounding factors, the outcomes of self-reported exposure and higher urinary DEET levels associated with increased risks of incident respiratory disease death persisted.
There was no evidence indicating the violation of the “proportional hazards” assumption in Cox regression models (Supplementary Table S4, Supplemental Digital Content 8).
Mediation analysis results indicated that household pesticide exposure increased the risks of respiratory disease-related death mainly by raising the levels of urinary DEET and DHMB metabolites (Fig. 1). Individuals exposed to household pesticides had urinary DEET metabolite levels that were 5.54% (95% CI 5.52%–5.56%) higher.
An increase of one log10 unit in urinary DEET was associated with a higher risk (adjusted HR 1.524, 95% CI 1.504–1.544) of subsequent respiratory disease-related death. Of the total association between household pesticide exposure and the risk of death from respiratory diseases, 3.36% was due to increased urinary DEET levels. Likewise, the participants exposed to household pesticides presented a 12.3% (95% CI 6.4%–13.1%) higher urinary DHMB metabolite level.
A per log10-unit increase in urinary DHMB was associated with a greater risk (adjusted HR 1.235, 95% CI 1.225–1.244) of subsequent respiratory disease-related death. A modest 1.50% of the total association between household pesticide exposure and the risk of respiratory disease-related death occurred through increased urinary DHMB levels (Supplementary Fig. S4, Supplemental Digital Content 4).
3.4. The results of stratified and sensitivity analyses
The results of the stratified analyses, along with specific details, are shown in Figure 2. Substantial evidence shows alterations in the effects of the study associations by age (P for interaction <.05). Among NHANES participants over 60 years old, there is an association between exposure to household pesticides and a higher risk of respiratory disease-related mortality.
The adjusted HR is 9.85, with a 95% CI of 1.39 to 69.97. There was no significant evidence of effect modifications across various factors, such as marital status, citizenship, BMI, family income, sex, education, alcohol consumption, smoking, race, physical activity, diabetes, hypertension, renal failure, and occupation (P for interaction >.05). Among NHANES participants, the association between household pesticide exposure and an increased risk of respiratory disease-related mortality remains largely consistent after adjusting for high-risk factors such as diabetes, renal failure, hypertension, occupational pesticide exposure risks, and education (Supplementary Methods 1–3, available in Supplementary Tables S5, S7, and S9, respectively; Figure 3 and Supplementary Table S6, S7, and S10, Supplemental Digital Content 9).
4. Discussion
Using nationwide data from NHANES, we presented evidence from a prospective cohort study showing that exposure to household pesticides is linked to a higher risk of respiratory disease mortality. We confirmed these associations by connecting higher levels of urinary metabolites from household pesticides to an increased risk of respiratory disease mortality. Finally, our analysis revealed that elevated urinary DEET levels accounted for 3.36% of the association between household pesticide exposure and respiratory disease mortality, while elevated urinary DHMB levels accounted for 1.50%.
This association is particularly pronounced among the elderly. We provided strong observational evidence to establish a prospective association between household pesticide exposure and subsequent respiratory disease mortality in adults. Based on the well-designed characteristics, we adjusted for confounders, such as demographic, socioeconomic, and behavioral differences.
We have conducted various sensitivity analyses, and the results indicated robust associations.
4.1. Compared with the existing evidence related to pesticide exposures
Numerous epidemiological investigations have established significant associations between pesticide exposure and various adverse health outcomes, including neoplastic diseases, endocrine dysfunction, hypersensitivity reactions, and pulmonary impairments.[21,22] Both environmental and occupational contact with these chemical agents may contribute to the development of respiratory pathologies. Research conducted in Ethiopia demonstrated that agricultural workers with occupational pesticide exposure exhibit substantially elevated risks of developing pulmonary complications.[18] Furthermore, scientific evidence indicates that certain pesticides may be implicated in respiratory morbidity among occupationally exposed adult populations and pregnant women.[18,22,23] In particular, Liu et al[24] identified a potential link between gestational exposure to a pyrethroid metabolite, 3-PBA, and child cough at age 5 to 6 years, and Islam et al[25] indicated that current pyrethroid exposure may affect children’s respiratory and allergic health. Furthermore, Hu et al[26] found that maternal urinary 3-PBA concentrations were associated with a small decrease in FEV1/FVC in school-aged children.
More studies have implicated exposure to pesticides in the development of wheeze, a precursor to asthma.[27–29] The underlying pathophysiological mechanisms may involve several mechanisms[30]: direct airway irritation; initiation of persistent inflammatory reactions; immunosuppressive effects; and endocrine-disrupting characteristics that may exacerbate respiratory diseases. Nevertheless, the current literature lacks evidence regarding the impact of household pesticide exposure on respiratory-related mortality in adult populations.[31] Our investigation provides novel prospective evidence indicating that both self-reported household pesticide exposure and objective measurements of urinary pesticide metabolites are associated with increased mortality from respiratory diseases in the U.S. adult population. The results consistently supported the associations of household pesticide exposure with elevated subsequent respiratory disease mortality in the NHANES adults.
4.2. The associations of respiratory disease mortality with urinary metabolites related to household pesticide
Our analysis revealed that 3.36% of the link between household pesticide exposure and respiratory disease mortality in NHANES participants was influenced by increased urinary DEET levels from insect repellents. However, the study showed a minimal link between urinary metabolites from household insecticides and respiratory disease mortality. Several factors may explain these differences, primarily the varied application methods and exposure frequencies of insect repellents compared to insecticides.[32] Household insect repellents, which generally have lower toxicity, are usually applied by spraying on clothes or hands.
This method poses a prolonged risk of respiratory inhalation or skin exposure, thereby increasing the risks of significant respiratory disease mortality.[33] In contrast, household insecticides are typically sprayed directly on mosquitoes or used indoors, resulting in shorter periods of respiratory inhalation or skin exposure.[34] Moreover, compared with household insecticides, household insect repellents are more widely used in daily life.[35]
4.3. The risk varies with age
The investigation found that older age is linked to a higher risk of dying from respiratory diseases after exposure to household pesticides. Current epidemiological evidence shows that children, pregnant women, and the elderly are particularly vulnerable to the health effects of pesticides.[31] This increased vulnerability can be explained by specific physiological mechanisms related to age. Children are especially vulnerable because their bodies are still developing and their metabolic pathways are not fully mature, which limits their ability to detoxify harmful substances.[36] Among geriatric populations, age-related physiological decline results in diminished pulmonary function, compromised antioxidant defenses, impaired lung tissue regeneration, and reduced metabolic detoxification capacity.[37] Additionally, studies have shown a strong link between pesticide exposure and chronic respiratory diseases, such as chronic bronchitis and COPD, which primarily affect older adults.[38] These findings underscore the necessity for implementing more stringent precautionary measures regarding pesticide use, particularly in residential environments frequented by elderly individuals.
4.4. The underlying mechanisms related to study association
Our study found that DEET and DHMB partially mediate the link between residential pesticide exposure and respiratory disease mortality. Current research shows that DEET acts as an acetylcholinesterase inhibitor, leading to an accumulation of acetylcholine that disrupts neuromuscular transmission in the muscles responsible for respiration.[39] This effect can cause bronchospasm and excessive mucus secretion, while severe exposure to DEET may lead to paralysis of the respiratory muscles. Additionally, DEET’s known neurotoxic properties negatively impact the pulmonary tissues and airways, leading to respiratory distress, reduced pulmonary function, and chronic respiratory issues.[40]
4.5. Limitations
This study has several limitations. Firstly, we were unable to evaluate the duration and routes of pesticide exposure, as well as the participants’ physiological states during exposure, due to the limited data available in the NHANES database. Second, we assessed participants’ household pesticide exposures using baseline data and we did not account for variations in exposures in the following years.
Adult household pesticide exposures are likely to increase due to global climate change. Thirdly, we assessed internal biomarkers for household pesticides by measuring urinary metabolites of DEET and pyrethroids, which are quickly metabolized in the body and may lead to misclassification of long-term exposure. However, evidence from experimental biomonitoring studies has shown that even for an environmental chemical with a short half-life, frequent exposure can compensate for rapid clearance, resulting in stable urinary concentrations.[41] Finally, due to data availability and some residual confounding factors, such as the lack of data on smoking dose stratification, we did not further classify current smokers by daily smoking amount, which may lead to residual confounding in our effect estimates.
5. Conclusions
Exposure to household pesticides, especially insect repellents, consistently increased the risk of respiratory disease mortality in adults. This association was partially mediated by elevated levels of urinary metabolites of these repellents, indicating that not only the parent pesticides but also their metabolic products may contribute to the increased health risk. The detection of these metabolites in urine serves as a biomarker of internal exposure and underscores the potential toxicity of both the parent compounds and their breakdown products.
This risk is especially notable among older adults.
Author contributions
Data curation: Jiayao Guo, Wenye Zhu.
Formal analysis: Jiayao Guo, Xianmei Li, Yang Yang.
Validation: Jiayao Guo, Guiping Liu, Airong Li.
Visualization: Jiayao Guo, Guiping Liu, Airong Li, Weimin Yang, Yang Yang.
Conceptualization: Min Li, Weimin Yang, Zhuang Luo, Wenye Zhu.
Funding acquisition: Min Li, Wenye Zhu.
Supervision: Min Li, Guiping Liu.
Methodology: Guiping Liu, Zhuang Luo, Wenye Zhu.
Writing – review & editing: Min Li, Xianmei Li, Airong Li, Weimin Yang, Zhuang Luo, Wenye Zhu.
Writing –original draft: Jiayao Guo.