Associations Between Ambient Exposure to Styrene, Benzene, Toluene, Ethylbenzene, and Xylenes and Hospitalizations in the US Gulf Region

Abstract

Background

Styrene, benzene, toluene, ethylbenzene, and xylenes (SBTEX) are hazardous air pollutants studied primarily in occupational settings, where high exposures are linked to adverse health outcomes and subject to strict regulation. In contrast, the health impacts of ambient SBTEX exposure, which occurs at lower concentrations, remain poorly characterized, largely because comprehensive exposure estimates are limited.

Methods

The study population comprised Medicare beneficiaries aged 65 years and older in the US Gulf region during 2011–2016. Outcomes were hospitalizations for cardiovascular (n = 601,631) and respiratory (n = 385,162) diseases. Ambient SBTEX exposures were derived from validated prediction models with high spatial (4 × 4 km) and temporal (daily) resolution. A time-stratified case-crossover design with conditional logistic regression was applied to estimate associations between daily ambient SBTEX concentrations and hospitalization risk.

Findings

We observed significant associations between ambient concentrations of benzene, toluene, ethylbenzene, and xylenes (BTEX) and cardiovascular and respiratory hospitalizations, but not for styrene. An interquartile range increase in daily mean benzene concentration (0.087 ppb) was associated with a 0.38% (95% CI: 0.08, 0.68) increase in cardiovascular hospitalizations at lag 0 and a 1.00% (95% CI: 0.60, 1.41) increase over lags 0–7. The corresponding increases in respiratory hospitalizations were 0.88% (95% CI: 0.50, 1.25) at lag 0 and 2.74% (95% CI: 2.29, 3.19) over lags 0–7. Similar associations were observed for toluene, ethylbenzene, and xylenes. An interquartile range increase in total BTEX (0.297 ppb) was associated with increases of 0.85% (95% CI: 0.44, 1.27) and 1.87% (95% CI: 1.41, 2.32) in cardiovascular and respiratory hospitalizations over lags 0–7, respectively.

Interpretation

Even at low ambient concentrations, BTEX exposures may increase the risk of cardiovascular and respiratory hospitalization among older adults. These findings support greater public health attention to ambient BTEX, extending beyond traditional concerns focused on occupational and industrial settings.

Graphical Abstract

1. Introduction

Styrene, benzene, toluene, ethylbenzene, and xylenes (SBTEX) are volatile organic compounds predominantly emitted from anthropogenic sources, including vehicular exhaust, petroleum refining, and other industrial activities.^1,2 They are widely used as solvents and chemical feedstocks and are present in fuels and various commercial products.^1,2

The toxicological properties of SBTEX at high concentrations, particularly in occupational settings, are well documented. In 1987, the International Agency for Research on Cancer (IARC) classified benzene as a Group 1 carcinogen (carcinogenic to humans), indicating sufficient evidence of carcinogenicity in humans.³ Styrene and ethylbenzene were subsequently classified as Group 2A (probably carcinogenic to humans) and Group 2B (possibly carcinogenic to humans), respectively.^4,5 Beyond carcinogenicity, high-level SBTEX exposures has been linked to central nervous system effects, including dizziness, headaches, and cognitive impairment.^6,7 Such exposures have also been associated with reproductive toxicity (e.g., sperm abnormalities and reduced fetal growth) and with cardiovascular diseases, respiratory dysfunction, asthma, and sensitization to common antigens.⁷

Due to their toxicological properties, exposure to SBTEX is strictly regulated in occupational settings. In the United States (US), agencies including the Occupational Safety and Health Administration (OSHA), the National Institute for Occupational Safety and Health (NIOSH), and the American Conference of Governmental Industrial Hygienists (ACGIH) have established permissible exposure limits, including time-weighted averages (TWAs) and short-term exposure limits (STELs), to protect worker health.^8–10 For example, NIOSH recommends that occupational benzene exposure not exceed 0.1 parts per million (ppm) averaged over a 10-hour shift, and not exceed 1 ppm at any time.⁹ The World Health Organization (WHO) further notes that no safe threshold exists for benzene exposure, underscoring the importance of minimizing both acute and chronic exposures.¹¹ In addition, the US Environmental Protection Agency (EPA) classified all SBTEX compounds as Hazardous Air Pollutants (HAPs), and regulates their emissions under the Clean Air Act to protect public health.¹²

Although regulatory efforts primarily target occupational settings and industrial emissions, SBTEX compounds are also present in ambient environments, typically at much lower concentrations. Nevertheless, evidence regarding health effects of these ambient concentrations remains limited, largely because exposure assessment outside occupational settings is challenging. Compared with other regulated air pollutants, ambient SBTEX monitoring is characterized by sparse spatial coverage and infrequent sampling. Ground-based monitoring stations are often located near known emission sources, leading to uneven geographic coverage that hampers accurate exposure estimation for the general population and constrains large-scale epidemiological research. For example, in the highly industrialized US Gulf region, ambient HAP monitoring is limited to four EPA Ambient Monitoring Technology Information Center (AMTIC) sites in Louisiana and 42 in Texas. These sites are predominantly clustered around major urban centers such as Houston, Dallas, and Baton Rouge, which reduces coverage and may limit the generalizability of findings to rural areas and broader populations.¹³

Consist with these exposure-assessment limitations, epidemiological evidence linking ambient SBTEX to adverse health outcomes remains scarce. To our knowledge, US-based research is limited to a single study of associations between ambient benzene, toluene, ethylbenzene, and xylenes (BTEX) and mortality, based on data from 2001–2006.¹⁴ Given changes in industrial emission profiles and environmental regulations over time, updated evidence is needed to reflect contemporary atmospheric conditions and health risks. Moreover, no large-scale US studies have evaluated associations between ambient SBTEX exposure and hospitalizations.

Because ambient exposure affects broad populations, including vulnerable groups such as older adults, there is a critical need for evidence on the health risks of ambient SBTEX. In this study, we aimed to address this gap by leveraging modeled ambient SBTEX concentrations with enhanced spatial and temporal resolution across the US Gulf region. Specifically, we examined associations between short-term ambient SBTEX exposure and hospitalizations for cardiovascular and respiratory diseases among Medicare beneficiaries aged 65 years and older. The US Gulf region, home to more than 120 operational petroleum refineries, provides a compelling setting for this investigation because of its substantial industrial activity and corresponding elevated SBTEX emissions.

2. Methods

2.1. Study population.

The study area included 1,643 ZIP codes in the US Gulf region (Figure 1). The study population comprised Medicare fee-for-service beneficiaries aged 65 years and older who resided in these ZIP codes during 2011 to 2016. Because Medicare covers nearly all US citizens and legal permanent residents aged 65 years and older, the study population is expected to be broadly representative of older adults residing in the Gulf region. Using Medicare Part A Medicare Provider Analysis and Review (MedPAR) and enrollment files, we obtained admission dates, diagnosis codes, ZIP code of residence, age, sex, race/ethnicity, and Medicaid eligibility. We included only the first hospitalization per beneficiary for cardiovascular or respiratory diseases, identified using the International Classification of Diseases, Ninth Revision (ICD-9), and Tenth Revision (ICD-10) codes: 390-459 (ICD-9) or I00-I99 (ICD-10) for cardiovascular hospitalizations, and 460-519 (ICD-9) or J00-J99 (ICD-10) for respiratory hospitalizations. To focus on community-dwelling beneficiaries, we restricted admissions to those originating from non-healthcare settings, clinic referrals, or the emergency department.

Figure 1.

Study area in the US Gulf region. The blue box denotes the modeling domain used for estimating ambient SBTEX concentrations, and solid black lines indicate the boundaries of individual ZIP code areas comprising the study area.

2.2. Exposure assessment.

Hourly ambient concentrations of SBTEX were estimated using gridded hourly data simulated by the Comprehensive Air Quality Model with Extensions version 7.0 (CAMx 7.0). This modeling incorporated emissions data from the National Emissions Inventory of the US EPA, augmented by the Hazardous Air Pollution Imputation program to address missing data.¹³ Hourly, spatially resolved emissions inputs for CAMx were generated using the Sparse Matrix Operator Kernel Emissions (SMOKE) system, and a reactive tracer module was used to simulate SBTEX concentrations, while accounting for chemical transformation and removal processes. The model output provided hourly concentrations of SBTEX at a spatial resolution of 4 × 4 km across the Gulf region. Model performance was evaluated against measurements from the EPA AMTIC network, demonstrating good agreement with observations. The overall correlation coefficient (R) was 0.75 (styrene: 0.64, benzene: 0.68, toluene: 0.46, ethylbenzene: 0.77, xylenes: 0.77), and the normalized mean bias was −5.6% (styrene: 32.1%, benzene: 12.5%, toluene: −6.7%, ethylbenzene: −21.4%, xylenes: −24.9%). Detailed modeling methods and validation results are presented elsewhere.¹³ Subsequently, the 4 km gridded hourly predictions of SBTEX concentrations were aggregated to ZIP code-level daily mean concentrations to align with the smallest geographic unit available in the Medicare data. The top 0.5% of SBTEX values were excluded to minimize the influence of extreme observations. ZIP code-specific daily estimated SBTEX concentrations were then linked to each hospitalization record using ZIP code of residence and admission date.

2.3. Covariates.

Daily mean fine particulate matter (PM2.5) concentrations were obtained from a previously published dataset based on deep learning methods.¹⁵ Daily mean temperature were obtained from Daymet Version 4, which provides gridded daily surface weather for North America using station-based interpolation and elevation models.¹⁶ Daily mean relative humidity was obtained from gridMET, a gridded meteorological dataset developed for ecological applications.¹⁷ These environmental datasets were aggregated to the ZIP code-level and linked to each hospitalization by ZIP code of residence and admission date.

We also compiled ZIP code-level socioeconomic variables. Median household income and the percentage of residents who were Black were obtained from the US Census Bureau. Urbanicity, based on Rural-Urban Commuting Area (RUCA) codes, was obtained from the US Department of Agriculture.¹⁸

2.4. Statistical analysis.

We employed a time-stratified case-crossover design with conditional logistic regression to examine associations between short-term ambient SBTEX exposure and hospitalizations for cardiovascular and respiratory diseases. In this design, each individual serves as their own control by contrasting pollutant exposure on the hospitalization date (case day) with exposure on matched control days. A key strength of this approach is that it inherently controls for individual-level confounders that remain constant over short time periods, such as age, sex, smoking status, and underlying chronic conditions.¹⁹ Control days were selected by matching the same day of the week within the same month and year as the hospitalization date, thereby accounting for day-of-week effects and long-term and seasonal and temporal trends in exposure.²⁰ We fitted conditional logistic regression models adjusting for the following covariates: daily mean temperature modeled using natural splines with 3 degrees of freedom (df), daily mean PM2.5 concentration modeled as a linear term, and relative humidity modeled using natural splines with 3 df.

To characterize the concentration-response relationship, we compared several functional forms (linear, log-transformed, quadratic, natural splines with 3 df, and penalized spline). Model selection for the optimal concentration-response relationship was guided by the Bayesian information criterion (BIC), selecting the model with the lowest BIC, in conjunction with visual inspection of concentration-response plots. We also examined the lagged effects of SBTEX on hospitalizations. We estimated 7-day cumulative effects using distributed lag models (DLMs), in which the lag-response relationship was modeled using a natural spline with 3 df. Analyses were conducted for each SBTEX compound individually, and we additionally combined effects using the sum of concentrations for compounds showing significant associations.

We assessed potential heterogeneity by conducting stratified analyses by age group (65-74, 75-84, ≥85 years), sex (male, female), race/ethnicity (White, Black, Asian, Hispanic), Medicaid eligibility (eligible, ineligible), ZIP code-level median household income (low, medium, high), ZIP code-level percent of Black residents (low, medium, high), and urbanicity (metropolitan, small town/rural). To formally assess heterogeneity across these strata, interaction terms between SBTEX concentrations and each stratifying variable were included in separate conditional logistic regression models, and statistical significance was evaluated using Wald tests for these interaction terms. Results are reported per IQR increase to facilitate comparisons across pollutants and to reflect typical exposure variability.

We performed several sensitivity analyses. Because temperature and PM_2.5 may have lagged effects that could confound the associations, we additionally adjusted for lag structures of temperature and PM_2.5. Temperature was modeled over lags up to two days (lags 0–2), using a B-spline with 3 df for the exposure-response function and a natural spline for the lag-response function. PM_2.5 was modeled over lags 0–2 using a moving-average term.²¹ We also performed sensitivity analyses using alternative specifications of the lag structure for SBTEX, including models with 3-, 5-, and 7-day lag windows and natural spline lag functions with 3 or 4 df. Finally, we adjusted for ozone as a co-pollutant and repeated analyses using alternative trimming thresholds for extreme SBTEX concentrations (excluding the top 1%, 2%, and 3%).

3. Results

3.1. Descriptive statistics.

We identified 601,631 hospitalizations for cardiovascular diseases and 385,162 hospitalizations for respiratory diseases during 2011–2016. Among beneficiaries hospitalized for cardiovascular diseases, 40.4% were aged 65–74 years, 45.5% were male, 79.9% were White, and 18.5% were eligible for Medicaid. Overall, 89.4% of hospitalizations occurred among beneficiaries residing in metropolitan areas. Characteristics of beneficiaries hospitalized for respiratory diseases were similar to those of beneficiaries hospitalized for cardiovascular diseases (Table 1).

Table 1.

Demographic characteristics of the study population by type of hospitalization (cardiovascular or respiratory), US Gulf region, 2011-2016

	Cardiovascular, n (%)	Respiratory, n (%)
Total	601,631	385,162
Individual-level characteristics
Age group (years)
65-74	243,006 (40.4)	138,427 (35.9)
75-84	219,193 (36.4)	141,310 (36.7)
≥85	139,432 (23.2)	105,425 (27.4)
Sex
Male	273,731 (45.5)	170,112 (44.2)
Female	327,900 (54.5)	215,050 (55.8)
Race/Ethnicity
White	480,572 (79.9)	310,676 (80.7)
Black	99,932 (16.6)	60,798 (15.8)
Asian	6,167 (1.0)	4,373 (1.1)
Hispanic	7,637 (1.3)	5,149 (1.3)
Medicaid eligibility
Ineligible	490,287 (81.5)	298,862 (77.6)
Eligible	111,344 (18.5)	86,300 (22.4)
Area level characteristics (ZIP code-level)
Median household income
Low	199,323 (33.1)	129,790 (33.7)
Medium	196,875 (32.7)	125,419 (32.6)
High	205,402 (34.1)	129,930 (33.7)
Percent Black residents
Low	204,293 (34.0)	130,120 (33.8)
Medium	198,871 (33.1)	128,693 (33.4)
High	198,457 (33.0)	126,340 (32.8)
Urbanicity
Metropolitan	537,998 (89.4)	343,522 (89.2)
Small town/Rural	62,284 (10.4)	40,734 (10.6)

Table 2 summarizes the distribution of daily mean concentrations of ambient SBTEX and other environmental covariates (PM_2.5, temperature, and relative humidity) at the ZIP code-level in the US Gulf region during 2011-2016. Among the SBTEX compounds, toluene and benzene had the highest mean concentrations (0.149 and 0.117 ppb, respectively), followed by xylenes (0.070 ppb), ethylbenzene (0.018 ppb), and styrene (0.005 ppb). The mean PM2.5 concentration during the study period was 9.194 μg/m³, with a mean temperature of 20.03°C and a mean relative humidity of 65.35%. Correlation analysis revealed strong positive correlations among SBTEX compounds, with particularly high correlations observed among toluene, ethylbenzene, and xylenes (r = 0.94-0.97) (eTable 1). Styrene showed moderate correlations with other SBTEX compounds (r = 0.47-0.52), and correlations between SBTEX and PM_2.5 were relatively weak (r < 0.18).

Table 2.

Summary statistics of ZIP code-specific daily mean concentrations of ambient air pollutants (styrene, benzene, toluene, ethylbenzene, xylenes, and PM_2.5) and meteorological variables (temperature and relative humidity), US Gulf region, 2011–2016

	Min	Q1	Median	Mean	Q3	Max
Styrene (ppb)	0.000	0.001	0.001	0.005	0.004	0.119
Benzene (ppb)	0.001	0.050	0.083	0.117	0.137	0.890
Toluene (ppb)	0.001	0.042	0.079	0.149	0.173	1.354
Ethylbenzene (ppb)	0.000	0.006	0.010	0.018	0.022	0.154
Xylenes (ppb)	0.000	0.020	0.038	0.070	0.083	0.618
PM2.5 (μg/m3)	0.098	6.464	8.561	9.194	11.180	61.360
Temperature (°C)	−9.21	14.15	21.45	20.03	26.97	36.24
Relative humidity (%)	7.68	58.16	66.85	65.35	73.59	100.00

3.2. Concentration-response relationships.

For simplicity and interpretability, we selected the linear model as the main model. Although the log-transformed model generally exhibited the lowest BIC among evaluated models (linear, log-transformed, quadratic, natural splines, and penalized spline), the improvement was typically small. For instance, when examining the association between benzene exposure and cardiovascular hospitalizations, the linear model (BIC = 1,688,234) and the log-transformed model (BIC = 1,688,232) differed by only 2 BIC units, suggesting a limited practical advantage of using the log-transformed specification (eTable 2). Visual inspection of the concentration-response curves further supported the similarity between these two models (eFigures 1 and 2).

3.3. Health effect estimates.

We observed statistically significant associations between ambient BTEX concentrations and increases in cardiovascular and respiratory hospitalizations (Table 3). At lag 0, an IQR increase in daily mean ambient benzene concentration (0.087 ppb) was associated with a 0.38% (95% CI: 0.08, 0.68) increase in cardiovascular hospitalizations and a 0.88% (95% CI: 0.50, 1.25) increase in respiratory hospitalizations. Similar associations were observed for toluene, ethylbenzene, and xylenes, indicating consistent positive associations between ambient BTEX concentrations and hospitalizations. The combined effect of BTEX at lag 0 remained significant, with an IQR increase in the sum of BTEX concentrations associated with a 0.34% (95% CI: 0.05, 0.64) increase in cardiovascular hospitalizations and a 0.80% (95% CI: 0.42, 1.17) increase in respiratory hospitalizations. No statistically significant associations were observed between ambient styrene concentrations and either hospitalization outcome.

Table 3.

Percent increase (95% CI) in cardiovascular and respiratory hospitalizations associated with an interquartile range increase in ambient concentrations of styrene, benzene, toluene, ethylbenzene, and xylenes, US Gulf region, 2011-2016

Pollutant	Cardiovascular disease		Respiratory disease
	Lag 0	Lags 0–7	Lag 0	Lags 0–7
Styrene	0.03 (−0.09, 0.16)	0.17 (−0.00, 0.35)	0.09 (−0.07, 0.24)	0.17 (−0.02, 0.36)
Benzene	0.38 (0.08, 0.68)	1.00 (0.60, 1.41)	0.88 (0.50, 1.25)	2.74 (2.29, 3.19)
Toluene	0.31 (0.04, 0.59)	0.54 (0.15, 0.92)	0.68 (0.34, 1.03)	1.16 (0.74, 1.58)
Ethylbenzene	0.41 (0.11, 0.71)	0.66 (0.25, 1.07)	0.83 (0.46, 1.20)	1.43 (0.98, 1.88)
Xylenes	0.40 (0.11, 0.69)	0.68 (0.27, 1.09)	0.88 (0.52, 1.24)	1.61 (1.16, 2.06)
BTEX	0.34 (0.05, 0.64)	0.85 (0.44, 1.27)	0.80 (0.42, 1.17)	1.87 (1.41, 2.32)

Lag 0 indicates the effect on the day of exposure, and lags 0-7 represent the cumulative effect over the 7 days following exposure.

Interquartile ranges for pollutant concentrations: Styrene = 0.003 ppb, benzene = 0.087 ppb, toluene = 0.131 ppb, ethylbenzene = 0.016 ppb, xylenes = 0.063 ppb, and BTEX = 0.297 ppb.

Cumulative associations up to 7 days of lagged exposure showed stronger effect estimates between BTEX and hospitalizations. Specifically, cumulative exposure to benzene over lags 0–7 was associated with a 1.00% (95% CI: 0.60, 1.41) increase in cardiovascular hospitalizations and a 2.74% (95% CI: 2.29, 3.19) increase in respiratory hospitalizations (Table 3). For the lag-specific associations, cardiovascular effect estimates declined steadily over the 7-day period, whereas respiratory effect estimates fell quickly after the first few days and then showed a slight rebound at later lags, particularly for benzene (Figure 3). Similar associations were observed for the sum of BTEX (eFigure 3).

Figure 3. Lag-response associations between ambient exposure to styrene, benzene, toluene, ethylbenzene, and xylenes and the risk of cardiovascular and respiratory hospitalizations (per interquartile range increase).

Interquartile ranges for pollutant concentrations: Styrene = 0.003 ppb, Benzene = 0.087 ppb, Toluene = 0.131 ppb, Ethylbenzene = 0.016 ppb, and Xylenes = 0.063 ppb.

We also observed significant effect modification by age and urbanicity for the associations between ambient SBTEX exposure and respiratory hospitalizations (Figure 2). Stratified analyses indicated higher effect estimates among individuals aged 85 years and older compared to those aged 65-74 years, with interaction p-values of 0.02, 0.01, 0.03, ≤ 0.01, and 0.03 for benzene, toluene, ethylbenzene, xylenes, and sum of BTEX, respectively (eTable 5). Additionally, individuals residing in small towns or rural areas tended to show higher effect estimates compared to those living in metropolitan areas, particularly for toluene (p = 0.03), ethylbenzene (p = 0.06), xylenes (p = 0.04), and sum of BTEX (p = 0.03).

Figure 2. Associations between ambient exposure to styrene, benzene, toluene, ethylbenzene, and xylenes and cardiovascular and respiratory hospitalizations, stratified by socioeconomic and demographic factors. (lag 0; per interquartile range increase).

Interquartile ranges for pollutant concentrations: Styrene = 0.003 ppb, Benzene = 0.087 ppb, Toluene = 0.131 ppb, Ethylbenzene = 0.016 ppb, and Xylenes = 0.063 ppb.

Sensitivity analyses demonstrated that the findings were robust. The results remained generally consistent after additional adjustment for lagged effects of PM_2.5 and temperature. For example, for the association between benzene and respiratory hospitalizations, the estimated effect decreased slightly from 0.88% (95% CI: 0.50, 1.25) to 0.60% (95% CI: 0.39, 0.80) after adjusting for lagged effects (lags 0–2) of PM_2.5 and temperature (eFigure 4). However, this difference was not statistically significant (Wald test p = 0.20). Further adjustment for ozone and the use of different cutoffs for extreme values did not materially change the main results (eFigures 5 and 6). The lag structure did not vary across different choices of df, and the results were consistent with the main lag specification (eFigure 7).

4. Discussion

In this large population-based study of Medicare beneficiaries aged 65 years and older in the US Gulf region (2011–2016), we found positive associations between ambient BTEX concentrations and hospitalizations for cardiovascular and respiratory diseases. An IQR increase in daily mean ambient benzene concentration was associated with a 1.00% (95% CI: 0.60, 1.41) increase in cardiovascular hospitalizations and a 2.74% (95% CI: 2.29, 3.19) increase in respiratory hospitalizations over lags 0–7. We observed consistent associations for toluene, ethylbenzene, and xylenes. The combined BTEX metric was similarly associated with increased cardiovascular and respiratory hospitalizations over lags 0–7 (0.85% and 1.87%, respectively). Associations tended to be larger among adults aged 85 years and older and among residents of small towns or rural areas.

A growing epidemiological literature suggests that even low-level ambient BTEX exposure may adversely affect cardiovascular and respiratory health outcomes.²² In a recent multi-country time-series study spanning 757 locations in 46 countries, an increase in short-term BTEX was associated with higher cardiovascular and respiratory mortality.¹⁴ Specifically, the authors reported that an IQR increase in short-term BTEX concentration was associated with 0.42% (95% CI: 0.30, 0.54) and 0.68% (95% CI: 0.50, 0.86) increases in daily cardiovascular and respiratory mortality, respectively. In the US-specific analysis (208 locations), an IQR increase in ambient BTEX was associated with a 0.46% (95% CI: −0.25, 1.68) increase in total mortality. Similarly, a study in Hong Kong reported that short-term exposure to ambient benzene and combined exposure to toluene, ethylbenzene, and xylenes were associated 5.8% (95% CI: 1.0, 10.8) and 3.5% (95% CI: 1.0, 6.1) increases in circulatory mortality, respectively.²³

Evidence also extends beyond mortality to morbidity outcomes. In Hong Kong, short-term increases in ambient benzene concentrations were associated with a 2.7% increase (95% CI: 0.39, 5.04) in emergency hospitalizations for heart failure (n = 54,003).²⁴ For respiratory morbidity, Ran et al. analyzed emergency hospitalizations for chronic obstructive pulmonary disease (COPD) from 42 public hospitals in Hong Kong (n = 75,113) and found that short-term exposure to ambient benzene and toluene was associated with 2.62% (95% CI: 0.17, 5.13) and 1.42% (95% CI: 0.16, 2.69) increases in COPD-related hospitalizations, respectively.²⁵ Adverse health impacts are not limited to acute exposures: a cohort study of 277,585 United Kingdom residents reported that long-term ambient benzene exposure was associated with an increased risk of heart failure (HR: 1.22; 95% CI: 1.07, 1.39),26 and another cohort study of 393,042 United Kingdom adults reported elevated risks of cardiovascular mortality (HR: 1.24; 95% CI: 1.21, 1.28) and respiratory mortality (HR: 1.25; 95% CI: 1.20, 1.30) with long-term ambient benzene exposure.^26,27 Cross-sectional evidence has further linked higher ambient BTEX concentrations with impaired lung function, including reduced peak expiratory flow and ventilatory dysfunction.^28,29 Taken together, these studies align with our results and support the hypothesis that ambient BTEX exposure contributes to the cardiovascular and respiratory disease burden in non-occupational settings.

Several biological mechanisms could plausibly explain the observed associations. Previous studies reported that even short-term exposure to BTEX over several hours to a few days may increase oxidative stress and trigger inflammatory responses, as reflected in elevated oxidative stress markers and pro- inflammatory cytokines and reduced antioxidant defenses.^30,31 In the respiratory system, benzene, toluene, and xylenes can irritate mucosal surfaces and provoke airway inflammation, potentially exacerbating underlying airway disease and precipitating decompensation among susceptible individuals.^22,32,33 Acute inhalation exposure may also damage lung tissue, reduce lung function, and worsen asthma.^22,32,33 The modest rebound observed at longer lags for respiratory outcomes could reflect delayed downstream pathways, such as inflammatory or oxidative stress responses, that evolve after the initial irritant effects. Future studies incorporating clinical measurements and time-resolved biomarkers could help clarify these mechanisms.

In contrast to BTEX, we did not observe significant associations between ambient styrene exposure and hospitalizations. This may reflect differences in toxicological potency between styrene and the other compounds. Prior studies summarizing the health effects of styrene suggest that styrene-related effects are more evident in chronic or neurological domains rather than in acute cardiovascular or respiratory outcomes.^34,35 This difference may be attributable to the toxicological pathway through which styrene acts, primarily involving the metabolism of styrene to reactive intermediates that can contribute to neurotoxicity and cellular damage.³⁵ Although styrene exposure can provoke acute respiratory irritation, such effects may require relatively high concentrations. For instance, a human experimental study involving five volunteers exposed to styrene vapor found that nasal irritation was reported in one subject at 216 ppm after 20 minutes, whereas mild eye and nasal irritation were observed in four of the five subjects at 375 ppm after 15 minutes.³⁶ In addition to these toxicological considerations, styrene had a lower mean concentration and a narrower exposure range compared with the other pollutants in our study, which may have contributed to the nonsignificant associations by limiting exposure contrast and reducing statistical power.

The larger effect estimates observed in small towns or rural areas may partly reflect differences in baseline health status, exposures, and access to care. According to the Centers for Disease Control and Prevention (CDC), rural populations in the US have higher COPD prevalence, with some estimates indicating approximately twice the prevalence compared with urban populations.³⁷ Higher smoking rates and increased exposure to respiratory irritants, such as agricultural dust and wood smoke, may further contribute to respiratory vulnerability among rural residents. These factors may increase susceptibility to the health effects of ambient BTEX, potentially leading to higher hospitalization risk even at comparable exposure levels.³⁸ In addition, disparities in healthcare access, including fewer primary care providers and greater travel distances to hospitals, may delay diagnosis and treatment, increasing the likelihood that acute respiratory episodes progress to severity requiring hospitalization. Prior research has shown that rural patients with COPD experience worse outcomes and higher mortality rates compared with urban patients, partly due to these healthcare barriers.³⁹ Further research is warranted to clarify how underlying health and socioeconomic disparities influence susceptibility to ambient BTEX exposure, and to inform strategies aimed at reducing urban-rural disparities in health outcomes.

This study has several strengths. First, it is, to our knowledge, the first large-scale US epidemiological study to evaluate ambient SBTEX exposure in relation to hospitalization risk. Second, the use of Medicare fee-for-service data provided near complete capture of hospitalizations among older adults in the region, supporting generalizability within this population. Third, high-resolution modeled SBTEX estimates enabled region-wide exposure assessments beyond the limited spatial coverage of monitoring networks. Finally, the case-crossover design inherently controlled for time-invariant individual-level confounders such as diet or smoking status whthin the referent window, which are difficult to measure in claims-based data.⁴⁰ Although uncertainty remains, this design may also reduce confounding by workplace-related exposures if such exposures are approximately constant over the referent window.

Several limitations should be considered. We assigned outdoor ambient SBTEX exposure and did not account for indoor concentrations, which can differ because of indoor sources (e.g., paints, adhesives, cleaning agents), ventilation, air filtration, potentially introducing exposure misclassification. Similarly, occupational exposures were not explicitly controlled for. Although the case-crossover design controls for time-invariant factors, it may not capture short-term occupational exposure spikes occurring within a few hours or days. Furthermore, we cannot entirely rule out residual confounding by unmeasured area-level characteristics, particularly if they exhibit short-term variation and are correlated with ambient SBTEX exposures.

The performance of the SBTEX prediction models is another limitation. The adjusted SBTEX exposure predictions achieved a daily correlation of R = 0.65 (seasonal average R = 0.75) and a normalized mean bias of −5.6%, indicating that the exposure model is not free from error. Although this level of performance is generally acceptable, it may still lead to exposure measurement error and attenuate the estimated associations toward the null.¹³ Moreover, predictions at very low concentrations may be less reliable where data are sparse. The generalizability of our findings should also be considered with caution. Our analysis included Medicare beneficiaries aged 65 years and older residing in the US Gulf region, a population previously identified as relatively vulnerable to air pollution and living in an area characterized by substantial industrial activity. Future research is needed to investigate whether these findings apply to the general population or to other geographic regions.

Because SBTEX compounds were highly correlated, we did not fit multi-pollutant models; therefore, estimates likely reflect combined mixture effects rather than effects of individual compounds. More advanced mixture modeling approaches may help disentangle contributions of correlated pollutants. Future work incorporating additional co-pollutants (e.g., NO₂, SO₂), addressing potential area-level confounding, and extending the study period beyond 2016 would strengthen the evidence base.

In conclusion, this large population-based study found significant associations between ambient BTEX exposure and an increased risk of hospitalization for cardiovascular and respiratory diseases among older adults in the US Gulf region. These findings suggest that even relatively low ambient BTEX concentrations may have measurable adverse health consequences and support targeted public health and regulatory efforts to reduce exposure, particularly for vulnerable populations.

Highlights.

• Health effects of ambient SBTEX (styrene, benzene, toluene, ethylbenzene, xylenes) remain understudied.

• We linked daily SBTEX levels to hospitalizations among older adults in the U.S. Gulf region.

• Short-term BTEX exposures were associated with increased risks of cardiovascular and respiratory hospitalizations.

• Even low-level ambient BTEX exposures may affect health of older adults.

Data Availability:

The Medicare claims data analyzed in this study are available from the Centers for Medicare & Medicaid Services (CMS) under a data-use agreement and cannot be shared by the authors. Access can be requested through the CMS Research Data Assistance Center (ResDAC; https://resdac.org) and is subject to CMS review and approval. A subset of the SBTEX exposure dataset is publicly available at Zenodo (https://doi.org/10.5281/zenodo.7967541). The complete SBTEX dataset is available from the corresponding author upon reasonable request and with permission of the data owners.