Impact of air pollution and population vulnerability on cancer mortality in the United States: A scoping review

Abstract

Introduction:

Ambient air pollution is an important contributor to cancer risk. While previous studies have examined the impact of air pollution on cancer mortality, they have largely overlooked how population vulnerability indicators, such as socioeconomic status, interact with air pollution to influence cancer mortality across cancer types. This review aimed to address this gap by systematically reviewing these effect modifications and providing insights to inform efforts aimed at reducing differential health burdens.

Methods:

We searched PubMed, Embase, Scopus, and Web of Science for studies conducted in the U.S. published between January 1, 1994, and January 27, 2024. Covidence was used to manage study selection and data extraction, including study design, air pollutants, population vulnerability indicators (e.g., urbanicity, proximity to pollution, socioeconomic status, race/ethnicity), exposure assessment methods, cancer types, and findings.

Results:

We identified 15 articles and found that the associations between air pollution and cancer mortality varied by urbanicity and socioeconomic status. A slight majority of studies reported stronger associations in less urbanized populations and lower-SES groups. Black individuals experienced higher cancer mortality risks from air pollution than Whites, while findings for Hispanics and Asians were inconsistent.

Conclusion:

Air pollution exposure and population vulnerability indicators jointly contribute to disparities in cancer mortality, disproportionately affecting vulnerable populations such as those in less urbanized areas and marginalized racial and ethnic groups. Targeted research and policies are needed to mitigate these inequities, improve air quality, and reduce cancer mortality in high-risk communities.

Graphical Abstract

1. Introduction

Ambient air pollution is a complex mixture of pollutants, including particulate matter with diameters less than 10 micrometers (μm) (PM₁₀) and less than 2.5 μm (PM_2.5), nitrogen oxides (NO₂, NO_x), sulfur dioxide (SO₂), carbon monoxide (CO), ozone (O₃), and other constituents. Despite regulatory measures, air pollution remains a significant concern in North America, contributing to health and environmental issues. It is estimated that approximately 131 million people in the U.S. live in areas where pollutant levels exceed national standards, especially in industrialized and densely populated regions (Canada and CC, 2021; Key findings ∣ state of the air, 2025). In 2013, the International Agency for Research on Cancer (IARC) classified outdoor air pollution as carcinogenic to humans (IARC group 1) (Loomis et al., 2014). Exposure to ambient air pollutants, along with other extrinsic environmental factors, is estimated to contribute to 70–90 % of cancer cases in humans (Ou et al., 2020).

Despite the wealth of research on the relationship between air pollution and cancer incidence, studies examining the impact of air pollution on cancer mortality remain limited and report mixed findings (Turner et al., 2020). Studying cancer mortality is critical for identifying disparities in cancer outcomes, guiding targeted interventions, and informing resource allocation for effective cancer control policies. While incidence can be influenced by factors like overdiagnosis and screening, mortality data offer a more definitive measure of cancer’s impact (Estimating mortality from COVID-19, 2020). Examining mortality trends offers a clearer understanding of cancer’s burden and, when compared with incidence trends, can offer evidence of improved treatments and healthcare effectiveness. This comprehensive approach enables more informed decisions to address cancer’s impact across diverse populations.

Studies have shown that the non-cancer health risks (e.g. respiratory disease, preterm birth, and mortality) associated with air pollution disproportionately overburden vulnerable populations, including those living in rural areas, those with lower socioeconomic status or from historically marginalized racial and ethnic groups (Cook et al., 2021; Deguen and Zmirou-Navier, 2010; Geldsetzer et al., 2024; Hooper and Kaufman, 2018; Li et al., 2020a; Son et al., 2020). These population vulnerability indicators are also independently associated with an increased risk of cancer mortality, including both all-cause cancer mortality and major site-specific cancers such as breast, colorectal, lung, and prostate cancers (Gupta and Akinyemiju, 2024; Singh et al., 2011; Singh and Jemal, 2017; SEER*Explorer, 2024). Population vulnerability factors may modify the impact of air pollution on cancer mortality. This may occur, for example, by influencing an individual’s level of exposure to air pollution and susceptibility to air pollution-induced health outcomes (US EPA O, 2021). However, research on the combined effects of air pollution exposure and environmental injustices on cancer mortality remains limited.

This study aims to systematically identify and review the effect modification between air pollution and population vulnerability on cancer mortality through a comprehensive scoping literature review.

2. Methods

2.1. Search strategy

We worked with an informationist from the Johns Hopkins Welch Medical Library to develop and conduct the searches. Database searches included PubMed, Embase, Scopus, and Web of Science to identify studies published between January 1, 1994 to January 27, 2024. The search strategies were developed using a combination of controlled vocabulary and keywords to define the key concepts; these strategies were translated for each database. Detailed methodological information is available in the OSF (Open Science Framework) registry (Ji et al., 2024). The full search strategy is provided in Table S1.

Search results were exported to EndNote 21 library (Clarivate, Philadelphia, PA, USA) and deduplicated before being imported into Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia) (EndNote, n.d.; Covidence, n.d.). We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The screening workflow is depicted in the study flow diagram (Fig. 1). In addition to the database results, we searched for key review articles on similar topics and reviewed the reference lists of studies that met the inclusion criteria to identify other relevant studies that may have been missed otherwise.

Fig. 1.

PRISMA flow diagram summarizing the study identification, screening, eligibility assessment, and inclusion process.

2.2. Inclusion and exclusion criteria

Studies were eligible for inclusion if they were conducted in the U.S., published between 1994 and January 2024, and focused on air pollution, population vulnerability, and cancer mortality. We captured a broad range of population vulnerability factors in combination with air pollution exposures, many of which are related to environmental justice (Table 1). Studies were excluded if they were published before 1994 due to the limited availability of PM_2.5 air pollution data, or after January 2024 when the search was conducted. We also excluded reviews, reports, or non-human studies (e.g., in vitro or animal studies). Furthermore, we excluded studies that lacked quantitative estimates or those that did not rely on direct and objective measurements of air pollution (e.g., used proximity to highways as a proxy of exposure to air pollution rather than directly measuring pollutant concentrations). However, we included one study that used proximity to traffic or industry as an indicator of population vulnerability and had objective measures of air pollution.

Table 1.

Summary of air pollutants and population vulnerability indicators considered.

Category	Variables/Measures*
Air Pollutants	PM2.5, PM2.5 constituents, Source-specific PM2.5 PM10, PM15, PM15–25, sulfate, black carbon (BC), total suspended particles (TSP), NO2, NOx, O3, SO2, CO, and Environmental Quality Index (EQI)
Urbanicity	- Urbanicity was assessed using different classification systems including: Rural-Urban Continuum Codes (RUCC), Rural Health Research Center categories, USGS Land Cover, and U.S. Census definitions. - Categorized as “urbanized” (e.g., RUCC1–2, metropolitan) vs. “less urbanized” (e.g., RUCC3–4, rural); some studies included a separate “suburban” group.
Proximity to Traffic	Proximity to traffic was defined as a ZIP code centroid within 500 m or 1000 m of a freeway.
Socioeconomic Status (SES)	Education level, household income, and/or a composite SES index based on seven census indicators: educational attainment, median household income, percent of population below 200 % of the federal poverty level, blue-collar employment, unemployment rate among those over age 15, median rent, and median home value.
Race/Ethnicity	Race and ethnicity were defined based on self-reported categories. Most studies included groups such as White, Hispanic, African American, and Asian, while some also reported American Indian/Alaska Native (AI/AN), Native Hawaiian/Pacific Islander, European American, or Japanese American populations, depending on study design and geographic focus.

2.3. Study screening and data extraction

Six reviewers participated in the study selection process, and each article was independently reviewed by a minimum of two reviewers. Titles and abstracts of identified studies were screened to exclude those that did not meet the inclusion criteria. While the search strategy included broad MeSH terms, including “environmental pollution” and its subcategories (e.g. noise, transportation) to ensure comprehensiveness, the review was subsequently focused exclusively on air pollution. Studies focusing solely on other environmental exposures (e.g., noise) without a primary air pollution component were excluded during the screening process. After these exclusions, full-text reviews were conducted to identify studies that provided quantitative estimates of the impact of air pollution on cancer mortality stratified by population vulnerability indicators. We also reviewed the reference list for key review papers and included the referenced studies in the screening and full-text review process (Zhang et al., 2022). One additional article was retrieved through this process (Moore et al., 2017). For the studies that met the inclusion criteria, we extracted data on the study design, air pollutants measured, population vulnerability indicators studied, exposure assessment methods, cancer types considered, study population characteristics, data analysis methods, and the main study findings, which included the associations reported between each air pollutant and cancer mortality, the stratification of associations by population vulnerability indicators. Since studies varied widely in how air pollution and population vulnerability were measured and reported, numerical effect estimates are not readily comparable. To preserve clarity and facilitate comparison, we used the terms “positive association”, “negative association”, and “no association” to summarize general patterns. The use of these terms reflects the direction of reported associations, not the statistical significance of findings. Comparisons between subgroups were based solely on point estimates and no statistical tests were conducted. We also extracted the strengths and limitations of each study as reported by the original authors and incorporated our critical review of each study’s strengths and limitations as related to the purpose of this review in Table 2. As this is a scoping review, we did not conduct a formal risk of bias assessment. Two reviewers performed the data extraction for each article, identified any discrepancies among reviewers, and reached a resolution when any discrepancy between reviewers was identified. Reviewers compared independent results after each stage of the study selection and data extraction process. For the present scoping review, no formal statistical analysis were conducted to compare the strength of associations between air pollution and cancer mortality across population vulnerability indicators.

Table 2.

Summary of the 16 studies included in this review, air pollutants, Population Vulnerability Indicators, and cancer mortality, sorted by study type.

Author, (Year)	Study Details (grouped by study design)	Air Pollutant Metrics, Measurement Details, Exposure Assessment, and data Analysis	Population Vulnerability Indicators Included	Results (cancer mortality)*	Strengths and Limitations
PROSPECTIVE STUDIES
Cheng et al. (2022)	Period of data collection: 1993–2013 Study Design: Prospective Cohort study Location: Los Angeles, California Population: Adults, N = 3089 Cancer Type: Breast	Primary air pollutants: PM2.5, PM10, NO2, NOx Covariate: None Exposure Assessment: 1. Pollutants were estimated at monthly resolution using data from U.S. EPA air monitoring stations (PM2.5, PM10, NO2, NOx) and land use regression models (PM2.5) Data Analysis: 1. Time-dependent Cox regression model for the period 1993–2013 2. For each cancer death (index death) that occurred during that period, the model compared mean air pollution concentration from date of diagnosis to the age of the index death (months) for all individuals uncensored at that age 3. The model was repeated for all index deaths	Primary indicators: 1. Neighborhood SES (nSES) index: a composite measure derived from principal component analysis of seven census indicators: education, median household income, poverty rate (<200 %), blue-collar employment, unemployment (>15 years), median rent, and median home value 2. Race: African Americans (AA), European Americans (EA), Japanese Americans (JA), Latin Americans (LA)	Primary results***: positive associations (PM2.5, PM10, NO2, NOx) Stratified§: 1. SES: Weaker positive associations in low nSES vs. high nSES population (NOx, NO2, PM2.5, PM10). 2. Race: Weaker positive associations in AA and LA vs. EA (NOx, NO2, PM2.5, PM10); no result reported for JA.	Strengths: 1. Diverse population in terms of race/ethnicity and SES 2. Reduced exposure misclassification through the use of detailed residential histories 3. Comprehensive adjustment for individual-level factors (e.g., demographics; lifestyle factors at baseline, including smoking and alcohol intake; clinicopathologic and treatment factors at diagnosis) [hereafter: Comprehensive confounder adjustment] 4. An index of nSES based on seven census-based indicators of SES 5. Time-varying exposure data in Cox regression, used in this study, innovatively captures dynamic shifts in exposure status over time Limitations: 1. Lacks key individual SES data (e.g., insurance) 2. Missing non-residential/indoor pollution and treatment data 3. Limited statistical power to detect associations for some outcomes due to the small sample size in subgroup analysis 4. Multiple comparisons increase false discovery risk
Jerrett et al. (2005)	Period of data collection: 1982–2000 Study Design: Prospective Cohort study Location: Los Angeles, California Population: Adults, N = 22,905 Cancer Type: Lung, gastrointestinal	Primary air pollutant: PM2.5, Covariate: O3 Exposure Assessment: 1. PM2.5 estimates for the year 2000 were derived from 23 state and local district monitoring stations in the Los Angeles basin using five interpolation methods Data Analysis: 1. Cox proportional hazards regression (Survival is not explicitly mentioned)	Primary indicator: 1. Proximity: distance from the zip code-area centroid to the freeway	Primary results: positive associations (Lung, gastrointestinal cancer) Stratified: 1. Proximity for lung cancer: positive association for participants in zip codes with centroids within 500 m; negative association for participants in zip codes with centroids within 1000 m of a freeway. 2. Proximity for gastrointestinal cancer: weaker negative association for participants in zip codes with centroids within 500 m vs. 1000 m of a freeway.	Strengths: 1. State-wide representative sample 2. The combined use of universal kriging and multiquadric models for air pollution exposure assessment, which takes advantage of the local detail in the multiquadric surface and the ability to handle trends in the universal surface 3. [Comprehensive confounder adjustment] 4. Tested for exposure error and co-pollutant effects 5. Implementation of a novel spatial random-effects Cox model, as used in this study, to reduce bias Limitations: 1. Not representative of general US population 2. Complete residential history information was unavailable, which may introduce exposure misclassification due to residential mobility [hereafter: Incomplete residential history] 3. Traffic exposure crudely estimated by traffic proximity at zip code centroid
Kazemiparkouhi et al. (2020)	Period of data collection: 2000–2008 Study Design: Prospective Cohort study Location: United States Population: Adults, N = 52.9 million, Cancer Type: Lung, cancer in general	Primary air pollutant: O3 Covariate: PM2.5, NO2 Exposure Assessment: 1. Daily maximum hourly O3 concentration was estimated using data from EPA AQS (Air Quality System) for 2000 through 2008. For each calendar year, daily 1-h maximum values were averaged from April through September to obtain warm season average, 1-h daily maximum O3 concentrations (warm-season 1-h maximum O3) 2. Two secondary exposure measures were calculated, including the (1) warm-season average of daily 8-h maximum and (2) warm-season average of 24-h average Data Analysis: 1. Log-linear regression models	Primary indicator: 1. Urbanicity: Urban, non-urban classified using USGS (United States Geological Survey) and Land Cover Trends classifications Covariate: 1. Race: non-White versus non-Hispanic White	Primary results: positive associations (lung cancer); null (cancer in general) Stratified: 1. Urbanicity: Positive associations in non-urban residents and negative associations in urban residents.	Strengths: 1. Representative sample of U.S. Medicare beneficiaries 2. [Comprehensive confounder adjustment] 3. Compared multiple O3 metrics (1-h max, 8-h max, 24-h avg) Limitations: 1. Limited generalizability and potential misclassification due to the use of Medicare data, which primarily represents an older adult population and may lack completeness or accuracy for certain variables [hereafter: Medicare data limitations] 2. Unmeasured confounding due to a lack of individual-level data (e.g., long-term temporal trends) [hereafter: Unmeasured confounding] 3. Exposure misclassification from assigning exposure based only on ZIP code of residence
Kazemiparkouhi et al. (2022)	Period of data collection: 2000–2008 Study Design: Prospective Cohort study Location: United States Population: Adults, N = 15.4 million Cancer Type: Lung; cancer in general	Primary air pollutants: Total PM2.5, PM2.5 constituents, Source-specific PM2.5** Covariates: None Exposure Assessment 1. Total PM2.5 estimated using generalized additive mixed models 2. PM2.5 constituents obtained at a daily resolution from the PM2.5 National Chemical Speciation Network (CSN) and the Interagency Monitoring of Protected Visual Environments (IMPROVE) networks for the period 2000–2008 3. Source-specific PM2.5 for which factor analysis was applied to PM2.5 constituents to identify eight source-related factors. 4. Exposure was assessed using 1- to 5-year moving averages for all examined outcomes for each month and location over the study period. Data Analysis: 1. Cox proportional hazards regression (Survival is not explicitly mentioned)	Primary indicators: 1. Race: Non-Hispanic White, Hispanic or Latino, Non-Hispanic Black, and Asian 2. SES: annual mean gross adjusted income Covariate: 1. Urbanicity: urban, non-urban	Primary results: positive associations for cancer in general and lung cancer (Total PM2.5) Stratified: 1. Race for lung cancer: stronger positive association for Asians vs. Whites; weaker positive association for Black vs. Whites; negative association for Hispanic individuals. 2. Race for cancer in general: weaker positive association for Black vs. Whites; negative associations for Hispanic and Asian individuals. 3. SES: stronger positive associations for Middle- and High-income groups vs. Low-income group for lung cancer and cancer in general.	Strengths: 1. Representative sample of U.S. Medicare beneficiaries 2. Compared multiple PM2.5 matrices (Total PM2.5, specific PM2.5 constituents, and source-specific PM2.5) 3. Detailed effect modification analysis to examine interactions between air pollution and various factors, including age, sex, race, and ZIP code-level SES 4. Evaluated critical exposure windows using 1–5 year moving averages of air pollution exposure Limitations: 1. [Medicare data limitations] 2. [Unmeasured confounding] 4. Exposure misclassification from assigning exposure based only on ZIP code of residence 3. Multiple comparisons increase false discovery risk
Krewski et al. (2005)	Period of data collection: 1982–1989 Study Design: Prospective Cohort study Location: 151 metropolitan areas in the United States Population: Adults, N = 552,138 Cancer Type: Lung	Primary air pollutants: PM2.5, Sulfate Covariates: PM10, SO2, NO2, CO Exposure Assessment 1. PM2.5 levels were measured using the EPA’s Inhalable Particle Monitoring Network (1979–1983) 2. Sulfate concentrations in the 151 metropolitan areas were assembled from multiple sources, most of which were derived from Özkaynak and Thurston. 3. The mean sulfate concentration (for 1980) and the median PM2.5 concentration (1979–1983) were used as air pollution indices Data analysis: 1. Cox proportional hazards regression	Primary indicator: 1. Education: less than high school (HS), completion of HS, education beyond HS	Primary results: positive association (Sulfate); null (PM2.5) Stratified: 1. Education: stronger positive association for individuals with lower vs. higher education levels (Sulfate); positive associations for individuals with less than HS or completion of HS, negative association for those with education beyond HS (PM2.5).	Strengths: 1. Use of a national representative sample, providing high generalizability [High generalizability from national representative samples] 2. [Comprehensive confounder adjustment] 3. Adjustment for air pollutants using multi-pollutant models Limitations: 1. [Incomplete residential history] 2. Restricted to a subset of adults residing in U.S. metropolitan areas with available monitoring data for sulfate (151 areas) and fine particles (50 areas)
Pope et al. (2002)	Period of data collection: 1982–1998 Study Design: Prospective Cohort study Location: United States Population: Adults, N = 500,000 Cancer Type: Lung	Primary air pollutants: PM2.5, PM10, O3, SO2, NO2, Sulfate, PM15, PM15–25, TSP, CO Covariate: None Exposure Assessment: 1. Air pollution data (PM2.5, PM10, SO2, NO2, CO, O3) were obtained from the EPA’s Aerometric Information Retrieval System (AIRS) and Inhalable Particle Monitoring Network (IPMN) 2. Data were averaged across sites, imputed when missing, and PM2.5 was back-casted using PM2.5:PM10 ratios 3. Sulfate levels were adjusted for filter artifacts, and O3 used daily 1-h maximums 4. The mean concentration of each pollutant from all available monitoring sites was calculated for each metropolitan area during the 1 to 2 years prior to enrollment 5. PM2.5 concentrations were averaged separately for 1979–1983, 1999–2000, and overall, and each was used in separate models 6. Survival times of participants who did not die were censored at the end of the study period Data Analysis: 1. Extended Cox proportional hazards regression model with a spatial random-effects component	Primary indicator: 1. Education: less than high school (HS), completion of HS, education beyond HS Covariates: 1. Urbanicity: Urban 2. Race: Non-Hispanic White vs. Minority	Primary results: positive associations (PM2.5, PM10, PM15, sulfate, SO2); negative associations (TSP, NO2, CO, and O3), null (PM15–25) Stratified: 1. Education: stronger positive associations for individuals with lower levels of education vs. those with higher education levels (for PM2.5); no results reported for other air pollutants.	Strengths: 1. [High generalizability from national representative samples] 2. Extended follow up 3. Used advanced statistical modeling, including an extended Cox proportional hazards model to reduce bias 4. [Comprehensive confounder adjustment] Limitations: 1. [Unmeasured confounding] 2. [Incomplete residential history]
Pope et al. (2019)	Period of data collection: 1986–2014 Study Design: Prospective Cohort study Location: United States Population: Adults, N = 635,539 Cancer Type: Lung	Primary air pollutant: PM2.5 Covariate: None Exposure Assessment: 1. PM2.5 was estimated at an annual resolution using data from regulatory monitoring and constructed within a universal kriging framework from 1999 to 2015 2. For each cause of death, a separate Cox model was used to compare the mean air pollution concentration from the date of the interview to the date of death for those who died and to the end of mortality follow-up (December 31, 2015) for survivors Data Analysis: 1. Cox proportional hazards regression with sample weights to account for the complex survey design	Primary indicators: 1. Urbanicity: Urban, non-urban, 2. SES: education, income 3. Race: Non-Hispanic White, Hispanic or Latino, and Non-Hispanic Black	Primary results: positive association Stratified: 1. Urbanicity: stronger positive association in individuals living in urban areas vs. those living in rural areas. 2. Race: stronger positive association in Hispanics vs. Whites; negative association in Black individuals. 3. Education: stronger positive associations in higher educated individuals vs. lower educated individuals.	Strengths: 1. [High generalizability from national representative samples] 2. [Comprehensive confounder adjustment] 3. Used advanced statistical modeling, including complex Cox proportional hazards models with sample weights to account for the complex survey design of this cohort study Limitations: 1. [Incomplete residential history] 2. Potential bias from missing PM2.5 data monitoring networks before 1999 3. [Unmeasured confounding]
Wang et al. (2020)	Period of data collection: 2000–2008 Study Design: Prospective Cohort study Location: United States Population: Adults, N = 52,954,845 Cancer Type: Lung, cancer in general	Primary air pollutant: PM2.5 Covariate: NO2 Exposure Assessment: 1. Daily PM2.5 concentrations were estimated using data from validated spatiotemporal generalized additive mixed models. Data Analysis: 1. The study utilized a Cox proportional hazards model with restricted cubic spline 2. For each cause of death, the model compared the mean air pollution concentration over the previous 12 months to the time of death 3. This analysis was repeated for all causes of death	Primary indicators: 1. Urbanicity: Urban and non-urban 2. SES: income 3. Race: Non-Hispanic White, Hispanic or Latino, Non-Hispanic Black, and Asian	Primary results: positive association (cancer in general); null (lung cancer) Stratified: 1. Urbanicity for cancer in general: no association for non-urban individuals; positive association in urban individuals. 2. SES for cancer in general: stronger positive association in high-income individuals vs. low-income individuals. 3. Race for cancer in general: stronger positive association in Black vs. White; no association for Asian; negative association for Hispanics. 4. Urbanicity for lung cancer: negative association in non-urban individuals; positive association in urban individuals. 5. SES for lung cancer: positive association in high income individuals; negative association in low-income individuals. 6. Race for lung cancer: stronger positive association in Asian and Black vs. White; negative association in Hispanic.	Strengths: 1. Representative sample of U.S. Medicare beneficiaries 2. Application of restricted cubic splines to characterize non-linearities in the PM2.5-mortality association Limitations: 1. Exposure misclassification from assigning exposure based only on ZIP code of residence 2. Multiple comparisons increase false discovery risk 3. [Unmeasured confounding] 4. [Medicare data limitations]
RETROSPECTIVE COHORT STUDIES
Eum et al. (2022)	Period of data collection: 2001–2008 Study Design: Retrospective Cohort study Location: United States Population: Adults, N = 49,712,702 Cancer Type: Lung, cancer in general	Primary air pollutant: NO2 Covariate: PM2.5, BC Exposure Assessment: 1. NO2 was estimated at a monthly resolution using data from land-use regression models Data Analysis: 1. The study utilized Cox regression model from 2001 to 2008 2. The model compared the mean air pollution concentration over the previous 12 months to the age at cancer death (in months) for all individuals who remained alive at that age	Primary indicators: 1. Urbanicity: urban, suburban, rural areas 2. Race: Non-Hispanic White, Hispanic or Latino, Non-Hispanic Black, Asian, American Indian or Alaska Native, and Native Hawaiian or Other Pacific Islander Covariate: 1. SES: annual mean gross adjusted income	Primary results: positive associations (lung and cancer in general) Stratified: 1. Urbanicity for lung cancer: positive associations in urban and rural residents; negative association in suburban residents. 2. Race for lung cancer: stronger positive associations in Asian and Black vs. White and null in Hispanics. 3. Urbanicity for cancer in general: positive associations in urban and rural residents; negative associations in suburban residents. 4. Race for cancer in general: stronger positive associations in Asian and Black vs. White; negative association in Hispanic.	Strengths: 1. [High generalizability from national representative samples] 2. Fine-resolution exposure estimates with validated NO2 model 3. PM2.5 and BC adjusted using two-stage and two-pollutant approaches Limitations: 1. [Medicare data limitations] 2. [Unmeasured confounding] 3. Exposure misclassification from assigning exposure based only on ZIP code of residence
Villanueva et al. (2021)	Period of data collection: 1996–2014 Study Design: Retrospective Cohort study Location: California Population: Adults, N = 29,844 Cancer Type: Ovarian	Primary air pollutants: PM2.5, O3, NO2 Covariate: None Exposure Assessment: 1. All pollutants were averaged from the date of diagnosis to the date of death or last follow-up 2. Data obtained from the California Air Resources Board’s online database, Air Quality and Meteorological Information System 3. Exposure was assigned to women by spatially joining their geocoded residential location to the air pollution data 4. NO2 exposure was categorized as: <20.0 ppb (low level), 20.0–30.0 ppb (intermediate level), and >30.0 ppb (high level) Data Analysis: 1. Cox proportional hazards models	Primary indicators: 1. SES: Yost Score (community-level measure using census block group-level variables) for pre-2006 diagnoses; Yang index (comparable measure but uses block group variables from the American Community Survey) for post-2006 diagnoses 2. Race: Non-Hispanic White, Hispanic or Latino, Non-Hispanic Black, Asian, and Native Hawaiian or Other Pacific Islander (Asian/PI) Covariate: 1. Proximity: Proximity to traffic or industry	Primary results: positive associations (PM2.5, NO2, O3) Stratified: 1. SES: stronger positive associations in lower SES individuals vs. those with higher SES (PM2.5, NO2, O3). 2. Race: stronger positive associations in Black vs. Whites (O3, high level NO2); weaker positive association in Black vs. Whites (intermediate level NO2); weaker positive associations in Black vs. Whites (PM2.5); stronger positive associations in Hispanics vs. Whites (PM2.5, NO2, O3); weaker positive associations in Asian/PI vs. Whites (PM2.5, intermediate level NO2).	Strengths: 1. Analysis of air pollution exposures linked to multiple SES indicators 2. Air pollution exposure was averaged over each subject’s survival period, providing a more robust and accurate measure compared to single-point (e.g., 1-year averaged) estimate 3. The availability of geocoded residential addresses at the time of diagnosis allowed for the interpolation of PM2.5 exposures at the individual level, reducing exposure misclassification compared to larger-area (e.g., zip code) estimates Limitation: 1. [Incomplete residential history]
ECOLOGICAL STUDIES
de Grubb et al. (2017)	Period of data collection: 1999–2009 Study Design: Ecological study Location: United States Population: Adults, N = not specified Cancer Type: Lung	Primary air pollutant: PM2.5 Covariate: None Exposure Assessment: 1. County-level averages of concentrations were calculated for the period 2003 to 2008 using data provided on the U.S. Centers for Disease Control and Prevention WONDER platform, derived from monitoring station data and modeling outputs Data Analysis: 1. Ordinary least squares multiple regression	Primary indicator: 1. Race: Non-Hispanic White and Non-Hispanic Black	Primary results: did not report Stratified: 1. Race and gender: stronger positive association in Black women vs. White women; weaker positive association in Black men vs. White men.	Strengths: 1. Analyzed race-, sex-, and geography-specific disparities 2. Incorporated multiple vulnerability indicators (smoking, PM2.5) 3. Hotspot mapping identified high-risk, underserved areas Limitations: 1. Reliance on death certificate data 2. The ecological study design limits the ability to infer causality and is primarily hypothesis-generating and has no individual-level data on exposure or confounders [hereafter: Limitations of ecological design] 3. [Unmeasured confounding]
Huang et al. (2019)	Period of data collection: 2014 Study Design: Ecological and cross-sectional study Location: United States Population: Adults, N = 3107 Cancer Type: Breast; Other: cervical, ovarian and uterine	Primary air pollutant: Air domain from Environmental Quality Index (EQI) Covariate: None Exposure Assessment: 1. Air domain from EPA’s EQI, which integrates 87 variables representing 6 criteria air pollutants (PM10, PM2.5, Sulfur Dioxide, Nitrogen Dioxide, Ozone, and Carbon Monoxide) and 81 hazardous air pollutants for 2000–2005 2. County-level averages calculated using principal component analysis (PCA) Data Analysis: 1. Spatial autoregressive (SAR) model	Primary indicator: 1. Urbanicity: metropolitan urbanized (RUCC1), nonmetropolitan urbanized (RUCC2), less urbanized (RUCC3), and thinly populated (RUCC4) areas Covariate: 1. Race: Non-Hispanic White	Primary results: positive associations (breast, ovarian, uterine cancer, and cervical cancer) Stratified: 1. Urbanicity: stronger positive associations in less urbanized areas (RUCC2 and RUCC4) vs. RUCC1 (breast, cervical, uterine); weaker positive associations in RUCC3 vs. RUCC1 (ovarian, uterine).	Strengths: 1. [High generalizability from national representative samples] 2. Applied the novel EQI from 219 environmental variables across five domains (air, water, land, built environment, and sociodemographic) that captures cumulative environmental quality Limitations: 1. [Limitations of ecological design] 2. Unable to isolate effects of specific factors within air domain of EQI
Ito and Thurston (1996)	Period of data collection: 1985–1990 Study Design: Ecological study Location: Cook County, Illinois Population: Adults, N = not specified Cancer Type: cancer in general	Primary air pollutants: PM10, O3, CO, SO2 Covariate: None Exposure Assessment: 1. All pollutants were assessed using the EPA’s Aerometric Information Retrieval System (AIRS) for 1985–1990, including sites with at least four years of data during the six-year period Data Analysis: 1. Poisson regression models	Primary indicator: 1. Race: Non-Hispanic White, and Non-Hispanic Black Covariate: 1. Urbanicity: Urban	Primary results: positive associations (PM10 and O3); null (CO, SO2) Stratified (PM10 only): 1. Race: stronger positive association in Black vs. Non-Hispanic White.	Strength: 1. Provided regional mapping to identify geographic disparities in cancer mortality Limitation: 1. [Limitations of ecological design]
Li et al. (2020b)	Period of data collection: 2006–2010 Study Design: Ecological study Location: United States Population: Adults, N = not specified Cancer Type: Tracheal, bronchus and lung (TBL) cancers	Primary air pollutant: Air domain from Environmental Quality Index (EQI) Covariate: None Exposure Assessment: 1. Air domain from EPA’s EQI, which integrates 87 variables representing 6 criteria air pollutants (PM10, PM2.5, Sulfur Dioxide, Nitrogen Dioxide, Ozone, and Carbon Monoxide) and 81 hazardous air pollutants for 2000–2005 Data Analysis: 1. Simultaneous autoregressive (SAR) models	Primary indicator: 1. Urbanicity: metropolitan urbanized (RUCC1), nonmetropolitan urbanized (RUCC2), less urbanized (RUCC3), and thinly populated (RUCC4) areas	Primary result: positive association Stratified: 1. Urbanicity: stronger positive associations for individuals living in less urbanized areas (RUCC2-4) vs. RUCC1.	Strengths: 1. [High generalizability from national representative samples] 2. Applied the novel EQI from 219 environmental variables across five domains (air, water, land, built environment, and sociodemographic) that captures cumulative environmental quality 3. Incorporates spatial heterogeneity Limitations: 1. [Limitations of ecological design] 2. Unable to isolate effects of specific factors within air domain of EQI
Moore et al. (2017)	Period of data collection: 2004–2014 Study Design: Ecological study Location: United States Population: Adults and children, N = 362 counties Cancer Type: Lung	Primary air pollutant: PM2.5 Covariate: None Exposure Assessment: 1. County-level PM2.5 were estimated using daily satellite data (MODIS) and EPA environmental data, modeled into continuous spatial surfaces from 2003 to 2011 2. Mean daily PM2.5 levels were calculated and categorized into quartiles for analysis Data Analysis: 1. Logistic regression	Primary indicator: 1. Urbanicity: Urban, non-urban Covariate: 1. Race: Non-Hispanic White, Hispanic or Latino, Non-Hispanic Black, Asian, and American Indian or Alaska Native 2. SES: household income	Primary result: positive association Stratified: 1. Urbanicity: stronger positive association for individuals living in non-urban areas vs. those living in urban areas.	Strength: 1. Used geospatial clustering to identify vulnerable regions Limitation: 1. [Limitations of ecological design]

Õ In brackets we define a short name for a strength or limitation that will be used in the table again, in order to facilitate comparability and reduce table size.

^*The terms “positive association,” “negative association,” and “no association” reflect the direction of the observed relationship and do not imply statistical significance.

^**Report only total PM_2.5 for comparability with other studies;

^***We include the primary results to help readers understand the overall direction of the association, and because there are cases where the direction of association differs across subgroups.

^§Comparisons of stratified analysis reflect a comparison of the point estimates and do not imply statistical significance of those differences.

2.4. Type of cancer analyzed

We analyzed all papers including solid malignancies without focusing on a specific type due to the limited availability of granular data on the relationship between environmental exposures and cancer mortality. Given the complexity of environmental risk factors and their potential to contribute to multiple cancer types, we sought to take a broader approach to identify overarching patterns and trends. By including all solid malignancies, we aimed to explore whether any discernible associations emerged, which could then inform more targeted research in the future.

2.5. Air pollutants and population vulnerability indicators analyzed

Table 1 summarized air pollutants and population vulnerability indicators examined in the studies included in the manuscript. Air pollutants in the analysis included PM_2.5, PM₁₀, sulfate, black carbon (BC), total suspended particles (TSP), NO₂, NO_x, O₃, SO₂, CO, and indexes that integrated multiple air pollutants. Among the included studies, urbanicity was defined in several ways, including classification by ZIP codes using the Rural Health Research Center Categorization, Rural-Urban Continuum Codes established by the U.S. Department of Agriculture, United States Geological Survey Land Cover Trends classifications, and definitions provided by the U.S. Census Bureau. The studies employed varying classifications; two studies used metropolitan urbanized (RUCC1), nonmetropolitan urbanized (RUCC2), less urbanized (RUCC3), thinly populated (RUCC4) (Li et al. (2020a) and Huang et al. (2019)), and five studies used the urban vs. suburban vs. non-urban classifications (Eum et al. (2022), Kazemiparkouhi et al. (2020), Moore et al. (2017), Pope et al. (2019), and Wang et al. (2020)). To address these differences in classification systems, we grouped these categories under the broader term ‘urbanicity.’ For binary classifications, we grouped urban and metropolitan as “urbanized” and non-urban, non-metropolitan, and rural as “less urbanized.” RUCC1 and RUCC2 were categorized as “urbanized,” while RUCC3 and RUCC4 were categorized as “less urbanized.” Only one study (Eum et al., 2022) included ‘suburban’ as a distinct category. Therefore, we chose to retain it as a separate group. Proximity to traffic was defined as ZIP code centroids located within either 500 or 1000 m of a freeway. SES indicators included income and education. Race and ethnicity categories included White, Hispanic, African American, Asian, American Indian/Alaska Native (AIAN), Native Hawaiian or Pacific Islander, European Americans, Japanese Americans. For one study, a ‘General minority’ group was used to reflect instances where specific racial/ethnic groups were not defined but were collectively compared to Non-Hispanic White individuals to ensure a comprehensive evaluation of disparities.

3. Results

A total of 11,818 potentially relevant studies were found during the search. After removing duplicates, 8004 references remained for screening. Of these, 7849 were excluded based on title and abstract review. The remaining 155 articles underwent full-text review, after which 139 were excluded. In total, 15 articles were included in the review, which includes eight prospective cohort studies, two retrospective cohort studies, and five ecological studies. The detailed article selection process is illustrated in Fig. 1. A summary of each study’s design, air pollutants measured, population vulnerability indicators considered, and main findings is presented in Table 2.

Three main categories of population vulnerability indicators were identified in our scoping review: urbanicity/proximity (n = 9), socioeconomic (n = 7), and race/ethnicity (n = 8). The articles featured in this scoping review included five primary cancer types: lung/bronchus cancer (n = 12), breast cancer (n = 2), ovarian cancer (n = 2), cervical cancer (n = 1), gastrointestinal cancer (n = 1), and cancer in general (n = 5). Furthermore, our review yielded studies that examined the following air pollutants: PM_2.5 (n = 13), PM₁₀ (n = 4), nitrogen oxides (NO₂: n = 7; NO_x: n = 1), O₃ (n = 5), SO₂ (n = 3), CO (n = 3), BC (n = 1), TSP (n = 1), index of multiple air pollutants (n = 3), and sulfate (n = 2).

3.1. Urbanicity/proximity indicators

Eight of the 15 articles examined the impact of urbanicity/proximity indicators on the associations between air pollution and cancer mortality. Of those eight articles, seven focused on urbanicity, and only one assessed proximity to pollution sources. The results reveal notable disparities in cancer mortality linked to air pollution by urbanicity, with distinct patterns emerging across studies as described in the following sections.

3.1.1. Urbanicity

Two studies (Li et al. (2020a); Huang et al. (2019)) examined the impact of air pollution indexes, including the environmental quality index (EQI) from the US EPA), on mortality due to various cancers, including tracheal, bronchus, and lung cancers, breast, cervical, ovarian, and uterine cancers (Li et al., 2020b; Huang et al., 2019). The EQI is an estimate of U.S. county ambient environmental quality across five domains: air, water, land, sociodemographic, and built environment and is meant to be used by researchers to better understand how health outcomes relate to cumulative environmental exposures usually viewed in isolation. The air domain from the EPA’s EQI integrates 87 variables representing 6 criteria air pollutants (PM10, PM2.5, sulfur dioxide, nitrogen dioxide, ozone, and carbon monoxide) and 81 hazardous air pollutants. These studies suggested stronger positive associations among residents in less urbanized areas compared to urbanized areas (e.g., Li et al. reported that for each 1-unit increase in the EQI-air domain score, the percent change in tracheal, bronchus, and lung (TBL) cancer mortality was 2.38 % (95 % CI: 1.90, 2.87) and 3.37 % (95 % CI: 2.42, 4.33) in less urbanized areas (RUCC3 and RUCC4, respectively), compared to 1.68 % (95 % CI: 1.27, 2.09) and 2.19 % (95 % CI: 1.33, 3.05) in urban areas (RUCC1 and RUCC2, respectively). The remaining six studies focused on individual air pollutants. Moore et al. (2017) suggested stronger positive associations between PM_2.5 exposure and lung cancer mortality among less urbanized residents than among urban residents. However, two studies suggested stronger associations for mortality among populations living in urban environments. Compared to residents in less urbanized environments, Pope et al. (2019), suggested stronger positive associations between PM_2.5 exposure and lung cancer mortality among urban residents. Wang et al. (2020), reported positive associations between PM_2.5 exposure and overall cancer mortality in urban populations while no associations for non-urban populations (Wang et al., 2020; Pope et al., 2019). Two studies found mixed effects of air pollution on cancer mortality by urbanicity. Wang et al. (2020) identified negative associations between PM_2.5 exposure and lung cancer mortality in less urbanized populations, but positive associations were observed in urban populations (Wang et al., 2020).

One study by Eum et al. (2022) examined the impact of long-term NO₂ exposure on cancer in general and lung cancer mortality among urban, suburban, and rural residents. The study found that exposure to NO₂ increased cancer mortality in urban and rural residents, but decreased cancer mortality among suburban residents (Eum et al., 2022).

3.1.2. Proximity to pollution sources

Jerrett et al. (2005) assessed the impact of traffic by assigning buffers that included zip code-area centroids within either 500 or 1000 m of a freeway, defining freeways as having “limited access,” a numbered assignment, and a speed limit of greater than 50 miles per hour (Jerrett et al., 2005). This study identified a positive association between PM_2.5 exposure and lung cancer mortality among individuals residing closer to freeways and a negative association for those living farther away. Additionally, stronger negative associations between PM_2.5 exposure and gastrointestinal cancer mortality were suggested in residents closer to freeways compared to those farther away (Jerrett et al., 2005).

3.2. Socioeconomic status indicators

Seven of the 15 articles explored the impact of socioeconomic status (SES) indicators on the associations between air pollution and cancer mortality (Wang et al., 2020; Pope et al., 2019; Villanueva et al., 2021). Specifically, three studies (Krewski et al. (2005), Pope et al. (2002), Pope et al. (2019)) used education as an SES indicator, three used income (Wang et al. (2020), Pope et al. (2019), and Kazemiparkouhi et al. (2022)), and two used an SES index (i.e., Yost score) based on seven SES indicators (Villanueva et al. (2021), Cheng et al. (2022)). Among these seven studies, four suggested stronger positive associations between air pollution and cancer mortality in low SES individuals compared to high SES individuals. Villanueva et al. (2021) conducted a cohort study and suggested stronger positive associations betweenPM_2.5, NO₂, O₃, and ovarian cancer mortality among individuals with lower SES index compared to those with higher SES index (Villanueva et al., 2021). For instance, the hazard ratio (HR) for ovarian cancer mortality was 2.69 (95 % CI: 2.24, 3.23) per interquartile range (IQR) increase in NO₂ for individuals in the lowest SES index quintile, compared to 2.37 (95 % CI: 2.01, 2.79) for those in the highest SES index quintile (Villanueva et al., 2021). Similar results were observed in two other cohort studies by Pope et al. (2002) and Krewski et al. (2005), which suggested stronger positive associations between air pollutants, such as PM_2.5 and sulfate, and lung cancer mortality in populations with lower levels of education (Pope et al., 2002; Krewski et al., 2005). Cheng et al. (2022) suggested stronger positive associations between PM_2.5, PM₁₀, NO₂, NO_x, and breast cancer mortality among individuals living in neighborhoods with higher SES index, which indicates lower SES, compared to those in neighborhoods with lower SES index (Cheng et al., 2022). However, the other three studies reported inconsistent findings. For example, Pope et al. (2019) observed stronger positive associations between PM_2.5 and lung cancer mortality among individuals with higher education (Pope et al., 2019). Wang et al., 2020 suggested that the positive association between PM_2.5 and overall cancer mortality was stronger among individuals with higher income (Wang et al., 2020). The study also reported positive associations between PM_2.5 exposure and lung cancer mortality among individuals with higher incomes, but negative associations were observed among those with lower incomes.

3.3. Race/ethnicity

Eight of the 15 articles evaluated how the associations between air pollution and cancer mortality varied by race and ethnicity (Wang et al., 2020; Pope et al., 2019; Eum et al., 2022; Villanueva et al., 2021; Cheng et al., 2022; de Grubb et al., 2017; Ito and Thurston, 1996; Kazemiparkouhi et al., 2022). According to four studies, Blacks had a higher risk of lung cancer and overall cancer mortality associated with air pollution exposure, compared to Whites (Wang et al., 2020; Eum et al., 2022; de Grubb et al., 2017; Ito and Thurston, 1996). For instance, Eum et al. (2022) reported a mortality HR of 1.22 (95 % CI: 1.19, 1.25) per 10 ppb increase in NO₂ among Black individuals, compared to 1.04 (95 % CI: 1.02, 1.07) among Whites (Eum et al., 2022). Additionally, three cohort studies (Wang et al. (2020); de Grubb et al. (2017) and Ito et al., (1996)) suggested stronger positive associations between PM exposure and lung and overall cancer mortality among Black compared to White individuals (Wang et al., 2020; de Grubb et al., 2017; Ito and Thurston, 1996). The other four studies reported inconsistent findings. Cheng et al. (2022) found weaker positive associations between PM_2.5, PM₁₀, NO_x, and NO₂ exposure and breast cancer mortality among African Americans compared to European Americans (Cheng et al., 2022). Kazemiparkouhi et al. (2022) suggested a stronger increased risk of lung cancer mortality associated with PM_2.5 exposure among White individuals compared to Blacks (Kazemiparkouhi et al., 2022). Pope et al. (2019) reported a negative association between PM_2.5 and lung cancer mortality among Black individuals, and a positive association among White individuals (Pope et al., 2019). Additionally, Villanueva et al. (2021) reported mixed results for different air pollutants, showing a weaker positive association between PM_2.5 exposure and ovarian cancer mortality among Black individuals compared to White individuals, while a stronger positive association was observed for O₃ and high level NO₂ (>30.0 ppb) exposure (Villanueva et al., 2021).

Findings for Hispanics were inconsistent when compared to Whites. For example, Pope et al. (2019), suggested that positive effects of PM_2.5 on lung cancer mortality were stronger among Hispanics (HR = 1.28, 95 %CI: 0.83, 1.96) than Whites (HR = 1.12, 95 %CI: 0.99, 1.28) per 10 μg/m³ increase in PM_2.5 (Pope et al., 2019). Similarly, Villanueva et al. (2021) suggested stronger positive associations between NO₂, PM_2.5, O₃, and ovarian cancer mortality in Hispanics compared to White individuals (Villanueva et al., 2021). However, two cohort studies (Wang et al. (2020) and Kazemiparkouhi et al. (2022)) found that PM_2.5 exposure was positively associated with lung cancer mortality among Whites but negatively associated with lung cancer mortality among Hispanics (Wang et al., 2020; Kazemiparkouhi et al., 2022). Eum et al. (2022) reported that NO₂ exposure was positively associated with lung cancer mortality and overall cancer mortality among Whites, while no association with lung cancer mortality and a negative association with overall cancer mortality were observed among Hispanics (Eum et al., 2022).

For Asian individuals, Wang et al. (2020), Eum et al. (2022), and Kazemiparkouhi et al. (2022) found that a higher risk of lung cancer mortality was associated with PM_2.5 and NO₂ exposure compared to Whites (Wang et al., 2020; Eum et al., 2022; Kazemiparkouhi et al., 2022). Conversely, Villanueva et al. (2021) suggested a lower increased risk of ovarian cancer mortality for PM_2.5, exposure among Asian/Pacific Islanders compared to Whites (Villanueva et al., 2021).

4. Discussion

By analyzing 30 years of research, this review explored the intersection of air pollution and population vulnerability indicators, focusing on urban/proximity, socioeconomic, and racial/ethnic disparities on cancer mortality among adults in the U.S. Summarizing the papers is challenging due to variations in inclusion of air pollutants, population vulnerability indicators, cancer types, and study designs. Nevertheless, this approach offers new insights into the impact of air pollution and population vulnerability on cancer mortality and highlights the need for targeted strategies to address potential differential health burdens.

4.1. Urbanicity/proximity indicators

This review highlights the nuanced and often inconsistent relationships between air pollution and cancer mortality when considering urban/proximity indicators. Three of the seven studies were ecological in design and consistently suggested stronger associations between air pollution and cancer mortality in less urbanized areas (Moore et al., 2017; Li et al., 2020b; Huang et al., 2019). These findings may reflect known disparities in healthcare access. For example, rural adults are less likely to be insured than their urban counterparts (Admon et al., 2023), and rural areas generally have fewer physicians, which can make access to care more challenging (Machado et al., 2021). The remaining four studies, all cohort designs, demonstrated mixed associations between air pollution, urbanicity, and cancer mortality (Wang et al., 2020; Kazemiparkouhi et al., 2020; Pope et al., 2019; Eum et al., 2022). While some identified stronger effects in urban populations, others observed greater risks in rural or non-urban groups, and several suggested divergent results across racial, ethnic, or socioeconomic subgroups. These inconsistencies may reflect differences in the population vulnerability indicators assessed, pollutants measured, and exposure windows considered. While ecological studies showed more consistent associations, their methodological limitations, including the inability to adjust for individual-level confounders and the temporal misalignment between exposure and outcomes, limit the strength of their findings. Cohort studies address these issues and offer stronger inferential value, but the divergent results across those studies highlight the need for harmonized methods, consistent exposure metrics, and populationspecific approaches.

Another key contributor to the inconsistent findings is the varying definition of urbanicity. Among the seven studies, four different classification systems were used: Rural-Urban Continuum Codes (RUCC), Rural/Urban Commuting Area (RUCA) codes, United States Geological Survey (USGS) Land Cover Trends classifications, and U.S. Census Bureau definitions. These inconsistencies complicate direct comparisons across studies. Eum et al. (2022) classified ‘suburban’ as a distinct category. Given that suburban populations in the U.S. typically demonstrate distinct demographic characteristics, such as age distribution and socioeconomic status, these findings may not be generalizable to all areas of the United States (Igielnik et al., 2018). Consistent with our methodological approach, we maintained the original categorizations as specified in each included study. To enhance comparability, future research should adopt a standardized definition of urbanicity.

Two of the three ecological studies used air pollution indexes. These indexes assess the overall burden of multiple air pollutants but are unable to distinguish the specific contributions of individual pollutants. These limitations make it challenging to compare their findings with studies that analyze individual pollutants. Nonetheless, our findings point to the critical need for policies that improve air quality across both urban and rural communities, recognizing their unique challenges and health infrastructures.

4.2. Socioeconomic status indicators

The studies included in our review revealed variability in the joint effects of air pollution and SES on cancer mortality, although they shared similarities in air pollutants, cancer types, and study designs. All were cohort studies; however, the specific confounders adjusted for varied across studies, which may partially explain the differences in their findings. Lower SES has been linked to an increased risk of conditions such as obesity, diabetes, and chronic lung diseases, all of which are significant risk factors for cancer mortality (Kivimäki et al., 2020). Additionally, individuals with lower SES may have more limited access to healthcare compared to those with higher SES, further exacerbating health disparities.

4.3. Race/ethnicity

Our review also revealed racial and ethnic disparities, with Black populations facing higher risks of cancer mortality due to increased exposure to air pollution, as compared to White populations. This disparity may be attributed to several factors, including co-exposures to additional environmental hazards, chronic psychosocial stressors, and a greater likelihood of having lower SES (Kazemiparkouhi et al., 2020; Özkaynak and Thurston, 1987; Williams et al., 2016; Kaufman and Hajat, 2021; Sternthal et al., 2011). For studies that report null, weaker positive, or negative effects of air pollution on cancer mortality in Black populations, several factors should be considered to explain the variation in results. For instance, the source of race/ethnicity data may be limited in terms of completeness and accuracy, which could influence the observed associations. Additionally, results from studies conducted in specific regions of the U.S. may not be directly comparable to those from studies using representative samples of the U.S. population (Villanueva et al., 2021).

Our review yielded mixed results for Hispanic individuals. All three studies that compared the effect of PM_2.5 on lung cancer mortality between Hispanics and Whites were prospective cohort studies with large sample sizes. However, their findings varied, with two studies reporting negative associations in Hispanics despite positive associations in Whites (Wang et al., 2020; Kazemiparkouhi et al., 2022), while one study suggested stronger positive associations in Hispanics compared to Whites (Pope et al., 2019) Potential explanations for these discrepancies include misclassification of air pollution exposure, potential information bias in race/ethnicity data, and residual confounding. Many studies, such as Moore et al. (2017) and Huang et al. (2019), adjusted for population vulnerability indicators like race and SES to control for potential confounding. While this approach helps produce less biased overall effect estimates, it does not necessarily clarify whether the strength of associations varies across groups. In contrast, the stratified analyses synthesized in this review (Table 2) offer additional insight by specifically examining how these associations differ across subpopulations. Asian populations generally exhibited stronger positive associations between air pollution and lung cancer mortality compared to Whites, which may be attributed to factors such as elevated psychological stress or different genetic susceptibilities to cancer (Grineski et al., 2017; Zhou and Christiani, 2011).

Several limitations should be acknowledged in interpreting these findings. First, this review was limited to studies conducted in the United States, which may exclude relevant findings from other regions that could provide broader insights into the relationship between air pollution, population vulnerability factors, and cancer mortality. Additionally, variability in the definitions, measurements, and categorizations of air pollutants and population vulnerability indicators across studies makes drawing direct comparisons and identifying consistent trends challenging. Furthermore, the selected population vulnerability indicators, such as SES, race/ethnicity, and proximity to pollution sources, may not fully capture all dimensions of environmental injustice. Ignoring factors like community resilience or the influence of local policies on exposure could result in an incomplete understanding of the structural drivers of disparities and reduce the effectiveness of targeted public health interventions. Another limitation is that while this comprehensive approach allows for a broad examination of potential links between environmental exposures and cancer mortality, aggregating all cancers may mask site-specific associations or unique risk factors that would be more apparent in an analysis of individual cancer types. However, this initial broad assessment provides a necessary foundation for future investigations that may focus on specific cancers with stronger environmental associations.

Results from this review highlight important opportunities for future research and policy action. Researchers can focus on investigating the mechanisms driving observed disparities and evaluating the effectiveness of targeted air pollution interventions to improve cancer outcomes for vulnerable populations. Given the complex associations between air pollution, population vulnerability indicators, and cancer mortality, future studies should incorporate population vulnerability indicators as covariates and perform stratified analyses. This information can be used by policymakers to improve environmental and public health policies. These policies could include prioritizing air quality improvements in high-risk areas, incorporating population vulnerability considerations into cancer prevention and treatment strategies, and enforcing air pollution emission standards. Finally, as existing studies predominantly focus on adults, future research on air pollution, population vulnerability, and cancer mortality should also focus on children as they represent a uniquely vulnerable population.

5. Conclusion

This review suggests that air pollution and population vulnerability indicators collectively influence cancer mortality, with substantial disparities observed across urban/proximity, socioeconomic, and racial/ethnic lines. Vulnerable populations, including those in less urbanized areas and marginalized racial and ethnic communities, may face a disproportionate burden. This scoping review is the first to systematically summarize the impact of air pollution and population vulnerability indicators on cancer mortality. These findings underscore the need for targeted research and policy interventions to address these inequities, improve air quality, and enhance cancer outcomes in high-risk and/or vulnerable populations.

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.scitotenv.2025.181079.

HIGHLIGHTS.

• Blacks had higher air pollution-related cancer risk; varied risks for other races.

• Mixed results between air pollution and cancer mortality by urbanicity and socioeconomic status.

• A consistent definition of urbanicity should be used in future studies.

Data availability

No data was used for the research described in the article.