Abstract
Background
Industrial facilities are not located uniformly across communities in the United States, but how the burden of exposure to carcinogenic air emissions may vary across population characteristics is unclear. We evaluated differences in carcinogenic industrial pollution among major sociodemographic groups in the United States and Puerto Rico.
Methods
We evaluated cross-sectional associations of population characteristics including race and ethnicity, educational attainment, and poverty at the census tract level with point-source industrial emissions of 21 known human carcinogens using regulatory data from the US Environmental Protection Agency. Odds ratios and 95% confidence intervals comparing the highest emissions (tertile or quintile) to the referent group (zero emissions [ie, nonexposed]) for all sociodemographic characteristics were estimated using multinomial, population density–adjusted logistic regression models.
Results
In 2018, approximately 7.4 million people lived in census tracts with nearly 12 million pounds of carcinogenic air releases. The odds of tracts having the greatest burden of benzene, 1,3-butadiene, ethylene oxide, formaldehyde, trichloroethylene, and nickel emissions compared with nonexposed were 10%-20% higher for African American populations, whereas White populations were up to 18% less likely to live in tracts with the highest emissions. Among Hispanic and Latino populations, odds were 16%-21% higher for benzene, 1,3-butadiene, and ethylene oxide. Populations experiencing poverty or with less than high school education were associated with up to 51% higher burden, irrespective of race and ethnicity.
Conclusions
Carcinogenic industrial emissions disproportionately impact African American and Hispanic and Latino populations and people with limited education or experiencing poverty thus representing a source of pollution that may contribute to observed cancer disparities.
Cancer disparities are well documented among African American, American Indian, Alaska Native, Asian, Pacific Islander, and Hispanic and Latino population groups, and those with low socioeconomic status (SES; ie, below the poverty line) (1,2). In the United States, people from these groups experience a disproportionate burden of exposure to environmental pollution, including criteria (eg, particulate matter, ozone) and hazardous (eg, acetaldehyde, formaldehyde) air pollutants from traffic, industry, and other sources (3-10). Industrial facilities are known sources of carcinogenic air pollutants, and similar patterns of unequal exposure burden have been shown among populations that live in proximity to these facilities (11-18), but few studies have evaluated specific chemicals emitted from industrial sources. Numerous national studies have illustrated racial, ethnic, and socioeconomic inequities in exposure to industrial pollution generally (19-24), most often using an aggregated air pollution estimate of burden, toxicity, or risk to reflect exposure. Although informative, these studies lack specificity and do not provide information on exposure patterns for specific carcinogenic chemicals emitted from industrial sources that may be associated with cancer risk, limiting their value for etiologic or translational research.
Small-area studies in California (25,26), Louisiana (27), Maryland (28), Missouri (29), Tennessee (30), and Texas (31,32), however, have examined population-level inequities in carcinogenic industrial pollution. These studies have shown disproportionate exposure among populations within urban areas and specific regions, including increased exposure to benzene and formaldehyde emissions among Black residents, and lower exposures to benzene, formaldehyde, and 1,3-butadiene exposures with increasing median household income (27). In Texas, the percentage of Hispanic populations in a census block group or tract was associated with higher estimated exposure to the total sum of carcinogenic air pollutants (31,32). However, it is unclear if these chemical-specific inequities persist across the United States. Nationally, little is known about the potential exposure to specific carcinogenic industrial air pollutants among the general population.
Industrial point source emissions may pose considerable environmental and public health burdens, especially when considering the cumulative impacts that some population groups may experience compared with others. Environmental racism is a recognized driver of inequality in the burden of environmental exposures to hazards, including industrial pollution and waste (33,34). A characterization of geographic and population patterns in exposure to carcinogenic industrial air emissions is a starting point for understanding their role in observed disparities in cancer outcomes. To address this critical research gap, we used a US regulatory database and data from the US census to estimate differences in census tract–level population exposure to emissions of carcinogens (eg, benzene, formaldehyde) released to air from industrial facilities in all 50 US states and Puerto Rico. We also examined emissions in the 4 other US territories (American Samoa, Guam, Northern Mariana Islands, and the US Virgin Islands) with permanent population.
Methods
Carcinogenic industrial air emissions
Carcinogenic industrial air emissions in US census tracts were based on 2018 data recorded in the US Environmental Protection Agency (EPA) Toxics Release Inventory (TRI) program (35). Briefly, the TRI program requires industrial facilities to annually report data on quantities and mode of chemical emissions if they meet certain regulatory requirements (eg, exceeding the minimum amount of allowed emissions; Supplementary Methods, available online). Of the 29 chemicals classified as carcinogenic to humans (group 1; Supplementary Table 1, available online) by the International Agency for Research on Cancer (36) that are tracked by the TRI program, 8 had no air emissions in 2018. Therefore, we identified air emissions (in pounds) for the 21 chemicals remaining. Bis(chloromethyl)ether, 4-aminobiphenyl, benzidine, and 2-Naphthylamine were emitted in less than 5 census tracts and therefore were excluded from carcinogen-specific analyses of population characteristics. We calculated the inverse distance-weighted sum of all the emissions within the tract for each carcinogen (Supplementary Methods, available online), weighting each facility’s emissions by the linear distance between the facility and the population-based tract centroid. We evaluated 2 metrics of burden per tract: 1) chemical-specific inverse distance-weighted sums of air emissions and 2) total sum of inverse distance-weighted air emissions from all known carcinogens. To make comparisons after accounting for differences in toxicity among different chemicals, we obtained toxicity-weighted concentrations from the EPA’s Risk-Screening Environmental Indicators (RSEI) model (Supplementary Methods, available online).
Population characteristics
Population data were obtained from the 2010 decennial US census and the 5-year 2006-2010 American Community Survey to characterize area-level sociodemographic characteristics (37). Tract-level characteristics included the percentage of population by self-reported race and ethnicity, educational attainment less than high school, unemployment, renter-occupied housing, and families below the poverty line, as well as median family income. We included tract-level persistent poverty and the Yost index (38,39), a composite of socioeconomic factors. Within each race and ethnicity group, we also assessed the percentage of adults with less than a high school education and families below the poverty line. We defined tracts as urban, suburban, or rural using the rural-urban commuting area codes from the 2010 US census (40) (Supplementary Methods, available online). We did not disaggregate population groups by ancestry because it is unclear that a nationwide analysis is ideal to describe inequities among groups that individually account for less than 10% of the US population.
Statistical analysis
We excluded 243 (0.3%) tracts that had zero population in the 2010 census (eg, airports, military installations). For all other tracts (n = 73 426), we estimated the air emissions for each carcinogen and determined the total carcinogenic air emissions by summing across all 21 chemicals. We excluded the 4 territories that do not have population data from subsequent analyses.
For the 9 carcinogens that were emitted in more than 5 but less than 50 tracts, the dependent variable for models was parameterized dichotomously to evaluate the presence or absence of carcinogen-specific emissions in the tract. For carcinogens emitted in a minimum of 50 tracts, we evaluated carcinogen-specific inverse distance-weighted emissions per tract (Supplementary Methods, available online). We categorized the pounds of inverse distance-weighted air emissions for the highly prevalent chemicals benzene, formaldehyde, and nickel (500 or more tracts with facilities) into their respective quintiles based on their distributions among all tracts with emissions of that carcinogen. We likewise created quintiles for the total sum of emissions based on the distribution among all tracts with emissions. We categorized arsenic, 1,3-butadiene, beryllium, ethylene oxide, and trichloroethylene (less than 200 tracts with facilities) into tertiles.
We used multinomial, population density–adjusted logistic regression models with robust variance estimation to estimate odds ratios (ORs) and 95% confidence intervals (CIs) comparing either the presence of emissions or the highest category of emissions (tertiles: T3 or quintiles: Q5) to the referent group of zero emissions (dependent variable), for continuous population characteristics (10%, $10 000, or 10-unit increases; Supplementary Methods, available online), and persistent poverty (categorical). For comparison with associations with inverse distance-weighted metrics, we also modeled RSEI concentrations (Supplementary Methods, available online). To evaluate if population patterns differed in urban and rural areas, we tested for interaction by urbanicity (Supplementary Methods, available online). All P values and 95% confidence intervals were 2-sided. Analyses were conducted using SAS version 9.4 and STATA/SE version 16.0.
Results
A total of 2196 facilities reported releases of 11 721 590 pounds of carcinogenic air emissions in 2018 and were located in 1763 (2.4%) tracts with an estimated population of 7 442 197 (Figure 1). Distributions of carcinogenic air emissions are presented in Supplementary Figure 1 and Supplementary Table 2 (available online). Carcinogens with the highest air emissions included formaldehyde, benzene, trichloroethylene, 1,3-butadiene, and vinyl chlorine (Table 1). When the carcinogens were ranked by their toxicity-weighted concentrations, the top 5 carcinogens were similar (eg, formaldehyde, benzene, and 1,3-butadiene) except for ethylene oxide and nickel, which had the 2 highest toxicity-weighted concentrations. Industrial facilities in the permanently inhabited island areas reported air releases of several carcinogenic compounds, including benzene (6594 pounds) and ethylene oxide (580 pounds) in Puerto Rico (Supplementary Results, available online).
Figure 1.
Locations of industrial facilities reporting air emissions of 21 chemicals classified as carcinogens by the International Agency for Research on Cancer across the conterminous United States, Alaska, Hawaii, and Puerto Rico by quintiles of pounds of air emissions per census tract.
Table 1.
Summary of carcinogenic air emissions and affected population in the conterminous United States, Alaska, Hawaii, and Puerto Rico for 21 known carcinogens
| Chemical | No. census tracts | % Urban or suburban | Population | Air releases,a pounds | Toxicity-weighted concentrations,b µg/m3 |
|---|---|---|---|---|---|
| Formaldehyde | 578 | 89 | 2 481 998 | 4 628 367 | 453 175 |
| Benzene | 524 | 93 | 2 086 814 | 3 011 602 | 678 623 |
| Trichloroethylene | 121 | 95 | 531 979 | 1 934 609 | 407 214 |
| 1,3-Butadiene | 138 | 95 | 574 015 | 1 079 350 | 1 007 273 |
| Vinyl chloride | 26 | 88 | 103 058 | 549 035 | 76 426 |
| Ethylene oxide | 84 | 100 | 386 957 | 245 989 | 9 749 240 |
| Nickel compounds | 703 | 93 | 2 923 178 | 215 469 | 1 282 977 |
| Arsenic and inorganic arsenic compounds | 165 | 85 | 686 957 | 32 527 | 172 035 |
| 1,2-Dichloropropane | 9 | 100 | 46 179 | 16 725 | 497.23 |
| ortho-Toluidine | 8 | 100 | 41 819 | 5541 | 22 312 |
| Beryllium and beryllium compounds | 52 | 86 | 222 698 | 917 | 8209 |
| Bis(chloromethyl)ether | 2 | 100 | 4482 | 663 | 998.1 |
| Cadmium and cadmium compounds | 18 | 88 | 84 467 | 211 | 1525 |
| Polychlorinated biphenyls | 25 | 84 | 111 003 | 181 | 61.2 |
| Pentachlorophenol | 16 | 93 | 61 532 | 178 | 43.2 |
| Asbestos | 12 | 83 | 44 466 | 171 | 26 993 |
| Lindane | 6 | 100 | 24 685 | 23 | 3.3 |
| 4-Aminobiphenyl | 1 | 100 | 2537 | 10 | 0 |
| 4,4′-Methylenebis (2-chloroaniline) | 6 | 83 | 18 468 | 8 | 401.2 |
| Benzidine | 1 | 100 | 2064 | 7 | 891.4 |
| 2-Naphthylamine | 1 | 100 | 2064 | 7 | 0 |
Emissions amounts self-reported by industrial facilities to the US Environmental Protection Agency Toxics Release Inventory Program in 2018.
Toxicity-weighted concentrations of reported emissions derived using the US Environmental Protection Agency Risk-Screening Environmental Indicators model.
Models showed that a 10% increase in the proportion of African American populations was associated with significantly greater odds (ranging from 10% to 20%) of living in a tract with the highest air emissions (T3 or Q5) of 1,3 butadiene, benzene, ethylene oxide, formaldehyde, nickel, and trichloroethylene compared with nonexposed (Table 2). The pattern for 1,3 butadiene, benzene, ethylene oxide, and formaldehyde was inverse for White populations (from -6% to -18%) and inverse for arsenic, benzene, and formaldehyde among Asian populations (−16% to -36% per 1% increase in the Asian population). Increases in Hispanic populations were associated with 16% to 21% higher odds of living in tracts with the highest air emissions of 1,3-butadiene, benzene, and ethylene oxide but with 17% and 22% lower odds for formaldehyde and trichloroethylene, respectively. For most of the carcinogens, except for beryllium, nearly all indicators of SES were associated with burden. The odds of the highest air emissions of these carcinogens were 19%-51% greater for populations with less than a high school education overall, and 19%-39% higher odds were observed for most carcinogens among White populations with less than a high school education. The odds of the highest air emissions were 24%-41% greater for families experiencing poverty overall and, by race and ethnicity, remained 11%-29% greater for White and African American families. Persistent poverty was significantly associated with 167%-228% higher odds of living in tracts with the highest air emissions of 1,3-butadiene, benzene, nickel, and trichloroethylene. Increases in the proportion of Asian families experiencing poverty were associated with high burden of arsenic exposure, as was the proportion of White adults with less than a high school education. Associations using toxicity-weighted emissions were most consistent among African American and White population groups and for some indicators of SES (Supplementary Table 3, available online).
Table 2.
Carcinogen-specific associations of tract-level sociodemographic characteristics with odds a tract having the highesta inverse-distance weighted carcinogenic air emissions compared with zero air emissions (nonexposed)
| Tract characteristics | Arsenic ORT3 (95% CI)b |
1,3-Butadiene ORT3 (95% CI)b | Benzene ORQ5 (95% CI)b |
Beryllium ORT3 (95% CI)b | Ethylene oxide ORT3 (95% CI)b | Formaldehyde ORQ5 (95% CI)b | Nickel ORQ5 (95% CI)b |
Trichloroethylene ORT3 (95% CI)b |
|---|---|---|---|---|---|---|---|---|
| Race and ethnicity, % | ||||||||
| African American | 0.99 (0.83 to 1.18) | 1.18 (1.04 to 1.33) | 1.20 (1.11 to 1.29) | 1.06 (0.78 to 1.45) | 1.20 (1.05 to 1.37) | 1.19 (1.12 to 1.28) | 1.10 (1.03 to 1.17) | 1.12 (1.01 to 1.25) |
| American Indian and Alaska Native | 0.97 (0.92 to 1.02)c | 0.96 (0.87 to 1.06)c | 0.99 (0.97 to 1.01)c | 0.94 (0.80 to 1.11)c | 0.71 (0.48 to 1.05)c | 0.99 (0.98 to 1.00)c | 0.97 (0.94 to 1.01)c | 0.76 (0.54 to 1.07)c |
| Asian | 0.64 (0.41 to 0.99)c | 0.86 (0.70 to 1.06)c | 0.70 (0.56 to 0.89)c | 0.83 (0.43 to 1.59)c | 0.86 (0.72 to 1.03)c | 0.84 (0.73 to 0.97)c | 0.65 (0.36 to 1.19) | 0.89 (0.53 to 1.49) |
| Hispanic | 0.96 (0.78 to 1.19) | 1.21 (1.10 to 1.32) | 1.16 (1.09 to 1.24) | 0.70 (0.42 to 1.17) | 1.18 (1.06 to 1.32) | 0.83 (0.72 to 0.96) | 0.97 (0.88 to 1.07) | 0.78 (0.63 to 0.97) |
| Multiracial | 0.96 (0.85 to 1.10)c | 0.65 (0.46 to 0.93)c | 0.99 (0.89 to 1.09) | 0.90 (0.68 to 1.19)c | 1.02 (0.97 to 1.07)c | 1.03 (1.01 to 1.05)c | 1.05 (1.02 to 1.08)c | 1.01 (0.94 to 1.09) |
| Native Hawaiian and Other Pacific Islander | 0.98 (0.88 to 1.09)c | — | 0.95 (0.74 to 1.21)c | — | 0.01 (0.001 to 1.69)c | 0.85 (0.50 to 1.42)c | 0.92 (0.70 to 1.22)c | 0.30 (0.03 to 2.31)c |
| Other | 0.96 (0.52 to 1.75)c | 0.04 (0.01 to 0.63)c | 0.73 (0.25 to 2.12)c | — | 0.10 (0.002 to 4.75)c | 1.04 (0.59 to 1.82)c | 0.91 (0.43 to 1.85)c | 1.06 (0.83 to 1.35)c |
| White | 1.05 (0.91 to 1.20) | 0.82 (0.75 to 0.90) | 0.83 (0.78 to 0.88) | 1.09 (0.80 to 1.49) | 0.82 (0.74 to 0.90) | 0.94 (0.88 to 0.99) | 0.95 (0.89 to 1.01) | 0.99 (0.89 to 1.11) |
| Socioeconomic status, % | ||||||||
| Less than high school education | 1.19 (1.01 to 1.40) | 1.37 (1.14 to 1.65) | 1.46 (1.31 to 1.64) | 1.17 (0.83 to 1.64) | 1.51 (1.30 to 1.75) | 1.29 (1.16 to 1.43) | 1.30 (1.18 to 1.42) | 1.04 (0.84 to 1.27) |
| Rented housing unitsd | 1.08 (0.90 to 1.31) | 1.18 (1.05 to 1.33) | 1.23 (1.14 to 1.33) | 1.08 (0.77 to 1.53) | 1.24 (1.09 to 1.41) | 1.29 (1.22 to 1.37) | 1.26 (1.17 to 1.36) | 1.30 (1.16 to 1.47) |
| Unemployed | 1.19 (0.78 to 1.81) | 1.35 (1.05 to 1.75) | 1.63 (1.44 to 1.84) | 0.76 (0.30 to 1.93) | 1.46 (1.20 to 1.78) | 1.43 (1.22 to 1.68) | 1.52 (1.33 to 1.75) | 1.25 (0.90 to 1.74) |
| Families below poverty threshold | 1.18 (0.99 to 1.41) | 1.41 (1.24 to 1.61) | 1.41 (1.29 to 1.55) | 1.17 (0.65 to 2.10) | 1.28 (1.10 to 1.49) | 1.35 (1.24 to 1.48) | 1.41 (1.28 to 1.54) | 1.24 (1.04 to 1.47) |
| Persistent poverty | 5.61 (1.93 to 16.24) | 3.25 (1.26 to 8.39) | 3.28 (1.68 to 6.42) | 3.57 (0.46 to 27.5) | 2.02 (0.44 to 9.23) | 2.61 (1.17 to 5.80) | 2.89 (1.67 to 4.99) | 2.67 (1.18 to 6.02) |
| Family income, per $10ke | 0.83 (0.72 to 0.94) | 0.83 (0.74 to 0.93) | 0.75 (0.69 to 0.83) | 0.90 (0.71 to 1.14) | 0.78 (0.68 to 0.89) | 0.77 (0.70 to 0.85) | 0.81 (0.74 to 0.89) | 0.83 (0.75 to 0.93) |
| Yost Index, per 10 unitf | 1.14 (1.03 to 1.26) | 1.14 (1.03 to 1.25) | 1.27 (1.18 to 1.37) | 1.07 (0.89 to 1.29) | 1.27 (1.13 to 1.43) | 1.28 (1.18 to 1.38) | 1.22 (1.15 to 1.30) | 1.15 (1.04 to 1.26) |
| Families below poverty, % | ||||||||
| African American | 1.07 (0.97 to 1.18) | 1.15 (1.06 to 1.25) | 1.11 (1.05 to 1.17) | 1.05 (0.85 to 1.29) | 1.12 (1.03 to 1.22) | 1.12 (1.06 to 1.18) | 1.12 (1.07 to 1.18) | 1.08 (0.99 to 1.18) |
| American Indian and Alaska Native | 0.93 (0.78 to 1.12) | 0.99 (0.85 to 1.14) | 1.06 (0.97 to 1.14) | — | 1.04 (0.89 to 1.22) | 1.05 (0.97 to 1.13) | 1.01 (0.93 to 1.11) | 0.81 (0.59 to 1.10) |
| Asian | 1.17 (1.07 to 1.27) | 0.87 (0.69 to 1.10) | 0.97 (0.84 to 1.13) | — | 1.04 (0.85 to 1.26) | 0.91 (0.75 to 1.10) | 0.96 (0.84 to 1.09) | 1.12 (1.01 to 1.23) |
| Hispanic | 1.02 (0.92 to 1.12) | 0.96 (0.88 to 1.05) | 0.93 (0.87 to 1.01) | 0.92 (0.78 to 1.09) | 0.97 (0.88 to 1.07) | 1.08 (1.02 to 1.15) | 0.99 (0.92 to 1.06) | 0.98 (0.86 to 1.11) |
| Multiracial | 1.00 (0.89 to 1.12) | 1.02 (0.92 to 1.13) | 0.99 (0.91 to 1.08) | 1.06 (0.90 to 1.26) | 0.97 (0.83 to 1.14) | 1.06 (0.99 to 1.13) | 1.06 (1.00 to 1.13) | 1.06 (0.95 to 1.17) |
| Native Hawaiian and Other Pacific Islander | 1.16 (0.94 to 1.42) | — | 1.08 (0.87 to 1.33) | — | — | 0.96 (0.77 to 1.19) | — | — |
| Other | 0.99 (0.89 to 1.11) | 1.00 (0.88 to 1.13) | 1.04 (0.97 to 1.11) | 0.88 (0.69 to 1.12) | 1.02 (0.93 to 1.12) | 1.03 (0.96 to 1.11) | 1.01 (0.94 to 1.08) | 0.90 (0.77 to 1.05) |
| White | 1.23 (1.08 to 1.40) | 1.29 (1.13 to 1.47) | 1.25 (1.15 to 1.37) | 1.03 (0.58 to 1.84) | 1.15 (0.97 to 1.37) | 1.26 (1.18 to 1.35) | 1.23 (1.15 to 1.32) | 1.15 (0.95 to 1.40) |
| Less than high school education, % | ||||||||
| African American | 1.01 (0.91 to 1.11) | 1.07 (0.96 to 1.20) | 1.09 (1.02 to 1.16) | 1.07 (0.92 to 1.25) | 1.08 (0.95 to 1.23) | 1.07 (1.01 to 1.14) | 1.09 (1.03 to 1.15) | 0.98 (0.87 to 1.11) |
| American Indian and Alaska Native | 1.00 (0.90 to 1.11) | 1.00 (0.89 to 1.12) | 1.02 (0.95 to 1.09) | 0.84 (0.57 to 1.23) | 1.05 (0.93 to 1.20) | 1.03 (0.97 to 1.10) | 1.03 (0.96 to 1.09) | 1.05 (0.94 to 1.17) |
| Asian | 0.97 (0.84 to 1.12) | 0.95 (0.80 to 1.12) | 0.92 (0.81 to 1.05) | 0.83 (0.62 to 1.11) | 1.10 (0.97 to 1.26) | 0.97 (0.88 to 1.07) | 0.83 (0.73 to 0.95) | 1.10 (0.99 to 1.22) |
| Hispanic | 0.96 (0.88 to 1.06) | 1.04 (0.96 to 1.13) | 1.01 (0.95 to 1.07) | 0.97 (0.83 to 1.14) | 1.05 (0.95 to 1.16) | 1.04 (0.98 to 1.11) | 1.02 (0.95 to 1.08) | 0.99 (0.89 to 1.10) |
| Multiracial | 0.96 (0.86 to 1.08) | 1.03 (0.92 to 1.16) | 1.01 (0.93 to 1.09) | 0.92 (0.74 to 1.13) | 1.09 (0.98 to 1.22) | 1.02 (0.94 to 1.09) | 0.93 (0.86 to 1.01) | 1.07 (0.97 to 1.17) |
| Native Hawaiian and Other Pacific Islander | 1.13 (0.97 to 1.32) | — | 0.75 (0.53 to 1.06) | — | — | 0.82 (0.61 to 1.10) | 0.95 (0.74 to 1.22) | — |
| Other | 0.97 (0.89 to 1.05) | 1.10 (1.02 to 1.17) | 1.05 (1.00 to 1.11) | 0.93 (0.81 to 1.07) | 1.06 (0.97 to 1.15) | 1.02 (0.96 to 1.07) | 0.98 (0.93 to 1.04) | 0.94 (0.83 to 1.05) |
| White | 1.39 (1.23 to 1.56) | 1.19 (1.04 to 1.37) | 1.29 (1.19 to 1.41) | 1.31 (1.03 to 1.67) | 1.36 (1.23 to 1.50) | 1.30 (1.21 to 1.40) | 1.30 (1.22 to 1.40) | 0.90 (0.72 to 1.14) |
For chemicals emitted in more than 50 but less than 200 tracts (arsenic, 1,3-butadiene, beryllium, ethylene oxide, and trichloroethylene), the distribution was categorized into tertiles (T), with estimates for the highest tertile (T3) compared with nonexposed presented. For chemicals emitted in more than 200 tracts (benzene, formaldehyde, and nickel), the distribution was categorized into quintiles, with estimates for the highest quintile (Q5) compared with nonexposed presented. “—” indicate odds ratios not presented due to unstable estimate. CI = confidence interval; OR = odds ratio.
Odds ratios are adjusted for population density (population per km2 land area of the tract) and reflect a 10% increase in the characteristic of the tract population (ie, a 10% increase in the percentage of the tract population with less than a high school education) unless otherwise noted; estimates for persistent poverty compare the odds of being in the highest emissions category among tracts experiencing persistent poverty with tracts that do not.
Odds ratios for 1% increase in population characteristic substituted when 10% increase was not able to be estimated.
Percentage of rented housing units missing for 214 tracts.
Median family income missing for 493 tracts.
Yost Index missing for 1469 tracts.
For the 9 carcinogens with emissions in less than 50 tracts, patterns generally showed higher emissions burdens for African American, Hispanic, and low SES populations and lower emissions burdens for White and Asian populations, but most associations were not statistically significant (Supplementary Results and Supplementary Table 4, available online ).
Models evaluating the total sum of emissions yielded patterns of association similar to those for individual carcinogens (Table 3). For example, a 10% increase in the proportion of African American populations in the tract was associated with 18% higher odds of living in a tract with the highest air emissions compared with nonexposed, whereas the pattern was inverse for White and Asian populations (11% and 62% lower odds, respectively). Low educational attainment was associated with greater odds of having the highest air emissions, overall and by race and ethnicity, mainly among White, African American, and Hispanic populations. The same was true for families experiencing poverty, mainly for White (ORQ5 vs nonexposed = 1.22, 95% CI = 1.15 to 1.28) and African American (ORQ5 vs nonexposed = 1.10, 95% CI = 1.06 to 1.13) families. Persistent poverty within a tract was associated with greater odds of having the highest air emissions (ORQ5 vs nonexposed = 2.42, 95% CI = 1.61 to 3.65). Patterns of association did not substantively change when race and ethnicity models were adjusted for SES, and vice versa, except that the association with persistent poverty was attenuated and no longer statistically significant.
Table 3.
Associations of tract-level sociodemographic characteristics with odds a tract having the highest quintile of total sum of inverse-distance weighted carcinogenic air emissions (quintile 5) compared with zero air emissions (nonexposed)
| Tract characteristics | Inverse-distance weighted sum of air emissions ORQ5 vs nonexposed (95% CI)a |
|---|---|
| Race and ethnicity, % | |
| African American | 1.18 (1.13 to 1.23) |
| American Indian and Alaska Native | 0.86 (0.74 to 0.99) |
| Asian | 0.38 (0.19 to 0.74) |
| Hispanic | 1.04 (0.98 to 1.09) |
| Multiracial | 1.19 (0.81 to 1.74) |
| Native Hawaiian and Other Pacific Islander | 0.83 (0.55 to 1.24)b |
| Other | 0.87 (0.37 to 2.02)b |
| White | 0.89 (0.86 to 0.92) |
| Socioeconomic status, % | |
| Less than high school education | 1.34 (1.26 to 1.43) |
| Rented housing unitsc | 1.24 (1.18 to 1.29) |
| Unemployed | 1.49 (1.34 to 1.65) |
| Families below poverty threshold | 1.34 (1.27 to 1.42) |
| Persistent poverty | 2.42 (1.61 to 3.65) |
| Family income, per $10kd | 0.79 (0.75 to 0.83) |
| Yost Index, per 10 unite | 1.23 (1.18 to 1.28) |
| Families below poverty, % | |
| African American | 1.10 (1.06 to 1.13) |
| American Indian and Alaska Native | 1.03 (0.98 to 1.08) |
| Asian | 1.00 (0.94 to 1.07) |
| Hispanic | 1.01 (0.97 to 1.05) |
| Multiracial | 1.03 (0.99 to 1.07) |
| Native Hawaiian and Other Pacific Islander | 0.98 (0.82 to 1.15) |
| Other | 1.02 (0.97 to 1.06) |
| White | 1.22 (1.15 to 1.28) |
| Less than high school education, % | |
| African American | 1.07 (1.03 to 1.11) |
| American Indian and Alaska Native | 1.01 (0.97 to 1.06) |
| Asian | 0.99 (0.94 to 1.05) |
| Hispanic | 1.04 (1.01 to 1.08) |
| Multiracial | 1.02 (0.98 to 1.06) |
| Native Hawaiian and Other Pacific Islander | 0.73 (0.56 to 0.96) |
| Other | 1.03 (1.00 to 1.06) |
| White | 1.26 (1.20 to 1.32) |
Odds ratios are adjusted for population density (population per km2 land area of the tract) and reflect a 10% increase in the characteristic of the tract population (ie, a 10% increase in the percentage of the tract population with less than a high school education) unless otherwise noted; estimates for persistent poverty compare the odds of being in the highest emissions category among tracts experiencing persistent poverty to tracts that do not. CI = confidence interval; OR = odds ratio.
Odds ratios for 1% increase in population characteristic substituted when 10% increase was not able to be estimated.
Percentage of rented housing units missing for 214 tracts (nonexposed n = 218; Q 1 n = 4; Q2 n = 3; Q3 n = 9; Q4 n = 4; Q5 n = 3).
Median family income missing for 493 tracts (nonexposed n = 468; Q1 n = 3; Q2 n = 4; Q3 n = 10; Q4 n = 6; Q5 n = 2).
Yost Index missing for 1469 tracts (nonexposed n = 1412; Q1 n = 10; Q2 n = 13; Q3 n = 17; Q4 n = 10; Q5 n = 7).
More than 90% of tracts in the highest quintile of air emissions were in urban or suburban areas. Patterns of association limited to urban areas were like the nonstratified results (Supplementary Results and Supplementary Table 5, available online).
Discussion
Our novel evaluation showed that there are millions of pounds of carcinogens emitted to air from industrial sources across the United States and US territories and that notable population differences are apparent between those who do and do not live near high levels of these emissions. Prior to this work, there had been limited characterization of population patterns of exposure to emissions of specific carcinogenic industrial air pollutants on a national scale. Specifically, our findings illustrate inequitable emissions among African American populations that are independent of low educational attainment or family poverty, including in rural areas. In contrast, it wasn’t until we accounted for educational attainment and poverty among White populations that we saw an increase in emissions burden in this group. Although we did not observe a statistically significant overall association with the sum of carcinogenic air emissions for Hispanic populations, when we stratified by urbanicity, the proportion of Hispanic populations in urban areas was associated with higher emissions burden. Importantly, SES was a critical predictor of exposure burden; multiple indicators of low SES were associated with high emissions.
Our investigation extends a limited body of previous work on a national scale (19-24), none of which evaluated specific carcinogens. One study that used census and TRI data from 1990 showed that African American, Asian and Pacific Islander, Hispanic, and other racial groups tended to live in counties with higher industrial emissions amounts relative to counties where White populations lived (18). In our analysis, Asian populations had lower emissions burden, consistent with another national study of 2010 census data that found tract-level Asian populations had one of the lower estimated cancer risk scores from hazardous air pollution from point sources but did not evaluate SES (41). Other studies of Asian populations have found increased exposure to particulate matter in metropolitan areas (42) and hazardous air pollutants from a mix of sources, including industry, both nationally (10) and in 4 US cities (43). However, our findings highlight the importance of simultaneously accounting for measures of SES; we no longer observed reduced odds of exposure among the Asian population with low SES in our analysis.
We evaluated emissions burden for 21 individual carcinogens, with emphasis on 8 carcinogens emitted in at least 50 tracts. Comparable investigation of specific carcinogenic exposure patterns has only been evaluated in small-area studies, like those conducted in California (25), Louisiana (27), Maryland (28), Missouri (29), Tennessee (30), and Texas (31,32). Similar to studies in Louisiana (27) and Tennessee (30), we observed disproportionate amounts of benzene and formaldehyde emissions among African American populations and benzene, formaldehyde, and 1,3-butadiene emissions among low SES populations, showing that disparities in these emissions are not limited to the smaller geographic areas included in previous studies. In our study, population-level indicators of low SES were associated with higher airborne emissions of benzene, formaldehyde, 1,3-butadiene, and nickel. For these carcinogens, low educational attainment and family poverty among White populations were associated a with greater emissions burden compared with being a White individual overall.
Our evaluation contributes a needed national characterization of levels of carcinogenic air emissions that has largely been lacking from the literature and is useful to guide etiologic investigation of cancer associations with specific chemicals in the environment, the study of which remains limited for many carcinogens. We acknowledge that the emissions are self-reported by each facility, may be subject to errors, and are difficult to verify, however, for some chemicals the TRI is the only routinely collected estimate of environmental emissions. These data have been leveraged in several ecologic and cohort studies (44-46) including a prospective cohort in multiple US states that observed an increased risk of incident intraductal breast cancer in association with relatively high estimated exposure to ethylene oxide emissions near the home (47). In 2018, regulated industries reported emissions of more than 1 million pounds each of benzene, formaldehyde, butadiene, and trichloroethylene, but we know very little about exposure levels and related cancer risks among the general US population. Identification of specific carcinogens with unequal exposure potential is informative not only to community members and policy makers but also to researchers planning environmental studies of cancer risk and disparities.
Major strengths of our study include use of emissions data for the United States, including the territories, and of census data that account for the diversity of the US population. To our knowledge, this is the first characterization of carcinogenic air emissions on the island territories. Our distance- and population-weighted emissions metrics provided a more detailed quantitative characterization than the simpler metrics (eg, counts of emission sources) that have been used in other national studies. Often, census tract–level estimates of environmental exposure rely on the geometric center of the tract, which assumes population is evenly distributed throughout the tract. Instead, we used the population-weighted centroid, which better reflects the actual areas where population reside and their exposure potential. Estimating exposure burden for these point source carcinogenic emissions is challenging given the limited measurement of these chemicals in the ambient environment. Studies in industrial areas in New York State (48) and Houston, Texas, have demonstrated associations between ambient benzene concentrations and industrial sources (49). In a monitoring study of Ohio homes within 2 km of a natural gas compressor station, indoor benzene levels were 2-17 times greater than the state’s indoor standard (50). Nonsmokers living approximately 0.8 km from an ethylene oxide–emitting facility had statistically significantly higher levels of hemoglobin adduct levels compared with those farther away, and median levels among all residents of the general area were higher than levels measured in the US general population (51). However, we acknowledge that residing near an emitting facility does not necessarily confer exposure to the carcinogen, and likewise, emissions cross administrative boundaries like census tracts. Our exposure metrics are also different from studies that have used toxicity concentrations or cancer risk scores, such as data from the EPA’s National Air Toxics Assessment or RSEI models that are based on TRI emissions. Although these datasets are created using sophisticated methods that incorporate important geospatial features and other factors influencing environmental fate and transport (eg, meteorological conditions), we did not use RSEI estimates in our main analyses because they are modeled using distances out to 50 km and subsequently would define large proportions of the US population as exposed at very low concentrations. In contrast, our goal was to characterize population patterns among highly exposed groups. We therefore limited our definition of exposure to populations within the tract where a facility is located to conservatively include populations with the greatest potential for exposure. Emissions quantities are one way to describe population burden, however, we acknowledge that each chemical has a different toxicity and that the toxicity may be relevant to subsequent cancer risk.
Although we present data from all racial and ethnic population groups in the 2010 census, there were small numbers for some (eg, American Indian, Alaska Native, Native Hawaiian) that sometimes resulted in wide confidence intervals and limited our ability to evaluate associations by educational attainment or family poverty. We presented results from models evaluating each population characteristic individually, but estimates did not substantively change when racial and ethnic characteristics were co-adjusted for SES or vice versa. We did not disaggregate population groups by ancestry. To evaluate patterns of emissions among disaggregated groups, a different approach would be better. For example, comparisons could be made of geographically distinct areas with and without relatively high proportions of residents of similar racial and ethnic ancestry. Additionally, we acknowledge that associations for Hispanics may have been underestimated because of their known underrepresentation in the US census (52). We used available data on persistent poverty but acknowledge the challenges of creating this construct for census tracts that may change boundaries over time. The placement of facilities that emit the carcinogens in our study are related to local land use and zoning rules, and we did not explore this in our analyses.
In this assessment, we demonstrated inequities in the burden of airborne carcinogenic emissions across the general population. These results highlight the influence of socioeconomic factors on living near industrial sources with the highest levels of carcinogenic emissions and underscore the importance of taking an intersectional approach that evaluates population characteristics jointly to elucidate environmental exposure disparities. Given ongoing concerns about socio-environmental drivers of health inequities, this work offers a timely characterization and is informative for environmental exposure assessments and studies of cancer risk.
Supplementary Material
Acknowledgements
This work was supported in part by the Intramural Research Program of the National Cancer Institute (NCI). Preliminary findings from this work were presented at the International Society for Environmental Epidemiology conference in 2022. The views expressed by the authors are their own and do not necessarily represent the views of the U.S. Department of Health and Human Services, the National Institutes of Health, or the National Cancer Institute. The funder had no involvement in the study design, data collection, analysis, or interpretation, writing, or in the decision to submit the paper for publication; it reviewed and approved the manuscript for submission.
Contributor Information
Jessica M Madrigal, Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Rockville, MD, USA.
Abigail Flory, Westat Inc, Rockville, MD, USA.
Jared A Fisher, Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Rockville, MD, USA.
Elizabeth Sharp, Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Rockville, MD, USA.
Barry I Graubard, Biostatistics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Rockville, MD, USA.
Mary H Ward, Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Rockville, MD, USA.
Rena R Jones, Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Rockville, MD, USA.
Data availability
The data used in this study are publicly available from the US EPA (https://www.epa.gov/toxics-release-inventory-tri-program) and the US Census Bureau (https://www.census.gov/).
Author contributions
Jessica Madrigal, PhD, MS (Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Supervision; Visualization; Writing - original draft; Writing - review & editing), Abigail Flory, MS (Data curation; Methodology; Software; Visualization; Writing—review & editing), Jared Fisher, PhD (Data curation; Investigation; Methodology; Validation; Writing—review & editing), Elizabeth Sharp, BA (Writing—review & editing), Barry Graubard, PhD (Methodology; Writing—review & editing) Mary Ward, PhD (Methodology; Writing—review & editing), and Rena Jones, PhD, MS (Conceptualization; Data curation; Methodology; Funding acquisition; Investigation; Resources; Supervision; Writing—review & editing).
Funding
This work was supported in part by the Intramural Research Program of the National Cancer Institute (NCI) (Project ZIA CP010125).
Conflicts of interest
The authors declare no conflict of interest.
References
- 1. Ward E, Jemal A, Cokkinides V, et al. Cancer disparities by race/ethnicity and socioeconomic status. CA Cancer J Clin. 2004;54(2):78-93. [DOI] [PubMed] [Google Scholar]
- 2. Zavala VA, Bracci PM, Carethers JM, et al. Cancer health disparities in racial/ethnic minorities in the United States. Br J Cancer. 2021;124(2):315-332. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Kodros JK, Bell ML, Dominici F, et al. Unequal airborne exposure to toxic metals associated with race, ethnicity, and segregation in the USA. Nat Commun. 2022;13(1):6329. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Liu J, Clark LP, Bechle MJ, et al. Disparities in air pollution exposure in the United States by race/ethnicity and income, 1990-2010. Environ Health Perspect. 2021;129(12):127005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Bravo MA, Anthopolos R, Bell ML, Miranda ML.. Racial isolation and exposure to airborne particulate matter and ozone in understudied US populations: environmental justice applications of downscaled numerical model output. Environ Int. 2016;92-93:247-255. [DOI] [PubMed] [Google Scholar]
- 6. Alvarez CH, Calasanti A, Evans CR, Ard K.. Intersectional inequalities in industrial air toxics exposure in the United States. Health Place. 2022;77:102886. [DOI] [PubMed] [Google Scholar]
- 7. Mikati I, Benson AF, Luben TJ, et al. Disparities in distribution of particulate matter emission sources by race and poverty status. Am J Public Health. 2018;108(4):480-485. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Tessum CW, Paolella DA, Chambliss SE, et al. PM(2.5) polluters disproportionately and systemically affect people of color in the United States. Sci Adv. 2021;7(18):1-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Wang Y, Liu P, Schwartz J, et al. Disparities in ambient nitrogen dioxide pollution in the United States. Proc Natl Acad Sci USA. 2023;120(16):e2208450120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Grineski SE, Collins TW, Morales DX.. Asian Americans and disproportionate exposure to carcinogenic hazardous air pollutants: a national study. Soc Sci Med. 2017;185:71-80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Perlin SA, Sexton K, Wong DW.. An examination of race and poverty for populations living near industrial sources of air pollution. J Expo Anal Environ Epidemiol. 1999;9(1):29-48. [DOI] [PubMed] [Google Scholar]
- 12. Perlin SA, Wong D, Sexton K.. Residential proximity to industrial sources of air pollution: Interrelationships among race, poverty, and age. J Air Waste Manag Assoc. 2001;51(3):406-421. [DOI] [PubMed] [Google Scholar]
- 13. Mohai P, Lantz PM, Morenoff J, et al. Racial and socioeconomic disparities in residential proximity to polluting industrial facilities: evidence from the Americans’ Changing Lives Study. Am J Public Health. 2009;99(suppl 3):S649-56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Wilson SM, Fraser-Rahim H, Williams E, et al. Assessment of the distribution of toxic release inventory facilities in metropolitan Charleston: an environmental justice case study. Am J Public Health. 2012;102(10):1974-1980. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Wilson S, Zhang H, Jiang C, et al. Being overburdened and medically underserved: assessment of this double disparity for populations in the state of Maryland. Environ Health. 2014;13(1):26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Johnson R, Ramsey-White K, Fuller CH.. Socio-demographic differences in toxic release inventory siting and emissions in metro Atlanta. Int J Environ Res Public Health. 2016;13(8):747. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Neumann CM, Forman DL, Rothlein JE.. Hazard screening of chemical releases and environmental equity analysis of populations proximate to toxic release inventory facilities in Oregon. Environ Health Perspect. 1998;106(4):217-226. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Perlin SA, Setzer RW, Creason J, Sexton K.. Distribution of industrial air emissions by income and race in the United States: an approach using the toxic release inventory. Environ Sci Technol. 1995;29(1):69-80. [DOI] [PubMed] [Google Scholar]
- 19. Ard K. Trends in exposure to industrial air toxins for different racial and socioeconomic groups: a spatial and temporal examination of environmental inequality in the U.S. from 1995 to 2004. Soc Sci Res. 2015;53:375-390. [DOI] [PubMed] [Google Scholar]
- 20. Ash M, Boyce JK, Chang G, Scharber H.. Is environmental justice good for white folks? Industrial air toxics exposure in urban America. Social Science Quarterly. 2013;94(3):616-636. [Google Scholar]
- 21. Ash M, Fetter TR.. Who lives on the wrong side of the environmental tracks? Evidence from the EPA’s risk-screening environmental indicators model. Soc Sci Q. 2004;85(2):441-462. [Google Scholar]
- 22. Zwickl K, Ash M, Boyce JK.. Regional variation in environmental inequality: industrial air toxics exposure in U.S. cities. Ecol Econ. 2014;107:494-509. [Google Scholar]
- 23. Downey L, Hawkins B.. Race, income, and environmental inequality in the United States. Sociol Perspect. 2008;51(4):759-781. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Morello-Frosch R, Jesdale BM.. Separate and unequal: residential segregation and estimated cancer risks associated with ambient air toxics in U.S. metropolitan areas. Environ Health Perspect. 2006;114(3):386-393. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Marshall JD. Environmental inequality: air pollution exposures in California’s South Coast Air Basin. Atmos Environ. 2008;42(21):5499-5503. [Google Scholar]
- 26. Sadd JL, Pastor M, Boer JT, Snyder LD.. “Every breath you take…” : the demographics of toxic air releases in Southern California. Econ Dev Q. 1999;13(2):107-123. [Google Scholar]
- 27. James W, Jia C, Kedia S.. Uneven magnitude of disparities in cancer risks from air toxics. Int J Environ Res Public Health. 2012;9(12):4365-4385. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Apelberg BJ, Buckley TJ, White RH.. Socioeconomic and racial disparities in cancer risk from air toxics in Maryland. Environ Health Perspect. 2005;113(6):693-699. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Abel TD. Skewed riskscapes and environmental injustice: a case study of metropolitan St Louis. Environ Manage. 2008;42(2):232-248. [DOI] [PubMed] [Google Scholar]
- 30. Jia C, James W, Kedia S.. Relationship of racial composition and cancer risks from air toxics exposure in Memphis, Tennessee, U.S.A. Int J Environ Res Public Health. 2014;11(8):7713-7724. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Collins TW, Grineski SE, Chakraborty J, McDonald YJ.. Understanding environmental health inequalities through comparative intracategorical analysis: Racial/ethnic disparities in cancer risks from air toxics in El Paso County, Texas. Health Place. 2011;17(1):335-344. [DOI] [PubMed] [Google Scholar]
- 32. Linder SH, Marko D, Sexton K.. Cumulative cancer risk from air pollution in houston: disparities in risk burden and social disadvantage. Environ Sci Technol. 2008;42(12):4312-4322. [DOI] [PubMed] [Google Scholar]
- 33. Bryant B, Mohai P.. Race and the Incidence of Environmental Hazards: A Time for Discourse. 1st ed.Routledge; 1992. 10.4324/9780429303661 [DOI] [Google Scholar]
- 34. Kaufman JD, Hajat A.. Confronting environmental racism. Environ Health Perspect 2021;129(5):51001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. United States Environmental Protection Agency. Toxics Release Inventory (TRI) 2018 National Analysis. https://www.epa.gov/trinationalanalysis. Accessed October 2, 2022.
- 36. International Agency for Research on Cancer I. Agents Classified by the IARC Monographs. Volumes 1-125. https://monographs.iarc.fr/agents-classified-by-the-iarc/. Accessed April 1, 2021.
- 37. Manson SM, Schroeder J, Van Riper D, et al. IPUMS National Historical Geographic Information System: Version 15.0 [Dataset]. Minneapolis, MN: IPUMS; 2020, https://www.ipums.org/projects/ipums-nhgis/d050.v15.0. Accessed April 1, 2021. [Google Scholar]
- 38. Yost K, Perkins C, Cohen R, et al. Socioeconomic status and breast cancer incidence in California for different race/ethnic groups. Cancer Causes Control. 2001;12(8):703-711. [DOI] [PubMed] [Google Scholar]
- 39. Boscoe FP, Liu B, Lee F.. A comparison of two neighborhood-level socioeconomic indexes in the United States. Spat Spatiotemporal Epidemiol. 2021;37:100412. [DOI] [PubMed] [Google Scholar]
- 40. United States Department of Agriculture. United States Department of Agriculture 2010 Rural-Urban Commuting Area (RUCA) Codes.2010. https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes/. Accessed February 1, 2023.
- 41. Rubio R, Grineski S, Collins T, Morales DX.. Ancestry-based intracategorical injustices in carcinogenic air pollution exposures in the United States. Soc Nat Resour. 2020;33(8):987-1005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Collins TW, Grineski SE, Shaker Y, Mullen CJ.. Communities of color are disproportionately exposed to long-term and short-term PM2.5 in metropolitan America. Environ Res. 2022;214(Pt 4):114038. [DOI] [PubMed] [Google Scholar]
- 43. Grineski S, Morales DX, Collins T, et al. The burden of carcinogenic air toxics among Asian Americans in four US metro areas. Popul Environ. 2019;40(3):257-282. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Bulka C, Nastoupil LJ, Koff JL, et al. Relations between residential proximity to EPA-designated toxic release sites and diffuse large B-cell lymphoma incidence. South Med J. 2016;109(10):606-614. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Bulka C, Nastoupil LJ, McClellan W, et al. Residence proximity to benzene release sites is associated with increased incidence of non-Hodgkin lymphoma. Cancer. 2013;119(18):3309-3317. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Williams SB, Shan Y, Jazzar U, et al. Proximity to oil refineries and risk of cancer: a population-based analysis. JNCI Cancer Spectr. 2020;4(6):pkaa088. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Jones RR, Fisher JA, Medgyesi DN, et al. Ethylene oxide emissions and incident breast cancer and non-Hodgkin lymphoma in a U.S. cohort. J Natl Cancer Inst. 2023;115(4):405-412. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Aleksic N, Boynton G, Sistla G, Perry J.. Concentrations and trends of benzene in ambient air over New York State during 1990–2003. Atmos Environ. 2005;39(40):7894-7905. [Google Scholar]
- 49. Gilman JB, Kuster WC, Goldan PD, et al. Measurements of volatile organic compounds during the 2006 TexAQS/GoMACCS campaign: industrial influences, regional characteristics, and diurnal dependencies of the OH reactivity. J Geophys Res. 2009;114(D7):D00F06-n/a. [Google Scholar]
- 50. Vollet Martin KA, Lin EZ, Hilbert TJ, et al. Survey of airborne organic compounds in residential communities near a natural gas compressor station: Response to community concern. Environ Adv. 2021;5:100076. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51. Szwiec E, Friedman L, Buchanan S.. Levels of ethylene oxide biomarker in an exposed residential community. Int J Environ Res Public Health. 2020;17(22):1-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52. O’Hare WP. Census coverage of the Hispanic population. In: O’Hare WP, ed. Differential Undercounts in the U.S. Census: Who is Missed? Cham: Springer International Publishing; 2019:71-82. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data used in this study are publicly available from the US EPA (https://www.epa.gov/toxics-release-inventory-tri-program) and the US Census Bureau (https://www.census.gov/).