Home and School Pollutant Exposure, Respiratory Outcomes, and Influence of Historical Redlining

Abstract

Background:

The discriminatory and racist policy of historical redlining in the United States (U.S.) during the 1930s played a role in perpetuating contemporary environmental health disparities.

Objective:

Our objectives were to determine associations between home and school pollutant exposure (fine particulate matter (PM_2.5), nitrogen dioxide (NO₂)) and respiratory outcomes (Composite Asthma Severity Index (CASI), lung function) among school-aged children with asthma and examine whether associations differed between children who resided and/or attended school in historically redlined compared to non-redlined neighborhoods.

Methods:

Children ages 6 to 17 with moderate-to-severe asthma (N=240) from 9 U.S. cities were included. Combined home and school exposure to PM_2.5 and NO₂ was calculated based on geospatially assessed monthly averaged outdoor pollutant concentrations. Repeated measures of CASI and lung function were collected.

Results:

Overall, 37.5% of children resided and/or attended schools in historically redlined neighborhoods. Children in historically redlined neighborhoods had greater exposure to NO₂ (median: 15.4 vs 12.1 ppb) and closer distance to a highway (median: 0.86 vs 1.23 km), compared to those in non-redlined neighborhoods (p<0.01). Overall, PM_2.5 was not associated with asthma severity or lung function. However, among children in redlined neighborhoods, higher PM_2.5 was associated with worse asthma severity (p<0.005). No association was observed between pollutants and lung function or asthma severity among children in non-redlined neighborhoods (p>0.005).

Conclusions:

Our findings highlight the significance of historical redlining and current environmental health disparities among school-aged children with asthma, specifically, the environmental injustice of PM_2.5 exposure and its associations with respiratory health.

Capsule summary

Our findings concerning present-day disparities in environmental exposures and childhood respiratory outcomes in school-aged children with asthma between redlined and non-redlined neighborhoods highlight the urgent need to acknowledge and address the impact of historical racist policies that contribute to current health inequities.

Graphical Abstract

1. Introduction

Studies have shown that predominately Black and Hispanic communities, particularly those of lower socioeconomic status, have higher exposure to traffic-related air pollution (TRAP) such as fine particulate matter (PM_2.5), nitrogen dioxide (NO₂) and black carbon (BC) ^1–4. Furthermore, recent studies report that disparities in air pollution levels have been linked to historical redlining, the place-based, government sponsored discriminatory housing practice of the 1930s^5,6. Redlining refers to the policy of the Home Owner’s Loan Corporation (HOLC) in the United States (U.S.) that characterized neighborhoods into one of four categories. “Grade D” or “redlined” neighborhoods were populated by people of color and systematically denoted as the highest risk (hazardous) for mortgage lending. Historical redlining has perpetuated racial⁷ and economic segregation⁸, social and environmental inequalities⁹, and health disparities^10–15 for residents living in these neighborhoods in present day. Present day disparities in asthma outcomes among adults have also been associated with historical redlining^10–12,14. However, little is known about the impact of historical redlining on the associations between TRAP exposure and respiratory outcomes among children with asthma, who are the most vulnerable to the effects of air pollution¹⁶.

Exposure to TRAP has been associated with the development of asthma, asthma exacerbations, and airway inflammation^17–19. Multi-center cohort studies rely on geospatial modeling of pollution concentrations near a child’s residence to estimate exposure to pollutants²⁰. Given that home and school each represent locations where school-aged children spend most of their time, it is important to consider a combined exposure metric that accounts for both home and school exposure. However, few studies have examined combined home and school exposure to TRAP and the impact on respiratory health outcomes in children^21–24.

In a sample of children ages 6 to 17 with moderate-to-severe asthma from 9 U.S. cities residing in urban pre-specified low income census tracts, our objectives were to: 1) determine associations between combined home and school exposure to PM_2.5 and NO₂ and respiratory outcomes, 2) examine the relationship between historical redlining and air pollutants, and 3) examine whether associations of lung function and pollutants would differ between children who resided and/or attended school in historically redlined neighborhoods compared to those in non-redlined neighborhoods. We hypothesized that children who resided and/or attended school in historically redlined neighborhoods would be exposed to higher levels of PM_2.5 and NO₂, compared to those in non-redlined neighborhoods. Additionally, we hypothesized that the associations between PM_2.5 and NO₂ and respiratory outcomes may be influenced by historical redlining.

2. Methods

2.1. Study Population

Data for this analysis were collected through the Pollution and Lung Health in Active Youth (PLAY) study that recruited participants from the Mechanisms Underlying Asthma Exacerbations Prevented and Persistent with Immune-Based Therapy: A Systems Approach Phase 2 (MUPPITS-2) Study²⁵. MUPPITS-2 was a randomized controlled trial (RCT) of mepolizumab adjunct treatment to reduce asthma exacerbations among children ages 6 to 17 with moderate-to-severe asthma from 9 U.S. cities: Boston, Massachusetts (MA); Chicago, Illinois (IL); Cincinnati, Ohio (OH); Dallas, Texas (TX); Denver, Colorado (CO); Detroit, Michigan (MI); NYC, New York (NY); St. Louis, Missouri (MO); and Washington, D.C. All MUPPITS-2 participants had an asthma diagnosis at least one year prior to RCT recruitment²⁵. Eligible participants were randomly assigned in a 1:1 ratio to undergo blinded monthly injections of mepolizumab or placebo added to guideline-based treatment for 52 weeks from 2017–2021. The trial and subsequent analysis for the current study enrolled participants from the Inner-City Asthma Consortium, which was restricted to children residing in predetermined low-income areas. A total of 240 children with difficult-to-control exacerbation-prone asthma (defined as ≥ 2 exacerbations requiring systematic corticosteroids in the previous year) and blood eosinophils >150 cells/mm³ were included in this analysis (Fig. S1) and completed between one and 6 repeated clinical visits with a total of 920 study visits (Table S1). Thirty-one percent of participants (75/240) completed all six visits. All study procedures were reviewed and approved by the Columbia University Institutional Review Board and informed consent and assent (for 7–18) were obtained at enrollment.

2.2. Estimation of combined home and school air pollutant exposure

Utilizing geographic information systems (GIS, ArcGIS pro), participants’ current home and school addresses were assigned latitude and longitude coordinates. Location-specific ambient air pollution measures (PM_2.5 and NO₂) were collected from the publicly available Air Quality System (AQS) by the Environmental Protection Agency. We used inverse distance weighting to estimate monthly average exposure to pollutants based on location to nearest AQS monitoring station. Ambient ozone [O₃], was also collected alongside PM_2.5 and NO₂ and included as a covariate in our models because of its reactivity with air pollutants and its known association with lower lung function²⁶. The details of estimation of monthly average exposures are provided in the supplementary materials.

We assessed PM_2.5 and NO₂, as a composite estimate of home and school exposure by weighting the concentrations in each location according to the number of hours spent in school per day based on state-specific requirements²⁷. On average (±standard deviation [SD]), children attended school for 7.5 hours (±0.87) daily. Details for calculating the composite estimate of home and school air pollutant exposure as well as the nearest distance to a primary roadway (i.e., highway proximity) are provided in the supplementary materials. Weighted air pollutant concentrations were considered temporal exposure, while highway proximity was treated as chronic exposure, accounting for the number of school hours and days (excluding weekends and holidays).

2.3. Historical redlining

Redlining data were retrieved from the publicly available Mapping Inequality project developed by the University of Richmond which is a compilation of the 1935 Home Owners’ Loan Corporation (HOLC) security maps²⁸. Each current home and school were assigned a letter grade representing the HOLC risk level based on their location relative to the 1935 HOLC maps: A (best), B (still desirable), C (definitely declining), or D (hazardous, also known as redlining). Each home and school were dichotomized as redlined (D or hazardous; n=62 and 71 for home and school, respectively) versus others (A, B, or C; n=178 and 169 for home and school, respectively). Children were further categorized based on whether their home and/or school were in redlined neighborhoods (n=90) versus both being in non-redlined neighborhoods (n=150) for the analysis.

2.4. Respiratory outcome measures

Asthma burden was repeatedly assessed between one and 6 times within the 52-week study period using the previously-validated composite asthma severity index (CASI) that includes asthma symptoms, medication usage, and exacerbation risk²⁹. Spirometry was also assessed by study certified technicians using the American Thoracis Society (ATS) and European Respiratory Society (ERS) guidelines³⁰. Four spirometry outcome measures were included for analysis: forced vital capacity (FVC), forced expiratory volume in one second (FEV₁), FEV₁/FVC, and forced expiratory flow at 25–75% of forced vital capacity (FEF_25–75).

The details of allergic sensitization are provided in the supplementary materials.

2.5. Statistical Analyses

Descriptive analyses used Chi-square, Mann-Whitney, Wilcoxon-rank sum, and Spearman correlation, as appropriate. The averages of repeated measures for combined home and school air pollution estimates, CASI, and lung function were utilized for the descriptive statistics.

We used multivariable linear regression in generalized estimating equation (GEE) models, with robust standard errors to assess associations between air pollution (PM_2.5 and NO₂), CASI, and lung function. Each measurement of lung function (i.e., FVC, FEV₁, FEV₁/FVC, and FEF_25–75) was included in the model as a raw value without normalization to z-score or percent predicted and analyzed as a separate outcome. Three types of models were applied: Model 1) adjusted models controlling for randomization status (treatment with mepolizumab vs placebo), age, sex, race (Black, or African American, vs. non-Black), ethnicity (Hispanic vs non-Hispanic), number of positive skin tests to indoor aeroallergens, current environmental tobacco smoke exposure (ETS; dichotomous variable based on the frequency of exposure to smokers: rarely vs others), study site location, season (three dummy variables for summer, fall and winter), height (used for lung function outcome only), proximity (distance to nearest highway), O_3, and historical redlining (dichotomous variable: home and/or school in historically redlined neighborhoods vs both in non-redlined neighborhoods); Model 2) stratified models by historical redlining, and Model 3) interaction models to test whether the associations between air pollution and respiratory outcomes were modified by historical redlining. A multiplicative interaction term (PM_2.5×historical redlining and NO₂×historical redlining) was included in the adjusted models of PM_2.5 and NO₂, respectively. Coefficients of PM_2.5 and NO₂ (effect estimate, β_adj) are presented for an interquartile range (IQR) increase in concentrations, equivalent to 1.81 μg/m³ and 5.75 ppb for PM_2.5 and NO₂, respectively.

Sensitivity analyses were conducted among those who lived and/or went to school in redlined neighborhoods as follows: 1) reanalysis after removing the upper and lower 5% of PM_2.5 and NO₂; 2) reanalysis after controlling for number of study visits 3) reanalysis after controlling for asthma controller medication usage (a dichotomous variable based on the addition of a long-acting beta agonist (LABA), which is an indicator of worse asthma control: inhaled corticosteroid (ICS) only vs. ICS plus LABA for lung function models only); 4) reanalysis after categorizing historical redlining into two levels (both home and school in redlined neighborhoods [n=43] vs either home or school in redlined neighborhoods [n=47]) to assess for dose response by redlining exposure, 5) reanalysis focused exclusively on the placebo group to eliminate any treatment effects and focused solely on the treatment group, and 6) reanalysis after excluding participants with only one visit (n=8) since one-time spirometry measurement may not be representative of typical best lung function in children, particularly in asthma. We conducted additional sensitivity analyses to assess the impact of study site on the associations between air pollution and respiratory outcomes by excluding children from each city in stratified models by historical redlining.

The significance level was alpha=0.05 for descriptive analyses. To address the issue of multiple testing in our analyses of multiple exposures and health outcomes, we applied Bonferroni correction to adjust the significance threshold for testing each air pollution-respiratory relationship (Bonferroni corrected p-value=0.005). All analyses were performed using SPSS version 26 (Chicago, IL, USA).

3. Results

3.1. Characteristics of participants

Participants were an average of 10.9 years of age (SD = 2.8) and most identified as Black, African American, or Afro-Caribbean (68.8%) and non-Hispanic (74.2%). Among the 240 children with asthma, 40.0% had a BMI > 95^th percentile, and 82.1% had sensitization to at least one indoor aeroallergen (Table 1). Overall, 37.5% (n=90) lived and/or attended schools in historically redlined neighborhoods (Fig. S2). The percentage of home and/or schools located in redlined neighborhoods were variable across the U.S. cities, with the highest percentage in Denver (93%), followed by Cincinnati (58%) and Detroit (52%), and the lowest percentage in Washington, D.C. (0%) (Fig. S2). When compared to non-redlined neighborhoods, children who lived and/or attended schools in historically redlined neighborhoods were, on average, older (Table 1. 12.0 vs 10.3 years old for historical redlining vs non-redlining, respectively; p<0.01), exposed to ETS more frequently (27.8% vs 20.7% for redlining vs non-redlining, respectively; p<0.01), and used more ICS-LABA (70.0% vs 44.0% for redlining vs non-redlining, respectively; p<0.01).

Table 1.

Characteristics of study participants

	Totala (n=240)		Redlined (n=90)	Non-redlined (n=150)	p-valueb
Age (mean, SD)	10.9 (2.8)		12.0 (2.8)	10.3 (2.7)	<0.01
Sex, male (n, %)	134 (55.8)		55 (61.1)	79 (52.7)	0.20
Race (n, %)					0.59
Black	165 (68.8)		60 (66.7)	105 (70.0)
non-Black	75 (31.2)		30 (33.3)	45 (30.0)
Ethnicity (n, %)					0.40
Hispanic	62 (25.8)		26 (28.9)	36 (24.0)
non-Hispanic	178 (74.2)		564 (71.1)	114 (76.0)
ETSc exposure (n, %)	56 (23.3)		25 (27.8)	31 (20.7)	<0.01
BMI z-scored (mean, SD)	1.12 (1.08)		1.25 (1.06)	1.05 (1.10)	0.23
BMI > 95th percentile	96 (40.0)		35 (38.9)	61 (40.7)	0.79
Number of positive indoor aeroallergen skin tests (n, %)
None	43 (17.9)		14 (15.6)	29 (19.3)	0.84
1	38 (15.8)		15 (16.7)	23 (15.3)
2	42 (17.5)		14 (15.6)	28 (18.7)
3	68 (28.3)		26 (28.9)	42 (28.0)
4	49 (20/4)		21 (23.3)	28 (18.7)
ICS plus LABA usagef (n, %)	129 (54.0)		63 (70.0)	66 (44.0)	<0.01
Randomization: Treatment groupg (n, %)	120 (50.0)		39 (43.3)	81 (54.0)	0.11
PM2.5h, median (IQR), μg/m3	8.5 (1.81)		8.9 (2.2)	8.5 (1.5)	0.09
NO2h, median (IQR), ppb	13.3 (5.75)		15.4 (4.3)	12.1 (5.1)	<0.01

^aIncludes only the children that had lung function data and air pollution concentrations (n=201);

^bp-values from Chi-square test presented;

^cEnvironmental tobacco exposure; dichotomous variable based on the frequency of exposure to smokers: rarely vs others

^dWeight (kg)/height (m)²; BMI: Body mass index; SD: standard deviation;

^eGrains per cubic meter; Tree pollen counts were not available n=16 and 34 for redlined and non-redlined neighborhoods, respectively.

^fICS: inhaled corticosteroid; LABA: long-acting beta agonist: fluticasone ≥250 mcg +long-acting beta agonist twice per day;

^gRandomization: Mepolizumab vs Placebo

^hAverage of composite exposures of air pollutants between 1^st and 5^th visits were used for descriptive analysis; Mann-Whitney test performed; Interquartile range (IQR)

There were substantial variations in lung function and home + school PM_2.5 and NO₂ across cities (Table S2). Notably, children from Denver demonstrated greater lung function (e.g., average [± SD] FEV₁: 2.5±1.1vs 2.1±0.6 L/sec for Denver and others, respectively; p=0.19) despite experiencing the highest NO₂ compared to children residing in other cities (Table S2. 19.1±3.6 vs 12.9±3.7 ppb for Denver and all other sites, respectively; p<0.001). Similarly, children residing in NYC had higher NO₂ levels (Table S2. Average± SD: 16.6±2.0 vs 12.6±3.9 ppb for NYC and others, respectively), and shorter distance to highways (NYC vs others: 0.6±0.3 vs 1.8±1.7 km; p<0.001), compared to those in other cities, yet had better average levels of FEV₁/FVC (0.81±0.06 vs 0.75±0.08; p<0.001) and FEF_25–75 (2.2±0.7 vs .1.8±0.9 L/sec; p<0.001).

3.2. Composite estimates of air pollution: Effect of historical redlining

Overall, there was no significant difference between home vs. school exposure to outdoor PM_2.5 or NO₂ (Table S3). When analyzing composite estimates of air pollution, children in redlined neighborhoods were exposed to 27% higher levels of outdoor NO₂, compared to those in non-redlined neighborhoods (Table 1 and Fig. 1. Median: 15.4 vs 12.1 ppb for redlined and non-redlined, respectively, p<0.01). PM_2.5 levels were slightly higher in redlined, compared to non-redlined neighborhoods although this difference was not statistically significant (Table 1 and Fig. 1. median: 0.89 vs 0.85 μg/m³ for redlined and non-redlined, respectively; p=0.09). Importantly, children in historically redlined neighborhoods had closer proximity to highways, compared to those in non-redlined neighborhoods (Fig. S3. Median: 0.86 km vs 1.23 km for historically redlined vs non-redlined, respectively; p<0.01).

Fig. 1.

Comparison of composite estimates of PM_2.5 and NO₂ exposure between children who resided and/or attended school in redlined neighborhoods vs those in non-redlined neighborhoods: a) PM_2.5, and b) NO₂ Violin plots with median values presented. Each black dot represents the average PM_2.5 and NO₂ exposure concentration for each individual participant. A box plot is used to show the minimum, first quartile, median, third quartile, and maximum. The width of the violin plot indicates the density of data (density plot) with wider regions indicating values that occur more frequently. Mann-Whitney test performed with p-value presented.

Composite estimates of PM_2.5 were not significantly correlated with NO₂ (p>0.05); however, both PM_2.5 and NO₂ were inversely correlated with proximity (r=−0.14 and −0.21 for PM_2.5 and NO₂, respectively, p<0.05). Moreover, when stratified by historical redlining, the significant correlations between air pollutants and proximity were observed only in redlined neighborhoods (r=−0.24 and −0.22 for PM_2.5 and NO₂, respectively, p<0.05), but not in non-redlined neighborhoods (p>0.05).

3.3. Associations between composite estimates of air pollution and respiratory outcomes: effect modification by historical redlining

In linear regression models, no significant associations were observed between the PM_2.5 and CASI or lung function in the overall sample (Table S4). When stratified by historical redlining, among children who lived and/or went to schools in historically redlined neighborhoods, higher levels of PM_2.5 were significantly associated with higher CASI, indicating greater asthma severity (Table 2 and Fig. 2. adjusted-p<0.005). Additionally, in these children, we found links between elevated PM_2.5 levels and lower lung function (i.e., FEV₁, FEV₁/FVC, and FEF_25–75). Yet, the significance of the findings diminished after applying Bonferroni correction for multiple comparisons. I n non-redlined neighborhoods, no association was observed between PM_2.5 and respiratory outcomes. A test for effect modification showed nonsignificant interaction between PM_2.5 and historical redlining (Table 2 and Fig. 2. p-value for interaction> 0.05 for all).

Table 2.

Associations between combined home and school traffic-related pollution exposure and respiratory outcomes: the effect of historical redlining

	Adjusteda βadj (95% CI)				p-value for interaction
	Redlined Nsubjects/Nvisits=90/382		Non-redlined Nsubjects/Nvisits=150/580
Respiratory outcomes	PM2.5	NO2	PM2.5	NO2	PM2.5	NO2
CASI	0.43* (0.15, 0.71)	0.18 (−0.24, 0.60)	0.05 (−0.19, 0.28)	0.21 (−0.27, 0.69)	0.22	0.59
FVC, L	−0.03 (−0.08, 0.02)	0.04 (−0.05, 0.13)	0.01 (−0.02, 0.03)	−0.03 (−0.10, 0.03)	0.29	0.22
FEV1, L/sec	−0.05 (−0.10, −0.01)	0.08 (−0.02, 0.18)	0.00 (−0.03, 0.03)	0.01 (−0.05, 0.07)	0.23	0.26
FEV1/FVC	−0.01 (−0.02, −0.003)	0.01 (−0.01, 0.03)	0.00 (−0.07, 0.07)	0.02 (0.001, 0.03)	0.51	0.59
FEF25–75, L/sec	−0.12 (−0.20, −0.03)	0.15 (−0.07, 0.37)	0.01 (−0.05, 0.06)	0.12 (0.01, 0.23)	0.12	0.59

CASI: the Composite Asthma Severity Index; forced expiratory volume in one second (FEV₁), FEV₁/FVC, and forced expiratory flow at 25–75% of forced vital capacity (FEF_25–75); N_subjects, number of subjects included for the analysis; n_visits, total number of visits;

Coefficients of PM_2.5 and NO₂ (effect estimate, β_adj) are presented for an interquartile range (IQR) increase in concentrations, equivalent to 1.81 μg/m³ and 5.75 ppb for PM_2.5 and NO₂, respectively;

^aAdjusted for randomization status, age, sex, race, ethnicity (Hispanic vs non-Hispanic), the number of positive indoor aeroallergen skin tests, current ETS exposure, study site location, season (three dummy variables for summer, fall and winter), height (lung function outcome only), highway proximity, and O₃.

^*p-value<0.005.

Fig. 2.

Effect modification of historical redlining on the association between air pollution and respiratory outcomes among children with asthma: a) PM_2.5 and b) NO₂ Coefficients of PM_2.55 and NO₂ (effect estimate, β_adj) with 95% CI are presented for an interquartile range (IQR) increase in PM_2.5 concentrations, equivalent to 1.81 μg/m³ and 5.75 ppb for PM_2.5 and NO₂, respectively; Models adjusted for randomization status, age, sex, race, ethnicity (Hispanic vs non-Hispanic), the number of positive sensitizations to indoor aeroallergens, current ETS exposure, study site location, season (three dummy variables for summer, fall and winter), height, proximity and O₃; *p-value<0.005; P_interaction: p-values for multiplicative interaction

Despite the lack of statistical significance, we observed a positive overall association between NO₂ and lung function (Table S4. β_adj [95% CI]: 0.07 [0.02, 0.10], 0.02 [0.01, 0.03], and 0.13 [0.02, 0.24] for FEV₁, FEV₁/FVC and FEF_25–75, respectively). However, there were no significant associations between NO₂ and respiratory outcomes when stratified by redlining (Table 2. p>0.005).

3.4. Sensitivity analyses

The results of the sensitivity analyses in redlined neighborhoods are shown in Table S5. First, when the top or bottom 5% of PM_2.5 or NO₂ were removed, a significant association between PM_2.5 and CASI was replicated with a larger effect estimate. Notably, the positive association between NO₂ and lung function outcomes shifted to negative (as hypothesized), upon removing the top 5% of NO₂. Second, after accounting for the number of visits, similar associations between PM_2.5 and respiratory outcomes presented in Fig. 2 and Table 2 persisted. Third, after adjusting for ICS-LABA use, associations between PM_2.5 and lung function remained consistent. Fourth, when comparing two levels of redlining categories (‘both home and school’ vs ‘either home or school’ in redlined neighborhoods), the effect estimates of the associations between PM_2.5 and lung function were greater among children who both resided and went to school in redlined neighborhoods, though not statistically significant. Fifth, when the analyses focused exclusively on the placebo group, the associations between PM_2.5 and respiratory outcomes strengthened, achieving statistical significance in the PM_2.5-FEF_25–75 relationship (Table S5. β_adj [95% CI]: −0.17 [−0.28, −0.07], p<0.001). This trend was not observed with the treatment group, suggesting that the primary findings shown in Table 2 were predominantly driven by children in the placebo group. Lastly, the findings from the analysis focused on participants with multiple visits (≥2) remained consistent with the main results.

The results of sensitivity analyses to evaluate the impact of study site on the association between air pollutants and respiratory outcomes by historical redlining are presented in Fig S4 and Table S6. Notably, excluding participants from Denver strengthened the relationship between PM_2.5 and CASI score (Fig S4 and Table S6. β_adj [95% CI]: 0.55 [0.23, 0.88], p<0.001), and shifted positive association between NO₂ and lung function outcomes to negative, although not statistically significant (Fig S4 and Table S6). Exclusion of participants from NYC or Detroit rendered the associations of PM_2.5 with FEV₁/FVC, and FEF_25–75 significant (Fig S4 and Table S6). However, regardless of study site, there were no associations between air pollution and respiratory outcomes.

4. Discussion

In our sample of 240 children with exacerbation-prone asthma across 9 U.S. cities, we observed that school-aged children who resided and/or attended schools in historically redlined neighborhoods were exposed to higher levels of TRAP, as indicated by higher levels of NO₂, and closer distance to highways, compared to those in non-redlined neighborhoods. Among children who lived and/or attended school in redlined neighborhoods, greater exposure to PM_2.5 was associated with increased asthma severity as indicated by CASI. Overall, we observed significant differences in present day air pollution exposures that are likely attributed to the historical practice of redlining and contribute to the respiratory health of children with asthma residing and/or attending school in redlined communities in present day.

To our knowledge, this is the initial investigation into the impact of historical residential redlining, a discriminatory policy prevalent from the 1930s to the 1970s, among children in general and those with asthma demonstrating associations between composite estimates of air pollution, lung function and asthma severity. The impact of historical policies that contribute to unequal burden of air pollution exposure and stress for children has been underexplored; yet, children may be most susceptible to the effects of air pollution given that their lungs are still developing,³¹. Several studies have reported respiratory health disparities associated with historical redlining among adults.^11,14,32 For example, studies revealed that living in redlined neighborhoods was associated with asthma-related emergency department visits in California,¹¹ uncontrolled and/or severe asthma in Pennsylvania¹⁴ and current day asthma prevalence across U.S. cities³². Our findings among school-aged children with asthma of associated potential reductions of lung function and increased exacerbations linked to higher levels of PM_2.5 in redlined neighborhoods, thus, extends this knowledge. Furthermore, given no observed correlations of PM_2.5 with NO₂ (primary marker of traffic emission) and minimal correlation with distance to a highway, our measures of PM_2.5 likely captured additional point sources of pollutants that are more common in historically redlined neighborhoods including industrial emissions and railroads^6,33.

In order to better understand the potential mechanisms underlying the links between historical redlining and current health disparities, it is crucial to consider various factors associated with historical redlining, including racial and ethnic inequities, socioeconomic factors, neighborhood characteristics, environmental pollution, and built environments. For instance, our previous study highlighted that schools located in historically redlined neighborhoods in NYC had a higher percentage of Black residents, greater community deprivation, lower levels of resilience, and reduced childhood opportunities in present day⁵. Although research on air pollution disparities related to historical redlining is limited, recent studies have demonstrated associations between worsening HOLC grades and increased levels of PM_2.5, NO₂, BC, and sulfur dioxide (SO₂)^5,6,14, as well as a higher number of industrial emission sources, primary roads⁶, and diesel exhaust particle (DEP) emissions¹¹ across U.S. cities. Our study adds to the existing evidence by highlighting disparities in current environmental exposures experienced by children with exacerbation prone asthma across nine U.S. cities. Beyond air pollution disparities, limited access to resources that include quality healthcare, healthy food options, green spaces, and opportunities for children, further contribute to health disparities associated with historical redlining^5,6,9,34,35. Additionally, poorly constructed and maintained public housing projects that have historically been placed in redlined communities, experience greater indoor pollutant exposure due to poor ventilation that is also linked to asthma.³⁶ Furthermore, there is a growing body of literature demonstrating the effects of allostatic load, or the wear and tear on the body due to chronic stress, and the impacts on health ^37,38. Studies have demonstrated that increased allostatic load can be measured in children and may make them more susceptible to environmental insults leading to poor respiratory outcomes^39–41. For example, a case study of adolescents using a composite allostatic load index based on eight biomarkers (e.g., total cholesterol, glucose, and cortisol) found that higher allostatic load was associated with a greater risk of asthma prevalence or onset among boys⁴¹. Thus, a plausible explanation for the observed findings between pollutants and respiratory outcomes could be that children who live and/or attend school in historically redlined neighborhoods experience greater allostatic load making them more susceptible to the effects of air pollution. We did not measure allostatic load in our cohort of children, which is a limitation. However, understanding the influence of allostatic load in communities that have been historically marginalized and better understanding of the impact on the relationship between environment and health is an important future direction for this field of research. Nevertheless, allostatic load measurements may still not account for other well-known drivers of asthma morbidity, including viral infections, allergen exposures, medication adherence, all of which may be influenced by adverse social determinants of health (SDOH). These adverse SDOH are likely to be more significant in redlined neighborhoods and strongly influence pediatric asthma^42,43.

The unexpected positive association between NO₂ and various lung function metrics persisted across analyses despite not reaching statistical significance. For instance, our sensitivity analyses, which involved excluding each city, revealed that associations between NO₂ and lung function could be attributed to children from Denver, CO. Despite having the highest levels of NO₂ in our cohort, children from Denver exhibited greater lung function compared to those in other cities, consistent with the known observation that high altitude results in larger lung volumes⁴⁴. Notably, the majority of children in Denver (93%) resided and/or attended school in redlined neighborhoods. Upon removing Denver from the analysis, associations between PM_2.5 and CASI were strengthened and the direction of the association between NO₂ and lung function shifted from positive to negative in redlined neighborhoods. Thus, Denver represents a unique city with a more nuanced relationship between redlining, air pollution and respiratory outcomes. Excluding children from NYC or Detroit also resulted in significant associations between PM_2.5 and FEV₁/FVC as well as FEF_25–75 in redlined neighborhoods. One potential explanation for the observed results could be due to differences in environmental exposures and socioeconomic factors between cities. For example, children from NYC in redlined neighborhoods experienced higher NO₂ levels and a shorter distance to a highway yet exhibited better FEV₁/FVC and FEF_25–75, compared to other cities. Additionally, different socioeconomic conditions and access to healthcare could further contribute to differences in respiratory outcomes among children in different cities⁴⁵. Such variations underscore the complex interplay between historical redlining, air pollution, and respiratory outcomes.

In the current study, we focused on composite estimates of home and school outdoor exposure to air pollutants. Various methods of assessing air pollution exposure among school-aged children have been interrogated in prior studies including: 6-day residential indoor PM_2.5 measurements¹⁹, weekly average NO₂ exposure in classrooms⁴⁶, proximity of children’s residences to the nearest major road²², and the total length of roads within 100-m or 200-m buffer at the participants’ residences⁴⁷. These prior studies indicate that both home and school environments are important sources of TRAP exposure for children. However, studies that consider integrated measures of both home and school exposures in relation to respiratory health outcomes are limited. For instance, a European birth cohort study involving 10-year-old children with persistent respiratory symptoms (e.g., wheezing and/or use of medication) utilized weekly BC estimates, accounting for both home and school exposure based on time spent in each environment²³. Their findings indicated an association between higher BC levels and increased airway inflammation, assessed by fractional exhaled nitric oxide (FeNO). Similarly, a study involving kindergarten and first grade children employed an annual estimate of combined home and school TRAP exposure²¹. The study suggested a stronger association between new-onset asthma and the combined exposure, compared to exposure in either the home or school alone, emphasizing the contribution of both home and school environments to the risk of asthma. In comparison, a large study of 12-year-old South Korean children did not find any associations between combined home and school exposure, assessed by proximity to the nearest major road or density of major roads within 300 meters, and asthma prevalence²². A recent study utilized a composite measure of both home and school traffic exposure, and found that major roadway proximity was associated with increased asthma symptoms, reported health care use, and poor asthma control, but not with lung function among school-aged children with asthma²⁴. Our study extends this approach by incorporating time spent in school, based on state-specific requirements, and home environments to derive composite estimates of PM_2.5 and NO₂, reaffirming the significance of combined TRAP effects in home and school environments on respiratory health.

We acknowledge several limitations of our study. First, the sample size was relatively small, especially when stratifying into 3 groups of both, either, and neither home and school in redlined neighborhoods. However, sensitivity analyses showed that individuals who both resided and attended school in redlined neighborhoods had higher effect estimates for PM_2.5-FEV₁ or PM_2.5-FEF_25–75 relationships, despite their lack of statistical significance. Second, this is a selected population of high risk children with asthma who are more susceptible to the effects of air pollutants³¹. Therefore, our findings may not be generalizable to children who do not have an underlying diagnosis of asthma. Third, our approach to estimating ambient PM_2.5 and NO₂ concentrations at participant’s home and school addresses using EPA regulatory monitor data may result in exposure misclassification. Regulatory monitors may not be uniformly distributed within the neighborhoods across the nine cities studied, which could particularly affect NO₂ measurements due to its rapid diminishment with distance from its source, such as vehicular traffic. To mitigate, though not eliminate this issue, we included highway proximity as a covariate in the models. Furthermore, while we attempted to capture exposures both at home and school in a composite measure, the approach is relatively new. Future studies are needed to confirm the validity of this composite metric. Also, we were not able to account for changes in homes or schools in our study, which could further contribute to exposure misclassification. Fourth, the absence of time-activity data or details on housing quality could further amplify exposure misclassification and influence lung function ^48–50. Children typically spend more time indoors than outdoors both in home and the school settings, potentially leading to increased exposure to indoor pollutants compared to outdoor pollutants. This is especially relevant for obese children, which comprise 40% of our study participants. Research has shown that obesity is associated with less outdoor play time and increased hours of television watching⁵¹, further emphasizing the likelihood of increased indoor exposure. Moreover, poor housing quality, often characterized by inadequate ventilation, and overcrowding can contribute to poor indoor quality (e.g., air pollutants, pest, mold, and ETS etc)⁵⁰. These, in turn, negatively affect respiratory health in children. Fifth, despite our adjustment for allergic sensitization, we were not able to control for other pro-inflammatory exposures such as indoor aeroallergens (e.g., dust mites, rodents, and molds) at home and school. Children are exposed to clinically significant levels of indoor aeroallergens not only within their homes but also in settings like schools and daycare centers⁵². Sixth, given the complexities and nuances in the relationship between NO₂ and respiratory health, future research is warranted to further explore this relationship, considering other TRAP such as BC and other proxies of air pollution. Finally, we were unable to account for gentrification, a process involving the displacement of long-term residents that often occurs in many redlined neighborhoods. One study demonstrated that gentrification is linked to changes in redlined neighborhoods, economic inequality, and segregation,⁸ that could influence present-day air pollutant concentrations and have subsequent impacts on health. Despite these limitations, our sensitivity analyses demonstrated the robustness of the associations between PM_2.5 and asthma exacerbation in redlined neighborhoods. Furthermore, the robustness of this study is enhanced by employing repeated measures of home and school air pollution exposure along with respiratory outcomes over 52 weeks, spanning up to 6 visits.

5. Conclusions

Children with exacerbation-prone moderate-to-severe asthma, residing and/or attending school in historically redlined neighborhoods exhibited increased asthma severity associated with combined home and school PM_2.5 exposure. An association between PM_2.5 and lower lung function was also observed among those in redlined communities, but only when children from NYC or Detroit were excluded from the stratified analysis. Given the results from the sensitivity analysis excluding Denver, intervention studies focusing on improving lung volumes through physical therapy or exercise could offer valuable insights into effective public health strategies for better respiratory health. Our findings concerning present-day disparities in environmental exposures and childhood respiratory outcomes between redlined and non-redlined neighborhoods highlight the urgent need to acknowledge and address the impact of historical racist policies that contribute to current health inequities.

Key messages:

• We observed significant differences in present day air pollution exposures (e.g., higher levels of NO₂ and closer proximity to highways) by historical redlining.

• Among children who lived and/or attended school in redlined neighborhoods, higher levels of PM_2.5 exposure were significantly associated with increased asthma severity as indicated by CASI.

• We present evidence that the historical practice of redlining contributes to the negative respiratory health of children residing and/or attending school associated with PM_2.5 exposure in present day.

Abbreviations:

AQS

Air Quality System

BC

black carbon

CASI

Composite Asthma Severity Index

BMI

Body mass index

DEP

Diesel exhaust particle

FEF_25–75

Forced Expiratory Flow at 25–75% of Forced vital capacity

FeNO

Fractional exhaled nitric oxide

FEV₁

Forced expiratory volume in one second

FVC

Forced vital capacity

GEE

Generalized estimating equation

GIS

Geographic information systems

HOLC

Home Owners’ Loan Corporation

ICS

Inhaled corticosteroid

NAB

National Allergy Bureau

LABA

Long-acting beta agonist

LUR

Land use regression

MUPPITS

Mechanisms Underlying Asthma Exacerbations Prevented and Persistent with Immune-Based Therapy

NYC

New York City

NO₂

Nitrogen dioxide

O₃

Ozone

PM

Particular matter

PLAY

Pollution and Lung Health in Active Youth

PFTs

Pulmonary function tests

RCT

Randomized controlled trial

SD

standard deviation

SDOH

Social determinants of health

SO₂

Sulfur dioxide

TRAP

Traffic-related air pollution