Characterizing everyday exposure to volatile organic compounds and upper respiratory health effects

Abstract

Volatile organic compounds (VOCs) are relatively understudied pollutants compared to particulate matter but are ubiquitous in outdoor and indoor environments. Prior studies on VOCs on sinonasal health have been difficult to generalize due to limitations in their definitions of VOC exposures. We took a novel, holistic approach to characterize a major fraction of daily VOC exposure and its link to sinonasal health. Our study included 981 participants (mean ages = 58.8 years, ranges = 40–80 years) from the 2013–2014 National Health and Nutrition Examination Survey (NHANES) with available urinary VOC metabolites (mVOC) data. We used two unsupervised techniques to address the association between VOC exposure and sinonasal health outcomes. First, we applied factor analysis to identify the sources of urinary mVOCs. Logistic regression was employed to analyze associations between each source and sinonasal health questionnaires. Uniform Manifold Approximation and Projection (UMAP) was used to identify and compare the exposure patterns in subgroups. Factor analysis found three likely sources of exposure: “household goods”, “occupation contaminants”, and “fuel emissions”. Logistic regression showed that exposure to “household goods” was associated with a 22.2% higher likelihood of multiple sinus infections (p = 0.003), while “fuel emissions” were linked to a 16.4% increase (p = 0.026). UMAP identified subgroups where individuals with lower socioeconomic status, coupled with specific behavioral and lifestyle habits, may face an increased risk of VOC exposure and negative sinonasal health outcomes. Our findings provide evidence that usage of certain everyday goods, exposure to fuel emissions, and the characteristics of one’s home and built environment could play a prominent role in an individual’s overall VOC exposure and the manifestation of upper respiratory disease.

Introduction

Air pollution exposure is the single greatest environmental threat to health as identified by the World Health Organization¹. A large body of research on ambient air pollution (AAP) has already investigated the adverse impact of exposure to particulate pollutants on health^2,3. Recently, there has been increased attention on volatile organic compounds (VOCs), which can be found in both outdoor and indoor environments. These ubiquitous organic chemical pollutants are commonly emitted from sources like gas stoves, household products, and vehicle exhaust.

The effects of VOCs on health are becoming increasingly appreciated. To date, there is robust evidence for the impact of single VOC exposures on leukemia⁴, infectious⁵, and neurologic⁶ ailments, low birth rate⁷, hypertension and vascular dysfunction^8,9, liver damage¹⁰, and pulmonary symptoms^8,11. However, despite the increased recognition of VOC exposure’s association with certain diseases, uncertainty remains on the effects of VOCs on the nose and sinuses. The paranasal sinus tissue, being the initial point of contact with polluted air upon inhalation, may be particularly susceptible to the effects of VOCs. Given the complexities and growing concerns surrounding VOC exposure and its understudied effects on paranasal sinus health^12,13, there is a pressing need for a comprehensive study elucidating these associations in the real-world.

Historically, epidemiological studies have focused on single pollutant exposures or through exposure gestalts such as “air pollution” or “smoking” without the resolution to track how each component interacts with an individual’s health and any summative effects of concurrent exposures¹⁴. More realistically, multiple groups of pollutants collectively contribute to health risks, alongside particulate matter, and may cause independent or synergistic effects¹⁵. To overcome these limitations of single-exposure investigation, the concept of the exposome is used in environmental epidemiology to describe how the intricate and interrelated nature of various environmental pollutants contributes to specific health outcomes¹⁶. Therefore, in this study, we examined VOCs exposure patterns in a holistic manner (the interrelatedness of each compound) with a specific focus on how these exposures impact sinonasal health.

As shared by a recent exposome study, we propose that the cumulative impact of source-specific VOCs over time can be useful in understanding the potential health effects associated with VOCs exposure¹⁷. This involves considering different sources of personal exposure to the repertoire of VOCs encountered in households (e.g., furniture varnish, stoves), workplaces (e.g., industrial reagents, building materials), or exhaust. The degree of exposure can be affected by various factors, including the effects of socioeconomic forces that place individuals into a stratus of common behavioral, environmental, and lifestyle patterns such as smoking, diet, living and occupational conditions, which in turn influence an individual’s biology²⁰. Social forces exert their influence on an individual’s life over long spans of time, and consistent with the exposome paradigm, must be considered to understand how one’s biography contributes not only the types and abundance of pollutants encountered in one’s life but also health outcomes. Beyond the consideration of the presence of pollutants in an environment, it is crucial to consider internal exposures of the individuals in this environment and their effects on xenobiotics, commensal microbial communities, and endogenous signaling pathways. This involves the assessment of biomarkers from biological specimens such as blood or urine^18,19.

In this study, we take a unique and multidisciplinary approach to show the clinical importance of VOCs. In the first study on this topic, we characterize VOC exposure patterns in the general population by using unsupervised machine learning methods-clustering analysis. After identifying and clustering similar VOC exposure patterns, we calculate factor scores that represent the extent of individual exposure to each cluster of VOCs. Using these factor scores as predictors, we then apply logistic regression to determine which VOC exposure patterns significantly influence sinonasal outcomes. Additionally, we utilize Uniform Manifold Approximation and Projection (UMAP), which is easy and fast to use, to visualize VOC exposure patterns in a lower-dimensional space. UMAP offers the advantage of preserving both local and global data structures, facilitating the detection of subtle patterns within high-dimensional urinary VOC metabolites (mVOC) data. By integrating these methodologies, we aim to elucidate the associations between VOC exposure patterns and sinonasal outcomes, thereby advancing our understanding of their clinical implications.

Methods

Study design and participants

All data were acquired from the 2013–2014 National Health and Nutrition Examination Survey (NHANES), a nationally representative survey conducted by the Centers for Disease Control and Prevention’s (CDC) National Center for Health Statistics (NCHS)²¹. For our study, we follow the inclusion criteria as illustrated in Fig. 1. This NHANES cycle was selected based on the availability of data on urinary mVOC and taste and smell questionnaires. NHANES participants who are over 40 years old were eligible to participate in taste and smell questionnaires. Thus, participants 40 years and older in the cycles who had complete lab results for urinary mVOCs were considered for inclusion in the study cohort. Participants with missing covariate information were excluded from the analyses. Additionally, to enhance the robustness of our analysis, we removed extreme outlier values in mVOC measurements, defined as those falling above the third quartile plus 30 times the interquartile range (IQR). Individual characteristics extracted as covariates for this study include demographics (age, sex, race/ethnicity, income, education) and serum cotinine levels to measure exposure to tobacco.

[See PDF for image]

Fig. 1

The flow gram of screening out eligible participants from NHANES 2013–2014 survey. Creation of the study cohort from the NHANES dataset.

Urinary laboratory data

The level of mVOC and creatinine from urine samples were collected from the 2013–2014 NHANES laboratory databases. Urine samples were assayed for 26 unique urinary metabolites derived from 18 parent VOCs using ultra-performance liquid chromatography coupled with electrospray tandem mass spectrometry²². Among the mVOCs, six mVOCs were removed due to the reporting of only some similar values, or the number of low detection values exceeded 70% of our study sample. The parent compound and corresponding metabolites are summarized in Table S.1. Urinary creatinine was measured through a quantitative enzymatic assay and spectrophotometry²³. Total serum cotinine was measured through a tandem isotope-dilution high-performance liquid chromatography/atmospheric pressure chemical ionization mass spectrometry²⁴. Absolute urinary concentrations of mVOC analytes (ng/mL) were all normalized to the subject’s urinary creatinine concentration (mg/dL) by dividing the two values (ng VOC/10 ng Cr). These creatinine-corrected mVOC values were used for all subsequent analyses. The mVOC concentrations are summarized in Table S.2.

Assessment of smell and taste outcome from questionnaire data

Chemosensory and sinonasal health outcomes were obtained through responses to the taste and smell questionnaire administered in 2013–2014 NHANES. To identify if individuals have taste or smell dysfunction, we selected 1 question related to smell ability, 1 question related to taste ability, and 2 questions related to sinonasal health. (Table 1) All responses that were “don’t know” or “refused” were removed.

Table 1. Questionnaires used to assess the smell and taste outcome.

Category	Questionnaire
Smell ability	“Had a problem with smell in the past 12 months?”
Taste ability	“Had a problem with taste in the past 12 months?”
Sinonasal health	“Frequent nasal congestion in the past 12 months?” “Ever had 2 or more sinus infections?

Assessment of covariates

Demographic information (age, sex, race, and the ratio of household income to poverty) and serum cotinine levels were collected and set as the covariates for statistical analysis. To determine exposure to tobacco, we used serum cotinine (ng/mL) as a measure of tobacco smoke exposure instead of using the self-reported smoking status. We used 10 (ng/mL) as a threshold to determine study participants as smoke-exposed or not. We chose this threshold based on previous literature, since active smokers have levels higher than 10(ng/mL)²⁶. The distribution of serum cotinine was compared by self-reported smoking status (Fig. 2).

[See PDF for image]

Fig. 2

Comparison between self-reported smoker status and serum cotinine levels. Box plots of log-transformed serum cotinine levels across participants who reported no exposure to tobacco smoke, exposure to secondhand smoke, and active firsthand smoke. The cutoff of 10 ng/mL cotinine was selected as the cutoff for objective smoke exposure²⁵.

Statistical analysis

Factor analysis

Factor analysis is a useful tool to reduce the data by identifying factors responsible for groupings of inter-correlated variables²⁶. Each factor is characterized by variables that strongly correlate with it, as designated by their factor loadings. To group the urinary mVOCs into similar groups, we first used exploratory factor analysis via minimum likelihood with oblique rotation to extract factors. We used oblique rotation because we assumed that the mVOCs could be correlated. 981 participants with 21 normalized creatinine-corrected mVOCs were included in our factor analysis. To choose the optimal number of factors, we used eigenvalues²⁷ and scree plots²⁸ to identify where the sharp drop-off around the eigenvalues equal to 1. From the exploratory factor analysis, we identified 4 factors of mVOCs with factor loadings of 0.30 or more. However, we discovered that the mVOCs in factors 3 and 4 share similar exposure sources and both contained only 2 variables per factor. We combined factor 3 and 4 into a single factor in accordance with best practices to load each factor with at least 3 variables. To estimate the factor scores for each factor and participant, we conducted a confirmatory factor analysis. The factor score, which has a mean 0 and a standard deviation 1, can be interpreted as a value that represents an increased risk of exposure to a group of VOCs, where higher values correspond to greater risk.

Logistic analysis

After the mVOC typology was established, we conducted logistic regression analyses in order to assess the association between sinonasal outcomes and factor scores. The outcomes were yes (= 1) or no (= 0) answers for the four questionnaires related to smell and taste (Table 1). In these logistic regression models, the factor scores obtained from the factor analysis were used as predictor variables. Each factor score represents a degree of exposure based on patterns identified using the mVOC data. The aim of the model is to understand the extent to which each underlying factor was associated with sinonasal outcomes. Odds ratios were obtained from the models, which represent the likelihood of sinonasal outcomes.

UMAP analysis

We were interested in identifying significant sub-populations in the data and further sought to characterize these sub-populations based on the socio-demographic information. We performed UMAP analysis using all measured 21 urinary mVOCs as feature inputs to reduce to lower dimensions that can be used to visualize clustering patterns. To reduce the high-dimensional data onto two dimensions, UMAP used the distance between two vectors by using the metric measures: Euclidean. For the UMAP calculation, we use number of neighboring sample points to 15 and an effective minimum distance between embedded points to 0.01. After that, we used the unsupervised clustering approach to group participants on their similarity in mVOC profiles.

After that, we distinguished socio-demographic features (age, sex, race and ethnicity, education level, poverty income ratio), volatile toxicant questionnaire responses, taste and smell questionnaire responses, and factor scores between clusters. For the questionnaires, we calculated the percentage of individuals who answered ‘yes’ among participants who responded to the questionnaires. Although not all the individuals with mVOC data answered the volatile toxicant questionnaires, we focused on the trend within each group of people who share the similar urinary mVOC patterns. All UMAP analyses were performed using the uwot package²⁹ [ver 0.1.10] in R version [ver 4.2.3]. We used the Rphenoannoy package³⁰ [ver 0.1.0] for the unsupervised clustering of participants.

Results

Demographic characteristics

Table 2 presents the demographic characteristics of the study cohort. Our study included 981 participants aged 40 years and above. The average age of the participants was 58.8 years. Based on the self-reported race and ethnicity categories of NHANES, 45.2% of the participants were non-Hispanic white. 40.3% of the participants showed exposure to smoke, as determined by their serum cotinine levels. The average cotinine level in the serum samples of those exposed to smoke was 271.5 ug/ml. More than half of the participants in our study had completed education above the high school graduate level.

Table 2. Demographic characteristics of the of the NHANES respondents reporting taste and smell questionnaires and have complete information.

Characteristic	Study participant
Age (mean, yrs)	58.8 (40–80)
Male (%)	50.2
Race
White, non-Hispanic (%)	45.2
Others (%)	54.8
Exposure to smoke (cotinine above 10 ug/mL, (%))	40.3%
Average value of Cotinine who expose to smoke (mean, min-max)	271.5 (11.9-1,820)
Education
No high school diploma (%)	25.7
High school graduate (%)	24.0
Above high school graduate (%)	50.35
Poverty income ratio (mean, min-max)	1.9 (1.0–3.0)
Total (N)	981

Factor analysis

We found that factor 1 includes 11 mVOCs, factor 2 has 5 mVOCs, and factor 3 and factor 4 both have 2 mVOCs. Xylene in factor 3 is produced mainly as part of the benzene aromatics and all mVOCs loaded to factor 3 and 4 are major contributors to vehicle exhaust. Therefore, we combined factor 3 and 4 into one factor 3. Table 3 provides the list of the VOCs included in three factors. We also included the parent compound of VOCs used to identify the main exposure sources of these chemicals. Based on their common exposure sources, we named factor 1 as “household goods”, factor 2 as “industrial contaminants”, and factor 3 as “fuel emissions”. Factor 1 consisted of mVOCs from 1,3-butadiene, acrolein, acrylamide, acrylonitrile, vinyl chloride, crotonaldehyde, N N-dimethylformamide, styrene, which can be commonly found in household sources. For example, acrolein and acrylamide are from cooking processes, acrylonitrile can be from acrylic fiber clothing or carpeting, and styrene can be found in many commercial products or operating printers. Factor 2 includes mVOCs found mostly in industrial settings: ethylbenzene, styrene, 1-bromopropane, 1,3-butadiene, propylene oxide, and toluene. Lastly, Factor 3 includes mVOC from xylene, benzene, and cyanide, all of which share a common source of exposure, gasoline and vehicle exhaust.

Table 3. Grouping urinary mVOCs based on factor analysis outcomes.

Factor	VOC metabolite	Parent VOC	Exposure source	Category
1	N-Acetyl-S-(4-hydroxy-2-butenyl)-L-cysteine	1,3-Butadiene	Rubber and Plastics Production/Exhaust Emissions/Smoke	Household good
1	N-Acetyl-S-(2-carboxyethyl)-L-cysteine	Acrolein	Chemical Manufacturing/Smoke/Cooking Processes
1	N-Acetyl-S-(3-hydroxypropyl)-L-cysteine	Acrolein	Chemical Manufacturing/Smoke/Cooking Processes
1	N-Acetyl-S-(2-carbamoylethyl)-L-cysteine	Acrylamide	Chemical Manufacturing/Water contamination/Cooking Processes
1	N-Acetyl-S-(2-carbamoyl-2-hydroxyethyl)-L-cysteine	Acrylamide	Chemical Manufacturing/Water contamination/Cooking Processes
1	N-Acetyl-S-(2-cyanoethyl)-L-cysteine	Acrylonitrile	Rubber and Plastics Production
1	N-Acetyl-S-(2-hydroxyethyl)-L-cysteine	Acrylonitrile, vinyl chloride, ethylene oxide	Rubber and Plastics Production/Healthcare and Pharmaceuticals
1	N-Acetyl-S-(3-hydroxypropyl-1-methyl)-L-cysteine	Crotonaldehyde	Chemical Manufacturing//Exhaust Emissions/Smoke
1	N-Acetyl-S-(N-methylcarbamoyl)-L-cysteine	N, N- Dimethylformamide	Chemical Manufacturing
1	Mandelic acid	Styrene	Rubber and Plastics Production/Exhaust Emissions/Building Materials
1	N-Acetyl-S-(phenyl-2-hydroxyethyl)-L-cysteine	Styrene	Rubber and Plastics Production/Exhaust Emissions/Building Materials
2	Phenylglyoxylic acid	Ethylbenzene, styrene	Chemical Manufacturing/ Fuel Additives and Evaporation/Paints and Varnishes	Industrial contaminant
2	N-Acetyl-S-(n-propyl)-L-cysteine	1-Bromopropane	Chemical Manufacturing/Fuel Additives and Evaporation
2	N-Acetyl-S-(3,4-dihydroxybutyl)-L-cysteine	1,3-Butadiene	Rubber and Plastics Production/Exhaust Emissions/Smoke
2	N-Acetyl-S-(2-hydroxypropyl)-L-cysteine	Propylene oxide	Chemical Manufacturing
2	N-Acetyl-S-(benzyl)-L-cysteine	Toluene	Exhaust Emissions/Paints and Varnishes/Fuel Additives and Evaporation
3	2-Methylhippuric acid	Xylene	Exhaust Emissions/Fuel Additives and Evaporation/Paints and Varnishes	Fuel emission
3	3- and 4-Methylhippuric acid	Xylene	Exhaust Emissions/Fuel Additives and Evaporation/Paints and Varnishes
3	N-Acetyl-S-(phenyl)-L-cysteine	Benzene	Chemical Manufacturing/Exhaust Emissions:/Paints and Varnishes
3	2-Aminothiazoline-4-carboxylic acid	Cyanide	Chemical Manufacturing/Exhaust Emissions/Smoke

Table presents the list of mVOCs loaded to each factor, parent VOC of mVOC, known exposure sources of VOC, and category based on the common exposure source⁷⁷.

With newly defined factors, we calculated the factor scores using the confirmatory factor analysis. Using a logistic regression analysis, we tested whether the factor scores are associated with sinonasal health outcomes, where 1-unit increases of standard deviation in factor scores are corelated with proportional increases in the likelihood of reporting particular symptoms (Table 4). With four outcomes of interest and three VOC groups, we conducted a total of 12 analyses. Our findings showed that higher factor scores of “fuel emissions” and “industrial contaminants” made participants 19% (p = 0.089) and 13% (p = 0.084) more likely to report frequent nasal congestion in the past 12 months, respectively, but this did not reach statistical significance. All exposure sources except for “industrial contaminants” were significantly associated with participants reporting multiple sinus infections. In particular, exposure to household environments increased the likelihood of multiple sinus infections by 22.2% (p = 0.003) and exposure to fuel source emissions increased it by 16.4% (p = 0.026). After controlling tobacco exposure, we found ‘household goods’ to be the only significant predictor of risk of recurrent sinus infections.

Table 4. Factor scores as predictors of sinonasal health outcomes.

Exposure sources	had problem with smell in past 12 months	had problem with taste in past 12 months	frequent nasal congestion in past 12 months	ever had two or more sinus infections
Household	1.160	1.034	1.090	1.222**
	(0.113)	(0.154)	(0.075)	(0.082)
Occupation	1.158	1.027	1.131†	1.146†
	(0.113)	(0.164)	(0.083)	(0.083)
Fuel emissions	1.191†	0.982	1.102	1.164*
	(0.122)	(0.161)	(0.078)	(0.079)
Adjust for tobacco exposure
Household	1.128	0.958	1.078	1.232*
	(0.133)	(0.187)	(0.089)	(0.102)
Occupation	1.130	0.980	1.121	1.114
	(0.120)	(0.184)	(0.088)	(0.086)
Fuel emissions	1.161	0.9110	1.089	1.138†
	(0.134)	(0.177)	(0.087)	(0.087)
N	981	981	981	981

Note: ^†p < 0.1; ^*p < 0.05; ^**p < 0.01.

Two logistic regression models of selected health outcomes from each factor either adjusting for tobaccos smoke exposure or not. Values are reported as odds ratios with 95% confidence intervals in ().

UMAP – characterization of clusters

We characterize individuals into four subgroups through unsupervised clustering of 21 mVOC values, where each group consists of participants with similar urinary mVOC profiles (Fig. 3.A). The UMAP Fig. 3.B presents the heatmap of the average z-scores for the exposure sources we identified through factor analysis (household goods, industrial contaminants, and fuel emissions) and demographic characteristics, in each group (More details can be found in Table S.3). 981 individuals were divided into group 1 (N = 232), group 2 (N = 234), group 3 (N = 196), and group 4 (N = 319). The mean age for group 3 was 60.52 (SD = 10.69) and for group 1 was 59.28 (SD = 12.38), whereas groups 2 and 4 had lower mean ages of 57.92 (SD = 12.27) and 55.83 (SD = 10.09), respectively. We further characterized each group below, including demographics, SES, and lifestyle patterns.

[See PDF for image]

Fig. 3

UMAP of urinary mVOCs identified four group of participants, which shows distinct characteristics. (A) Uniform Manifold Approximation and Projection (UMAP) clustering of the 21 urinary mVOCs data reveals that 981 individuals could be grouped into four clusters, each colored according to its respective group. Axes is the reduced dimensions of the data. (B) The heatmap illustrates that the groups identified from the UMAP analysis exhibit distinct demographic characteristics, lifestyle behaviors, sinonasal outcomes, smoke exposure and factor scores derived from factor analysis across the groups. Demographic characteristics include the poverty income ratio (PIR), education level (no high school diploma, high school diploma, above high school diploma), age, sex, and race (Hispanic, non-Hispanic white, non-Hispanic Black, Asian, other race). Lifestyle characteristics include the percentage of individuals who store paint or fuels inside home, have an attached garage, pump gas into car in the last 48 h, and cook with natural gas. Serum cotinine levels greater than 10 ng/mL indicate smoke exposure. Health outcomes include experiencing problems with smell in the past 12 months, having had two or more sinus infections, experiencing problems with taste in the past 12 months, frequent nasal congestion in the past month, and experiencing taste/smell problems affecting daily life. (C) The heatmap of B but excluding UMAP cluster 4 because of its dissimilarity to the other clusters.

• Group 1 is likely to include individuals who are non-Hispanic white (49.1%), male (58.6%), likely to store paints and fuels at home (49.0%), have an attached garage (57.8%), pump with gas (39.9%), and have higher fuel emission factor scores compared to other groups. However, people in group 1 are less likely to have sinonasal problems compared to other groups.

• Group 2 is more likely to include individuals who are Asian (13.7%) and non-Hispanic black (27.8%), cook with natural gas (29.7%).

• Group 3 was more likely to compromise of older individuals (Group3 vs. Group1 p-value: 0.074; Group3 vs. Group2: 7.55e-05; Group3 vs. Group4: 1.054e-08), as well as a higher percentage of females (58.7%), Hispanic (18.4%), and those at a higher risk of having a problem with smell past 12 months (9.7%) and frequent nasal congestions (32.1%) compared to other groups. These individuals are more likely to be exposed to VOC sources compared to those in Group 1 and Group 2.

• Group 4 is likely to include individuals who are non-Hispanic (82.4%), low income (Group4 vs. Group1 p-value: 4,47e-0.5, Group4 vs. Group2 p-value: 7.62e-0.5, Group4 vs. Group3 p-value: 1.49–0.8), without high school diploma (31.0%), more likely to be exposed to smoke (86.2%), higher factor scores, higher rate of reporting recurrent sinus infections (37.6%), nasal congestion (32.0%), and problems with taste (5.3%) compared to other groups.

Notable relationships emerge from this analysis. Firstly, individuals who were more exposed to smoke were likely to have lower socioeconomic status and likely to be exposed to multiple exposure sources, leading to increased risk of sinonasal issues compared to other groups. Secondly, when we exclude group 4 due to its highly distinct characteristics, Fig. 3.C. shows that group 3 has high percentage of Hispanic older females who were likely to have higher “household good” and “industrial contaminant” scores compared to group 1 and 2, and to have an elevated risk of olfactory dysfunction and nasal congestion. Next, individuals who cooked with a gas stove were more likely to encounter taste and smell issues in their daily lives. Lastly, we note a group of individuals who had high “fuel emissions” factor scores, who were likely to pump with gas or store paints or fuels at home, and which also had a high chance of answering “yes” to questionnaire asking whether have two or more sinus infections.

Discussion

Our study investigated the association between sinonasal health outcomes and VOC exposure by using the urinary mVOC levels from NHANES 2013–2014. Unlike other studies on this topic, we took a holistic approach to characterize VOC exposure by sorting our 27 measured mVOCs into groups of probable sources. We believe this is the most relevant way to convey VOC exposure to patients, clinicians, and policymakers. By identifying specific locations, behaviors, and occupations that are present in an individual’s real life, we avoid convoluted lists of compounds and correlations that likely may not capture the diversity of pollutants different members of a population encounter nor the spatiotemporal fluctuations that occur as individuals move throughout the built and natural environment³¹.

To our knowledge, this is the first study to examine the effect of VOC exposure on sinonasal health and inflammation, which has been understudied despite the growing literature on VOCs on human health^8,11,32,33. Studies using single VOC exposures have not shown any association with acute nasal inflammation or nasal symptoms, which may overlook potential synergistic or antagonistic effects between VOCs^12,34. The few studies that explore the effects of acute VOC exposure on nasal health outcomes fail to reflect the complexity of periodic, real-world VOC exposure, resulting in inconsistent findings. One study found there were no significant health effects of a 2-hour exposure to a diverse VOC mixture with follow-up measures up to 85 min post-exposure; however, this limited time frame does not represent real-life exposure^12,34. A similar study, however, reported a dose-dependent increase in respiratory and non-respiratory symptoms³⁵. In another single-home VOC questionnaire study, there were significant associations between VOC levels and complaints of mucus symptoms of nasal or airway irritation and coughing¹³.

Many of these studies involving VOCs, however, share the same limitation: a lack of generalizability due to each study’s definition of a VOC exposure falling into two extremes of the spectrum of specificity. Studies either narrowly focused on specific compounds from the background of VOCs and environmental factors^{11,36, 37–38} or use total VOCs as a monolith^{35,39, 40–41}. One recent study, however, stands out in its use of unsupervised machine learning to cluster blood VOC data from NHANES into three VOC categories, which is a similar approach to what we took in our study³². Other notable studies have grouped benzene, toluene, ethylbenzene, and xylene (BTEX) because of their frequent cooccurrence from common pollutant sources: vehicle exhaust and secondhand smoke. Consequently, BTEX studies show better consistency and replicability⁴², confirming the validity of our source-based approach and the need for the current study to refine this approach further by including as many other VOCs as possible. Through urinary metabolomics, we were able to obtain data on the sum of internal chemical exposures from many VOCs, a key component that makes the characterization of the totality of VOC exposure possible⁴³. Our choice of urinary mVOCs was based on previous metabolomic studies from NHANES biofluid data^44,45.

Our factor analysis distinguished urine metabolites into three groups that share similar exposure sources and for which we were able to characterize and name using the Agency for Toxic Substances and Disease Registry’s Toxicological Profiles⁴⁶: `household good`, `industrial contaminant`, and `fuel emissions`. Although nearly all the VOCs studied are known to be found in tobacco smoke, we recognize that these same pollutants can be found in other environments and have thus parsed them into more refined groups accordingly. For example, acrylamide is often found in foods and thus classified as a `household good`⁴⁷. Propylene oxide is primarily encountered in high concentrations in manufacturing facilities and is classified as an `industrial contaminant`⁴⁸. Automotive exhaust and industrial emissions account for a significant part of national exposures to benzene, which we classified as `fuel emissions`. However, several VOCs are known to share more than one source. For example, both 1,3-butadiene and toluene are known components of motor vehicle exhaust. However, our results suggest that they are more closely related to the VOCs in ‘industrial contaminant’ than ‘fuel emissions’. While we carefully categorized each factor based on the function, effects, and potential common sources of the VOCs, the interpretation should be approached with caution due to our lack of occupational information. Future research is needed to delineate exposure sources more clearly by incorporating more detailed information on occupation and lifestyle factors.

Logistic regression analysis revealed exposure to ‘household good’ and ‘fuel emission’ sources are significantly associated with recurrent sinus infections. Our study is the first to report this association between household goods and fuel emissions to increased sinus infections. However, one study reported longer periods of work in gas station workers correlated with longer nasal mucociliary transport time, which may indirectly contribute to increased sinus infections⁴⁹.

Substantial evidence supports the hypothesis that chemical and biological pollutants interact^50,51. Previous studies showed that exposure to VOCs increased the risk of infection, not limited to respiratory infections^{8,52, 53, 54, 55–56}. The increased susceptibility to infections due to VOC exposure is possibly mediated through the disruption of host immunity such as depletions in regulatory T cells⁵⁷, altered monocyte responses⁵⁸, NK-mediated cytotoxicity⁵⁹, and increases in allergic⁶⁰ and non-Th2 inflammation⁶¹. This inflammatory stress may damage the epithelial barrier allowing for pathogens to more easily infect target tissues including the sinuses^{62, 63–64}. In addition, benzene has been shown to be immunosuppressive and to induce oxidative stress^65,66.

Our UMAP analysis grouped individuals sharing similar mVOC profiles, which helped us to identify and characterize which subpopulations are more vulnerable to VOC exposure. Emerging research has found differences in associations across gender and racial groups for VOC exposure and liver injury markers^67,68. From our results from the factor and UMAP analyses, we also found sociodemographic characteristics and lifestyles variables can be closely related to the VOC exposure. For example, we found a group of people (group 1), which was more likely exposed to fuel emissions, showed a high proportion of individuals who stored paint, fuel inside their house, and recently pumped gas in their car, but were more affluent and less likely to report sinonasal symptoms. In addition, we found a group (group 3), comprising individuals who are older, female, Hispanic, with many living in poverty, tend to exhibit higher `household good` and `industrial contaminant` factor scores, and are likely to report frequent nasal congestion and changes to their smell, aligning with findings from prior studies^69,70. Group 2 contains the highest proportion of Asian and Black participants, gas stove users, and self-reported chemosensory problems that interfere with their daily lives. Group 4 is characterized by having the lowest overall education level and income. These people on average suffer from the greatest pollutant and health burden compared to other groups, having higher rates of chemosensory and sinonasal symptoms^71,72.

Our UMAP findings showed significant differences in mVOC patterns across racial, ethnic, and socioeconomic status (SES). From previous epidemiological research on AAP exposures has demonstrated that individuals from lower SES or specific racial and ethnic groups face a higher risk of adverse respiratory health outcomes, contributing to health disparities⁷³. Recent studies found that areas with higher-than-average white and Native American populations are exposed to lower average PM_2.5 levels than regions with larger populations of Black, Asian, and Hispanic or Latino individuals⁷⁴. This disparity can often result in increasing risks of mortality, morbidity, and healthcare utilization among these groups. While our UMAP findings suggest that mVOC patterns may vary by race and/or SES, we did not find significantly different risks for sinonasal health outcomes. Moreover, due to the limitation of UMAP, the directionality of these associations cannot be fully determined.

Overall, our study found that VOCs from ‘household goods’ and ‘fuel emissions’ can be a potential risk factor related to sinonasal health outcomes compared to other frequently encountered VOC sources through a novel approach. A concerning theme of the unrecognized hazards existing within buildings we live, work, and play permeates through our findings. Greater consideration, therefore, must be given to the way buildings are designed and used. In the US, indoor air quality has been relatively neglected relative to ambient air quality. Guidelines on indoor air quality have been left up to individual parties, leading to relatively little oversight or concern for indoor air.

Our findings provide evidence to show that the toxic compounds that lead to adverse health effects are not unique to tobacco smoke but also correlated with individual’s home environment or their behaviors. While smoking certainly plays a role in the adverse health effects described in this study, one’s occupation, one’s usage of daily household products, the characteristics of the built environment, and interaction with fuel and combusting materials, including natural gas⁷⁵, could also play a prominent role in the overall dose of VOC exposure and resulting physiological stress.

However, this study has its limitations, including the lack of environmental VOC data. With each VOC having multiple possible sources, the ambiguity about which source contributes the greatest to the internal exposures measured in this study makes it challenging to confirm significant routes of exposure. In addition, the half-life of VOCs circulating within our bodies is relatively short that the VOC levels measured at the time of the examination may not fully capture the lifetime exposure⁷⁶. Moreover, due to the limited availability of data, our analysis was not able to adjust for the different metabolizing times of the different VOCs which might potentially impact the association between urinary mVOCs and sinonasal health outcomes. Another limitation is due to data availability, we were unable to control for other confounding factors such as housing conditions and additional air pollutants, which could contribute to the negative sinonasal symptoms among individuals, particularly those with lower SES or other vulnerable groups. Our clinical outcome measures may be associated with both chronic exposures to VOCs and to short-term exposures, only the latter of which are captured in mVOC data. We included lifestyle information known to increase exposures to VOC (e.g. storing paints and fuels inside the home) to supplement the relatively short-term exposure data found in mVOCs. However, we recognize that we may not be able to distinguish the effects of both short- and long-term exposures on our outcomes. Outcomes were also answered through questionnaires subject to recall bias. In addition, NHANES was designed as a cross-sectional survey, so a longitudinal investigation on the effect of VOCs and the factors defined in this study on health is difficult.

While studying how individual pollutants may affect health may provide insight into the many ways the world around us may be contributing to the development of disease, we want to highlight that often it is also useful to consider multiple pollutants together, since each source of exposure that an individual encounters includes more than one pollutant. In addition, by tracking exposure sources, we can better design public health policy and targeted interventions that effectively reduce pollution before pollutants are introduced into the environment.

Author contributions

YJ and SC wrote the main manuscript text. YJ prepared figures and tables. KM, JS, MR, and TD participated in the preparation, analysis, and presentation of the dataset. CLN and JGA conceived the study design and selected appropriate methods. TSA, SEY, RK, MM, JZ, MB, ZA contributed to sections of the manuscript draft and provided revisions. AM, RR, RB, KN, SEL contributed to the interpretation of results and revision of the final manuscript. All authors reviewed and approved the manuscript.

Data availability

Data sets can be obtained from the CDC NHANES website. https://wwwn.cdc.gov/nchs/nhanes/continuousnhanes/default.aspx? BeginYear=2013 Demographics (DEMO_H) Examination: Taste and Smell (CSQ_H) Laboratory: Cotinine and Hydroxycotinine – Serum, Volatile Organic Compound (VOC) Metabolites – Urine (UVOC_H, UVOCS_H, COT_H) Questionnaire: Taste and Smell, Smoking - Cigarette Use (CSQ_H, SMQ_H) The code generated during the current study is available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

8. References

1. World Health Organization. WHO Global Air Quality Guidelines. Particulate Matter (PM2.5 and PM10), Ozone, Nitrogen Dioxide, Sulfur Dioxide and Carbon Monoxide. (2021).

2. Sierra-Vargas, MP; Teran, LM. Air pollution: impact and prevention. Respirol. Carlton Vic.; 2012; 17, pp. 1031-1038. [DOI: https://dx.doi.org/10.1111/j.1440-1843.2012.02213.x]

3. Liu, C et al. Ambient Particulate Air Pollution and Daily Mortality in 652 cities. N Engl. J. Med.; 2019; 381, pp. 705-715.1:CAS:528:DC%2BC1MXhs1ykurvL [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31433918][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7891185][DOI: https://dx.doi.org/10.1056/NEJMoa1817364]

4. Costantini, AS et al. Risk of leukemia and multiple myeloma associated with exposure to benzene and other organic solvents: evidence from the Italian Multicenter Case-control study. Am. J. Ind. Med.; 2008; 51, pp. 803-811. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/18651579][DOI: https://dx.doi.org/10.1002/ajim.20592]

5. Madani, NA; Jones, LE; Carpenter, DO. Different volatile organic compounds in local point source air pollution pose distinctive elevated risks for respiratory disease-associated emergency room visits. Chemosphere; 2023; 344, 140403.1:CAS:528:DC%2BB3sXitFektLbE [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37832881][DOI: https://dx.doi.org/10.1016/j.chemosphere.2023.140403]

6. Hester, SD; Johnstone, AFM; Boyes, WK; Bushnell, PJ; Shafer, TJ. Acute toluene exposure alters expression of genes in the central nervous system associated with synaptic structure and function. Neurotoxicol Teratol; 2011; 33, pp. 521-529.1:CAS:528:DC%2BC3MXht1ejsbfM [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21827849][DOI: https://dx.doi.org/10.1016/j.ntt.2011.07.008]

7. Chen, D et al. Exposure to benzene, occupational stress, and reduced birth weight. Occup. Environ. Med.; 2000; 57, pp. 661-667.1:CAS:528:DC%2BD3cXnt1Kgsb4%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10984337][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1739869][DOI: https://dx.doi.org/10.1136/oem.57.10.661]

8. Liu, N et al. Health effects of exposure to indoor volatile organic compounds from 1980 to 2017: a systematic review and meta-analysis. Indoor Air; 2022; 32, e13038.1:CAS:528:DC%2BB38XhsVahsr7O [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35622720][DOI: https://dx.doi.org/10.1111/ina.13038]

9. McGraw, KE et al. Exposure to volatile organic compounds - acrolein, 1,3-butadiene, and crotonaldehyde - is associated with vascular dysfunction. Environ. Res.; 2021; 196, 110903.1:CAS:528:DC%2BB3MXmtFais78%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33636185][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8119348][DOI: https://dx.doi.org/10.1016/j.envres.2021.110903]

10. Liu, W et al. Single-chemical and mixture effects of multiple volatile organic compounds exposure on liver injury and risk of non-alcoholic fatty liver disease in a representative general adult population. Chemosphere; 2023; 339, 139753.1:CAS:528:DC%2BB3sXhs1Gqs77F [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37553041][DOI: https://dx.doi.org/10.1016/j.chemosphere.2023.139753]

11. Lv, JJ et al. Assessing volatile organic compounds exposure and chronic obstructive pulmonary diseases in US adults. Front. Public. Health; 2023; 11, 1210136. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37475768][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10354632][DOI: https://dx.doi.org/10.3389/fpubh.2023.1210136]

12. Laumbach, RJ et al. Nasal effects of a mixture of volatile organic compounds and their ozone oxidation products. J. Occup. Environ. Med.; 2005; 47, pp. 1182-1189.1:CAS:528:DC%2BD2MXhtFOmsLzM [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16282880][DOI: https://dx.doi.org/10.1097/01.jom.0000183338.95778.f0]

13. Araki, A et al. Relationship between selected indoor volatile organic compounds, so-called microbial VOC, and the prevalence of mucous membrane symptoms in single family homes. Sci. Total Environ.; 2010; 408, pp. 2208-2215.2010ScTEn.408.2208A1:CAS:528:DC%2BC3cXjtFyjurY%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20188399][DOI: https://dx.doi.org/10.1016/j.scitotenv.2010.02.012]

14. Vineis, P et al. What is new in the exposome? Environ. Int; 2020; 143, 105887.

15. Hart, JE et al. Long-term ambient multipollutant exposures and mortality. Am. J. Respir Crit. Care Med.; 2011; 183, pp. 73-78.1:CAS:528:DC%2BC3MXhslOnsLk%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20656944][DOI: https://dx.doi.org/10.1164/rccm.200912-1903OC]

16. Vermeulen, R; Schymanski, EL; Barabási, AL; Miller, GW. The exposome and health: where chemistry meets biology. Science; 2020; 367, pp. 392-396.2020Sci..367.392V1:CAS:528:DC%2BB3cXhvValurg%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31974245][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7227413][DOI: https://dx.doi.org/10.1126/science.aay3164]

17. Sarigiannis, DA. Assessing the impact of hazardous waste on children’s health: the exposome paradigm. Environ. Res.; 2017; 158, pp. 531-541.1:CAS:528:DC%2BC2sXht1aku7zJ [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28711809][DOI: https://dx.doi.org/10.1016/j.envres.2017.06.031]

18. Rappaport, SM. Redefining environmental exposure for disease etiology. Npj Syst. Biol. Appl.; 2018; 4, pp. 1-6. [DOI: https://dx.doi.org/10.1038/s41540-018-0065-0]

19. Rappaport, SM; Smith, MT. Environ. Disease Risks Sci.; 2010; 330, pp. 460-461.1:CAS:528:DC%2BC3cXhsVaqs7zM

20. Vineis, P; Barouki, R. The exposome as the science of social-to-biological transitions. Environ. Int.; 2022; 165, 107312.1:CAS:528:DC%2BB38XhsVGhurvF [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35635963][DOI: https://dx.doi.org/10.1016/j.envint.2022.107312]

21. Centers for Disease Control and Prevention (CDC) & National Center for Health Statistics (NCHS). National Health and Nutrition Examination Survey Data (U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, 2013).

22. Victor De Jesus & & James, L. Pirkle. Laboratory Procedure Manual: Volatile Organic Compounds (VOCs) Metabolites (Centers for Disease Control and Prevention, 2018).

23. John Eckfeldt. Laboratory Procedure Manual: Creatinine (University of Minnesota, 2015).

24. Connie, S. & James, L. Pirkle. Laboratory Procedure Manual: Cotinine and Hydroxycotinine (Total)vol. 2017 (Centers for Disease Control and Prevention, 2016).

25. Pirkle, JL et al. Exposure of the US population to environmental tobacco smoke: the Third National Health and Nutrition Examination Survey, 1988 to 1991. JAMA; 1996; 275, pp. 1233-1240.1:STN:280:DyaK287os1Cmsg%3D%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/8601954][DOI: https://dx.doi.org/10.1001/jama.1996.03530400021033]

26. Stevens, J. P. Applied Multivariate Statistics for the Social Sciences, Fifth Edition (Routledge, 2012).

27. Kaiser, HF. The Application of Electronic Computers to Factor Analysis. Educ. Psychol. Meas.; 1960; 20, pp. 141-151. [DOI: https://dx.doi.org/10.1177/001316446002000116]

28. Cattell, RB. The Scree Test for the number of factors. Multivar. Behav. Res.; 1966; 1, pp. 245-276.1:STN:280:DC%2BC28nmsVOlsA%3D%3D [DOI: https://dx.doi.org/10.1207/s15327906mbr0102_10]

29. Melville, J., Lun, A., Djekidel, M. N., Hao, Y. & Eddelbuettel, D. The Uniform Manifold Approximation and Projection (UMAP) Method for Dimensionality Reduction. (2023).

30. stuchly stuchly/Rphenoannoy. (2023).

31. Delgado-Saborit, JM; Aquilina, NJ; Meddings, C; Baker, S; Harrison, RM. Relationship of personal exposure to volatile organic compounds to home, work and fixed site outdoor concentrations. Sci. Total Environ.; 2011; 409, pp. 478-488.2011ScTEn.409.478D1:CAS:528:DC%2BC3cXhsFaru7nM [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21112612][DOI: https://dx.doi.org/10.1016/j.scitotenv.2010.10.014]

32. Wang, Y et al. Associations between the compositional patterns of blood volatile organic compounds and chronic respiratory diseases and ages at onset in NHANES 2003–2012. Chemosphere; 2023; 327, 138425.1:CAS:528:DC%2BB3sXmtlemtrc%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36931402][DOI: https://dx.doi.org/10.1016/j.chemosphere.2023.138425]

33. Mendy, A et al. Urinary volatile organic compound metabolites and reduced lung function in U.S. adults. Respir Med.; 2022; 205, 107053. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36399896][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9869342][DOI: https://dx.doi.org/10.1016/j.rmed.2022.107053]

34. Norbäck, D; Hashim, JH; Hashim, Z; Ali, F. Volatile organic compounds (VOC), formaldehyde and nitrogen dioxide (NO2) in schools in Johor Bahru, Malaysia: associations with rhinitis, ocular, throat and dermal symptoms, headache and fatigue. Sci. Total Environ.; 2017; 592, pp. 153-160.2017ScTEn.592.153N [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28319702][DOI: https://dx.doi.org/10.1016/j.scitotenv.2017.02.215]

35. Pappas, GP et al. The respiratory effects of volatile organic compounds. Int. J. Occup. Environ. Health; 2000; 6, pp. 1-8.1:CAS:528:DC%2BD3cXitVKqt7s%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10637531][DOI: https://dx.doi.org/10.1179/oeh.2000.6.1.1]

36. Saijo, Y et al. Symptoms in relation to chemicals and dampness in newly built dwellings. Int. Arch. Occup. Environ. Health; 2004; 77, pp. 461-470.1:CAS:528:DC%2BD2cXhtVSitLrM [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15558298][DOI: https://dx.doi.org/10.1007/s00420-004-0535-0]

37. Wieslander, G; Norbäck, D; Nordström, K; Wålinder, R; Venge, P. Nasal and ocular symptoms, tear film stability and biomarkers in nasal lavage, in relation to building-dampness and building design in hospitals. Int. Arch. Occup. Environ. Health; 1999; 72, pp. 451-461.1:CAS:528:DyaK1MXmvVKrtbw%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10541910][DOI: https://dx.doi.org/10.1007/s004200050398]

38. Wolkoff, P; Clausen, PA; Larsen, ST; Hammer, M; Nielsen, GD. Airway effects of repeated exposures to ozone-initiated limonene oxidation products as model of indoor air mixtures. Toxicol. Lett.; 2012; 209, pp. 166-172.1:CAS:528:DC%2BC38XhvVGht7k%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22212438][DOI: https://dx.doi.org/10.1016/j.toxlet.2011.12.008]

39. Venn, AJ. Effects of volatile organic compounds, damp, and other environmental exposures in the home on wheezing illness in children. Thorax; 2003; 58, pp. 955-960.1:STN:280:DC%2BD3srhtVWqtg%3D%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/14586048][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1746513][DOI: https://dx.doi.org/10.1136/thorax.58.11.955]

40. Stamatelopoulou, A; Asimakopoulos, DN; Maggos, T. Effects of PM, TVOCs and comfort parameters on indoor air quality of residences with young children. Build. Environ.; 2019; 150, pp. 233-244. [DOI: https://dx.doi.org/10.1016/j.buildenv.2018.12.065]

41. Kharel, M et al. Assessing volatile organic compound level in selected workplaces of Kathmandu Valley. Heliyon; 2021; 7, e08262.1:CAS:528:DC%2BB38XitVertbzJ [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34765781][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8571507][DOI: https://dx.doi.org/10.1016/j.heliyon.2021.e08262]

42. Bolden, AL; Kwiatkowski, CF; Colborn, T. New look at BTEX: are ambient levels a Problem? Environ. Sci. Technol.; 2015; 49, pp. 5261-5276.1:CAS:528:DC%2BC2MXmsFagsbw%3D [DOI: https://dx.doi.org/10.1021/es505316f]

43. Athersuch, TJ; Keun, HC. Metabolic profiling in human exposome studies. Mutagenesis; 2015; 30, pp. 755-762.1:CAS:528:DC%2BC2MXitVKgsrfP [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26290610]

44. Bouatra, S et al. The human urine metabolome. PloS One; 2013; 8, e73076.2013PLoSO..873076B1:CAS:528:DC%2BC3sXhsVCrsrnJ [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24023812][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3762851][DOI: https://dx.doi.org/10.1371/journal.pone.0073076]

45. Rappaport, SM; Barupal, DK; Wishart, D; Vineis, P; Scalbert, A. The blood exposome and its role in discovering causes of disease. Environ. Health Perspect.; 2014; 122, pp. 769-774. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24659601][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4123034][DOI: https://dx.doi.org/10.1289/ehp.1308015]

46. Agency for Toxic Substances and Disease Registry (ATSDR). Toxic Substances Portal | ATSDR. https://wwwn.cdc.gov/TSP/index.aspx

47. U.S. Department of Health and Human Services. Public Health Service. Toxicological Profile for Acrylamide. Agency Toxic. Subst. Dis. Regist (2012).

48. U.S. EPA & SUMMARY REVIEW OF THE HEALTH EFFECTS ASSOCIATED WITH PROPYLENE OXIDE: HEALTH ISSUE ASSESSMENT. US Environ. Prot. Agency EPA (2004). /600/8–86/007F.

49. Rianto, BUD; Yudhanto, D; Herdini, C. The correlation between length of work and Nasal Mucociliary Transport Time of Gas/Fuel Station Workers. Kobe J. Med. Sci.; 2018; 64, pp. E6-E10.1:CAS:528:DC%2BC1MXhsVeisLbM [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30282892][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6192822]

50. Biagioni, B; Annesi-Maesano, I; D’Amato, G; Cecchi, L. The rising of allergic respiratory diseases in a changing world: from climate change to migration. Expert Rev. Respir Med.; 2020; 14, pp. 973-986.1:CAS:528:DC%2BB3cXhsFamsrzE [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32662693][DOI: https://dx.doi.org/10.1080/17476348.2020.1794829]

51. Papadopoulos, N. G. et al. Addressing Adverse Synergies between Chemical and Biological Pollutants at schools—The ‘SynAir-G’ Hypothesis. Allergy n/a.

52. Yang, J et al. Association between serum cotinine and volatile organic compounds (VOCs) in adults living with HIV, HBV, or HCV (NHANES 2005–2018). Sci. Rep.; 2022; 12, 21769.2022NatSR.1221769Y1:CAS:528:DC%2BB38XjtFOmsb%2FO [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36526719][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9758166][DOI: https://dx.doi.org/10.1038/s41598-022-26420-7]

53. Norbäck, D; Hashim, Z; Ali, F; Hashim, JH. Asthma symptoms and respiratory infections in Malaysian students-associations with ethnicity and chemical exposure at home and school. Environ. Res.; 2021; 197, 111061. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33785322][DOI: https://dx.doi.org/10.1016/j.envres.2021.111061]

54. Obeng, GM; Aram, SA; Agyei, D; Saalidong, BM. Exposure to particulate matter (PM2.5) and volatile organic compounds (VOCs), and self-reported health symptoms among fish smokers: a case study in the Western Region of Ghana. PloS One; 2023; 18, e0283438.1:CAS:528:DC%2BB3sXmsVyqsL8%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36961812][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10038312][DOI: https://dx.doi.org/10.1371/journal.pone.0283438]

55. Sheppard, N; Carroll, M; Gao, C; Lane, T. Particulate matter air pollution and COVID-19 infection, severity, and mortality: a systematic review and meta-analysis. Sci. Total Environ.; 2023; 880, 163272.1:CAS:528:DC%2BB3sXnslCgsL8%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37030371][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10079587][DOI: https://dx.doi.org/10.1016/j.scitotenv.2023.163272]

56. Pompilio, A; Di Bonaventura, G. Ambient air pollution and respiratory bacterial infections, a troubling association: epidemiology, underlying mechanisms, and future challenges. Crit. Rev. Microbiol.; 2020; 46, pp. 600-630.1:CAS:528:DC%2BB3cXitFGmtb3E [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33059504][DOI: https://dx.doi.org/10.1080/1040841X.2020.1816894]

57. Herberth, G et al. Maternal and cord blood miR-223 expression associates with prenatal tobacco smoke exposure and low regulatory T-cell numbers. J. Allergy Clin. Immunol.; 2014; 133, pp. 543-550e4.1:CAS:528:DC%2BC3sXhtlektLrO [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23978443][DOI: https://dx.doi.org/10.1016/j.jaci.2013.06.036]

58. Haro-García, LC et al. Production of IL-10, TNF and IL-12 by Peripheral Blood mononuclear cells in Mexican workers exposed to a mixture of benzene–toluene–xylene. Arch. Med. Res.; 2012; 43, pp. 51-57. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22300681][DOI: https://dx.doi.org/10.1016/j.arcmed.2012.01.008]

59. Bulog, A; Karaconji, IB; Sutić, I; Mićović, V. Immunomodulation of cell-mediated cytotoxicity after chronic exposure to vapors. Coll. Antropol; 2011; 35, Suppl 2 pp. 61-64. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22220405]

60. Bönisch, U et al. Volatile organic compounds enhance allergic airway inflammation in an experimental mouse model. PloS One; 2012; 7, e39817.2012PLoSO..739817B [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22802943][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3389035][DOI: https://dx.doi.org/10.1371/journal.pone.0039817]

61. Kwon, JW; Park, HW; Kim, WJ; Kim, MG; Lee, SJ. Exposure to volatile organic compounds and airway inflammation. Environ. Health; 2018; 17, 65. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30086760][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6081941][DOI: https://dx.doi.org/10.1186/s12940-018-0410-1]

62. Morrison, D; Rahman, I; MacNee, W. Permeability, inflammation and oxidant status in airspace epithelium exposed to ozone. Respir Med.; 2006; 100, pp. 2227-2234. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17023150][DOI: https://dx.doi.org/10.1016/j.rmed.2005.10.005]

63. Liu, J et al. Particulate matter disrupts airway epithelial barrier via oxidative stress to promote Pseudomonas aeruginosa infection. J. Thorac. Dis.; 2019; 11, pp. 2617-2627.1:CAS:528:DC%2BC1MXhtl2qsbs%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31372298][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6626802][DOI: https://dx.doi.org/10.21037/jtd.2019.05.77]

64. Glencross, DA; Ho, TR; Camiña, N; Hawrylowicz, CM; Pfeffer, PE. Air pollution and its effects on the immune system. Free Radic Biol. Med.; 2020; 151, pp. 56-68.1:CAS:528:DC%2BB3cXjsVOlu78%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32007522][DOI: https://dx.doi.org/10.1016/j.freeradbiomed.2020.01.179]

65. Uzma, N; Kumar, BS; Hazari, MAH. Exposure to benzene induces oxidative stress, alters the immune response and expression of p53 in gasoline filling workers. Am. J. Ind. Med.; 2010; 53, pp. 1264-1270.1:CAS:528:DC%2BC3cXhs1enu7bF [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20886531][DOI: https://dx.doi.org/10.1002/ajim.20901]

66. Kirkeleit, J; Ulvestad, E; Riise, T; Bråtveit, M; Moen, BE. Acute suppression of serum IgM and IgA in Tank workers exposed to Benzene. Scand. J. Immunol.; 2006; 64, pp. 690-698.1:CAS:528:DC%2BD28XhtlCktLnI [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17083627][DOI: https://dx.doi.org/10.1111/j.1365-3083.2006.01858.x]

67. St Helen, G et al. Differences in exposure to toxic and/or carcinogenic volatile organic compounds between Black and White cigarette smokers. J. Expo Sci. Environ. Epidemiol.; 2021; 31, pp. 211-223.1:CAS:528:DC%2BC1MXhsFylsL%2FP [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31406274][DOI: https://dx.doi.org/10.1038/s41370-019-0159-9]

68. Wahlang, B et al. Associations between residential volatile organic compound exposures and liver injury markers: the role of biological sex and race. Environ. Res.; 2023; 221, 115228.1:CAS:528:DC%2BB3sXht1ykur8%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36610539][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9957966][DOI: https://dx.doi.org/10.1016/j.envres.2023.115228]

69. Wu, XM et al. Exposures to volatile organic compounds (VOCs) and associated health risks of socio-economically disadvantaged population in a ‘hot spot’ in Camden, New Jersey. Atmospheric Environ. Oxf. Engl. 1994; 2012; 57, pp. 72-79.1:CAS:528:DC%2BC38XovFynt7k%3D

70. Schuyler, AJ; Wenzel, SE. Historical redlining impacts contemporary environmental and asthma-related outcomes in black adults. Am. J. Respir Crit. Care Med.; 2022; 206, pp. 824-837.1:CAS:528:DC%2BB38XivVCmsrfF [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35612914][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9799280][DOI: https://dx.doi.org/10.1164/rccm.202112-2707OC]

71. He, T; Jin, L; Li, X. On the triad of air PM pollution, pathogenic bioaerosols, and lower respiratory infection. Environ. Geochem. Health; 2023; 45, pp. 1067-1077.1:CAS:528:DC%2BB3MXhsFGhtbrK [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34236582][DOI: https://dx.doi.org/10.1007/s10653-021-01025-7]

72. Cohen, S. Social status and susceptibility to respiratory infections. Ann. N Y Acad. Sci.; 1999; 896, pp. 246-253.1999NYASA.896.246C1:STN:280:DC%2BD3c7ks1CrsA%3D%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10681901][DOI: https://dx.doi.org/10.1111/j.1749-6632.1999.tb08119.x]

73. Gaffney, AW; Himmelstein, DU; Christiani, DC; Woolhandler, S. Socioeconomic Inequality in Respiratory Health in the US from 1959 to 2018. JAMA Intern. Med.; 2021; 181, pp. 968-976. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34047754][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8261605][DOI: https://dx.doi.org/10.1001/jamainternmed.2021.2441]

74. Jbaily, A et al. Air pollution exposure disparities across US population and income groups. Nature; 2022; 601, pp. 228-233.2022Natur.601.228J1:CAS:528:DC%2BB38XhtVOlu7k%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35022594][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10516300][DOI: https://dx.doi.org/10.1038/s41586-021-04190-y]

75. Michanowicz, DR et al. Home is where the Pipeline ends: characterization of volatile Organic compounds Present in Natural Gas at the point of the residential end user. Environ. Sci. Technol.; 2022; 56, pp. 10258-10268.2022EnST..5610258M1:CAS:528:DC%2BB38XhsF2it7%2FI [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35762409][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9301916][DOI: https://dx.doi.org/10.1021/acs.est.1c08298]

76. Janasik, B; Jakubowski, M; Jałowiecki, P. Excretion of unchanged volatile organic compounds (toluene, ethylbenzene, xylene and mesitylene) in urine as result of experimental human volunteer exposure. Int. Arch. Occup. Environ. Health; 2008; 81, pp. 443-449.1:CAS:528:DC%2BD1cXislCnt7o%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17680265][DOI: https://dx.doi.org/10.1007/s00420-007-0233-9]

77. United States Environmental Protection Agency. Health Effects Notebook for Hazardous Air Pollutants.