Housing mobility program effects on ambient fine particulate matter exposure and childhood asthma morbidity

Abstract

Rationale

Air pollution in high-poverty neighborhoods contributes to disparities in asthma morbidity. Housing mobility programs help families move to better-resourced neighborhoods and improve asthma outcomes. Whether housing mobility-related improvements in asthma morbidity are explained by reductions in air pollution is unknown.

Objectives

Evaluate change in concentrations of fine particulate matter and its components and associated changes in asthma morbidity among a cohort of children with asthma who moved out of high-poverty neighborhoods.

Measurements and Main Results

Caregiver-reported asthma exacerbations, symptoms, and asthma control test scores were examined among 123 children (age 5–17) with persistent asthma participating in the Baltimore Regional Housing Partnership housing mobility program. Exposures considered are annual average concentrations of PM_2.5 and major PM_2.5 components (black carbon, organic carbon, ammonium, nitrate, and sulfate). Children moved to neighborhoods with 0.64 μg/m³ [95% CI: 0.50 – 0.82] lower PM_2.5 concentrations, with greatest reductions in black carbon and organic carbon. One standard deviation higher exposure to PM_2.5 was associated with a 60% [95% CI: 21%–112%] higher odds of exacerbations and 20% [14%–27%] higher odds of symptom days/2 weeks. Reductions in PM_2.5 exposure explained 77% [39% - 100%] and 22% [16%–36%] of moving-related reductions in exacerbations and symptom days, respectively. Some PM components were also associated with asthma exacerbations and symptoms and showed mediating effects.

Conclusions

Moving from high-poverty neighborhoods to better-resourced neighborhoods through a housing mobility program is associated with lower air pollution exposure, contributing to moving-related asthma improvements. Housing mobility may be a useful complement to pollution-focused environmental justice interventions.

Introduction

Neighborhood-level factors, including environmental exposures and psychosocial stressors, contribute to morbidity among children with asthma, and inequality in these neighborhood conditions contributes to racial disparities in childhood asthma (1–4). The Mobility Asthma Project (MAP) followed children with asthma whose families were provided assistance to move from disadvantaged to better-resourced neighborhoods as the result of a Fair Housing lawsuit (5). Children in the MAP cohort showed significant improvements in exacerbations and symptoms, attributable in part to improvements in neighborhood-related psychosocial stressors such as perceived neighborhood safety, social cohesion, and urban stress (4).

Acute exposure to outdoor air pollution, and in particular fine particulate matter (PM_2.5), is an established cause of asthma morbidity (6). In addition, recent work has shown differences in chronic PM_2.5 exposure as low as ~0.7 μg to be associated with population-based rates of asthma-related emergency department visits (7). Evidence also suggests that specific chemical composition of PM_2.5 is associated with higher PM_2.5 toxicity (8, 9). Concentrations of PM_2.5 components differs from spatial variation in total PM_2.5 in urban areas, as components are associated with specific sources and certain sources are systematically located closer to disadvantaged neighborhoods (10–13). It is unknown, however, whether moving to neighborhoods with lower chronic exposure to PM_2.5 leads to improvements in asthma morbidity, and the extent to which any such improvements are related to specific PM_2.5 components.

We leveraged the MAP cohort to investigate whether moving from a disadvantaged neighborhood to a better-resourced neighborhood through a housing mobility program lowered PM_2.5 exposure overall and for individual components, if exposure was associated with changes in asthma morbidity, and to what extent the change may mediate the asthma benefits of moving.

Methods

Study population

The study was conducted in the Baltimore metropolitan region, including Baltimore City and surrounding counties. As described previously by Pollack et al. (2023), MAP followed 140 children with asthma recruited from 2016 to 2019 whose families were part of a non-profit housing voucher program (Baltimore Regional Housing Partnership, BRHP) which is designed to support families to move from areas of high poverty to better-resourced “opportunity neighborhoods.” Pre-move neighborhoods were characterized by a median of 26% of families below the poverty level, compared to 4% after moving; only 1 participant moved to a high poverty neighborhood (defined by at least 20% of families below the poverty line). Inclusion criteria were age 5–17 years and persistent asthma or an exacerbation within the previous 12 months. This analysis of 123 MAP children excluded those who did not receive a voucher or move after enrollment (n=17). Follow-up continued for up to 12 months after moving, truncated in March 2020 due COVID-19. Further details about inclusion criteria have been described previously (4). Caregivers provided written informed consent and child participants provided assent to participate in the study. The study was approved by the Johns Hopkins University School of Medicine Institutional Review Board.

Asthma outcomes

Asthma exacerbation and symptom data were collected via caregiver questionnaire at home visits conducted every six months and telephone interviews at three-month points between visits. Exacerbations were defined as a systemic corticosteroid burst in the past three months. Maximum symptom days, a validated composite measure, were calculated as the maximum number of days in the past two weeks that the participant had (1) cough, wheezing, shortness of breath, or tightness in chest; (2) slowed activities due to asthma symptoms; or (3) nocturnal awakening due to asthma symptoms. Asthma control was assessed at home visits via the Asthma Control Test (ACT), with a score ≤19 indicating uncontrolled asthma (14).

Exposure assessment

Annual average concentrations were used to reflect air pollution as a structural determinant of health, representing “typical” neighborhood concentrations. Fine particulate matter (PM_2.5) estimates for years 2016–2019 were provided by the Atmospheric Composition Analysis Group (ACAG) at 1 km² spatial resolution (15). Total PM_2.5, expressed as mass concentration, consists of organic and inorganic components, with major components comprising black carbon (BC), organic carbon (OC), sulfate (SO₄), nitrate (NO₃), and ammonium (NH₄). Component concentration estimates for years 2016–2019 were produced by Amini et al. via ensemble machine learning with a resolution of 50×50 m within the census-defined Baltimore urban area and 1×1 km in the rest of the domain (16, 17). Validation of total PM_2.5 estimates from ACAG using 10-fold cross-validation showed R²=0.61, while PM_2.5 component models from Amini et al. achieved R² values of 0.91 for BC, 0.86 for OC, 0.96 for SO₄, 0.88 for NO₃, and 0.92 for NH₄ within the test set, with RMSE ranging from 0.13–0.37 μg/m³. Annual exposure estimates were assigned based on geocoded participant address (“residence”) as the value of the grid cell containing the address for 1 km² resolution data or the average value of all grid cells within a 500m buffer for higher resolution data. Multiple addresses occasionally fell within the same 1 km grid cell, as some pre-move homes were in close proximity (81 [106] of 122 pre-move residences had at least one neighbor within 500 m [1 km]; 57 [70] of 117 post-move residences). Thus, pollution comparisons reflect the set of individual exposures and not a spatially balanced sampling of Baltimore City.

Participant and neighborhood characteristics

Baseline characteristics including age, sex, and racialized group were collected at the initial home visit. Asthma controller medication treatment step was collected at each visit and phone interview. Neighborhood-related stress was assessed using five selected items from a validated questionnaire designed to assess urban stress (scale of 5–25, with a higher score indicating more stress). The child opportunity index (COI), one of the indicators used to define “opportunity neighborhoods,” was used to describe neighborhood socioeconomic status. COI, calculated at the census tract level, is a composite measure scored from 1 to 100 based on indicators in three domains (education, health and environment, and social and economic), with higher scores indicating greater opportunity (18).

Statistical Approach

We used annual-average pollution surfaces in several ways to address three questions: (1) whether moving to better-resourced neighborhoods was associated with lower outdoor concentrations of total PM_2.5 and its major components, (2) whether living in areas with lower PM_2.5 and component concentrations was associated with better asthma outcomes, and (3) whether the change in neighborhood air pollution mediated the asthma benefits of moving.

To answer question 1, and as illustrated in Figure S1, annual-average concentrations were used to describe typical long-term neighborhood differences between the sets of pre-move and post-move addresses (contextual spatial contrast); and to quantify an individual-level counterfactual spatial contrast by comparing each participant’s pre- and post-move addresses in a single common calendar year. These analyses were descriptive and did not incorporate health outcomes. The second and third questions were addressed by assigning the calendar-year annual average at the participant’s residence for each visit or call. All analyses were conducted using Stata version 15.1 (StataCorp).

We quantified whether the move to “opportunity neighborhoods” lowered exposure in two ways. First, we compared mean concentrations across all pre-move residences (n = 122) and post-move residences (n = 117) for 2016–2019 to characterize long-term neighborhood differences, allowing multiple addresses per participant. Second, we estimated the difference between each participant’s pre- and post-move addresses in 2019 to assess the counterfactual contrast had they not moved. Because all participants had moved by 2019, this comparison holds year constant and reflects spatial differences between addresses. This counterfactual analysis was limited to 93 participants with both addresses. Missingness arose from 13 enrolled after moving, 10 missing records of post-move address, and 7 excluded due to COVID-19 truncation. Both comparisons used Welch two-sample t-tests.

To examine associations between annual concentrations and asthma morbidity we fit cross-sectional regression models separate for each exposure and each asthma outcome, using generalized estimating equations to account for within-individual correlation. Pollution concentrations were assigned as the calendar-year annual average at the participant’s residence at the time of each visit or call. Models adjusted for move status (pre/post), age, sex, and calendar year of assessment. Analyses used all available visit-level observations with non-missing outcome and exposure information; observations missing either were excluded only for that analysis. Models used a log link function for exacerbations, a binomial distribution with 14 trials for maximum symptom days. Odds ratios represent change in odds of an asthma exacerbation in the past three months or maximum symptom days in the previous 14 days for a one standard deviation change in PM_2.5 or component concentration. ACT was modeled as a continuous variable using a Gaussian linear model. In sensitivity analyses to further assess the threat of confounding of air pollution-asthma associations by stress not captured by adjustment for move, we examined attenuation of associations between air pollution and asthma outcomes after additionally controlling for stress. We focused on the modified urban stress scale (“stress”), which was highly correlated with other stress measures, improved with moving, and had the greatest mediating effect (35% and 22% of the beneficial effect of moving on asthma exacerbations and symptom days, respectively) of previously examined factors (4).

Last, we estimated the contribution of reductions in exposure to improvements in asthma that occurred with moving using the Hicks and Tingley approach and medeff command in Stata.(19) The primary estimand of interest was the natural indirect effect of the treatment (i.e., moving) on asthma-related outcomes via the mediator.(20) In particular, we estimated the proportion of the total effect of moving on exacerbations, symptom days, and ACT score that could be attributed PM_2.5 and components.

Results

Approximately half of the cohort was male (52.8%) and the average age at baseline was 8.9 years old (Table 1). Participants reported experiencing at least 1 exacerbation in the last 3 months at 15% of pre-move visits, reported >5 symptom days over the past two weeks at 39% of pre-move visits, and participant ACT score indicated uncontrolled asthma at 50% of pre-move visits. In contrast, participants reported experiencing at least 1 exacerbation in the last 3 months at 8% of post-move visits, >5 symptom days over the past two weeks at 17% of post-move visits, and 27% of ACT scores indicated uncontrolled asthma at post-move visits. Mean urban stress scores were 12.7 out of 25 across pre-move visits versus 8.1 across post-move visits. Pre-move neighborhoods had an average COI score of 5.6 out of 100, increasing to 51.2 in post-move neighborhoods.

Table 1.

Cohort characteristics and comparison of pre- and post-move asthma and neighborhood characteristics.

Baseline Characteristic
N children	N = 123
Age (y), mean (SD)	8.9 (2.9)
Sex, n (%)
Male	65 (52.8%)
Female	58 (47.2%)
Asthma characteristics	Pre-move	Post-move	p-value
N (visits and phone interviews)	N = 306c	N = 425d
Treatment step, mean (SD) N (%)	1.38 (1.94)	1.28 (1.87)	0.48a
0–3	231 (75%)	329 (77%)
4–6	75 (25%)	96 (23%)
Exacerbation in last 3 months, N (%)
None	259 (85%)	386 (92%)	0.006b
At least 1	46 (15%)	36 (8%)
Symptom Days past 14 days, mean (SD) n (%)	5.1 (5.0)	2.7 (3.8)	<0.001a
0–4 days	187 (61%)	351 (83%)
5–14 days	119 (39%)	74 (17%)
Asthma control
N (visits only)	N = 107	N = 211
Asthma control test, mean (SD) n (%)	18.9 (4.3)	21.7 (3.5)	0.004a
20+	85 (50%)	153 (73%)
≤19	85 (50%)	58 (27%)
Neighborhood characteristics	Pre-move	Post-move
Modified urban stress scale (5–25)
N	N = 130	N=159	<0.001a
mean (SD)	12.7 (4.1)	8.1 (2.7)
Child opportunity index (0–100)e
N (visits and phone interviews)	N = 306	N = 425	<0.001a
mean (SD)	5.63 (9.3)	51.2 (17.0)

p-values are based on

^aStudent’s t-test, and

^bchi-squared test.

^cone observation is missing for exacerbations (N=305)

^dthree observations are missing for exacerbations (N=422)

^eCOI is used descriptively to characterize neighborhoods and is not included as a covariate in exposure–outcome or mediation models.

Total PM_2.5 and components at pre- and post-move residences

Concentrations of total PM_2.5 averaged through years 2016–2019 across post-move residences were lower than across pre-move residences by 0.65 (95% CI: −0.82 - −0.50) μg/m³ (Table 2). Concentrations of individual PM_2.5 components were lower at post-move residences to varying degrees, with greater differences in both mass and relative share of OC and BC.

Table 2. Mean (SD) annual PM2.5 concentrations (2016–2019), summarized across years, across residences, grouped by pre-move and post-move addresses.

This table characterizes typical neighborhood concentration profiles of the address sets and is not intended to represent exposure accrued by participants prior to moving.

PM component		All residences	Pre-move (n=122)	Post-move (n=117)	Difference by neighborhood type [95% CI]	p-valueb
Total PM2.5	Mean (SD)	7.72 (0.73)	8.00 (0.78)	7.34 (0.46)	−0.65 [−0.82 – −0.501	< 0.001
BC	Mean (SD)	0.62 (0.07)	0.65 (0.05)	0.58 (0.06)	−0.07 [−0.09 – −0.061	< 0.001
	Mean sharea (%)		13%	12%
OC	Mean (SD)	2.00 (0.22)	2.12 (0.11)	1.87 (0.12)	−0.25 [−0.28 – −0.221	< 0.001
	Mean sharea (%)		41%	39%
NO3	Mean (SD)	0.91 (0.09)	0.95 (0.04)	0.87 (0.09)	−0.08 [−0.09 – −0.061	< 0.001
	Mean sharea (%)		18%	18%
SO4	Mean (SD)	1.09 (0.22)	1.10 (0.03)	1.08 (0.04)	−0.01 [−0.02 – 0.001	0.01
	Mean sharea (%)		21%	23%
NH4	Mean (SD)	0.40 (0.06)	0.41 (0.04)	0.39 (0.04)	−0.02 [−0.03 – −0.021	< 0.001
	Mean sharea (%)		8%	8%

^a“Mean share” refers to the percentage of total PM_2.5 mass comprised of the individual component.

^bp-values calculated using Welch two sample t-test.

BC=black carbon; OC=organic carbon; NO₃=nitrate; SO₄=sulfate; NH₄=ammonium

Among participants with valid pre- and post-move address records (n=93), the observed 2019 post-move total PM_2.5 exposure was lower by a mean (sd) of 0.64 (1.02) μg/m³ (95% CI: −0.88 - −0.48) relative to the counterfactual 2019 concentration at pre-move addresses. Exposure did not improve for all participants: the interquartile range across all participants was −1.4 to 0.1 μg/m³, with 25% experiencing higher PM_2.5 concentrations than they would have had they had not moved and another 3% experiencing reductions <0.1 μg/m³ (Figure 1). Differences in component concentrations are shown in Table S1.

Figure 1. Distribution of difference in concentration for total PM_2.5 at pre-move and post-move addresses in the year 2019.

Figure reflects share of participants with valid pre- and post-move address records (n=93). The median (25^th-75^th%ile) for difference in PM_2.5 was −0.80 μg/m³ (−1.42–0.08).

Associations of PM_2.5 and components with asthma morbidity

In models using all available pre-and post-move visit-level observations, one standard deviation higher annual total PM_2.5 (0.73 μg/m³, Table 2) was associated with 60% higher odds of a reported exacerbation within the past three months (OR=1.60, 95%CI [1.21–2.12], Figure 2), 20% higher odds of a day of reported asthma symptoms (OR: 1.20, 95% CI [1.14–1.27]), and a 0.35 point lower ACT score (coefficient: −0.35), although this association was not statistically significant [95% CI: −0.79– 0.10]. Among PM_2.5 components, only BC was statistically significantly associated with odds of an exacerbation, while BC, OC and SO₄ showed statistically significant increases in the odds of symptom days. Full model results for PM components are shown in Table S2. Sensitivity analyses controlling for stress in addition to move did not attenuate the association between total PM_2.5 and asthma exacerbations (OR 1.75, 95% CI [1.23–2.50]), but did attenuate associations for symptom days and ACT score: OR for symptom days was reduced from 1.20 to 1.04 (95% CI: [0.96–1.13]) and was no longer statistically significant, and coefficient for ACT score was reduced from −0.35 to −0.25 (95% CI [−.73 - .22]) (Table S2).

Figure 2. Association of annual PM_2.5 (grey background) and component concentrations with a reported exacerbation within the past 3 months (left), symptoms within the past two weeks (center), and ACT Score (right).

Odds ratios and coefficients correspond to the standard deviation (μg/m³) of concentrations across all addresses and four years and are adjusted for move, age, sex, and year: PM_2.5: 0.73; Black Carbon (BC): 0.07, Organic Carbon (OC): 0.22; Nitrate (NO₃): 0.09; Ammonium (NH₄): 0.06; Sulfate (SO₄): 0.06.

PM_2.5 as a mediator of changes in asthma morbidity with move

We found that changes in PM_2.5 mediated 77% (95% CI: 39%–100%) of the reduction in risk of exacerbations, 22% (16% - 36%) of the reduction in risk of symptom days, and 14% (10 – 23%) of improvements in ACT score associated with moving to better-resourced neighborhoods. All three PM components with positive associations with exacerbations and symptom days—BC, OC, and SO₄—also showed mediating effects for those outcomes (Table 3). Although associations between ACT Score and the BC and SO₄ components did not exclude the null effect, reduction in those concentrations were estimated to mediate some improvement in ACT score.

Table 3. Share of moving-related benefits on asthma morbidity and control [95% CI] mediated by total PM_2.5 and components.

Only components with a positive estimated association with asthma morbidity are included.

PM component	Exacerbations	Symptoms	ACT
Total PM2.5	77% [39% – 100%]	22% [16%–36%]	14% [10%–23%]
BC	46% [27% – 100%]	11% [8%–18%]	19% [14%–30%]
OC	31% [18% – 100%]	1% [1%–2%]	-
SO4	18% [10%–77%]	10% [7%–16%]	11% [8%–18%]

BC=black carbon; OC=organic carbon; NO₃=nitrate; SO₄=sulfate; NH₄=ammonium

Discussion

We found that an intervention that supported families with children with asthma to move to better-resourced neighborhoods provided the co-benefit of reducing exposure to outdoor concentrations of PM_2.5 for approximately 70% of the sample, with a mean reduction of 0.64 μg/m³ (9%), a magnitude that is associated with health benefits (7). Among components, OC and BC showed the greatest reductions in terms of both concentration and relative share. Furthermore, total PM_2.5 was associated with significant increases in risk of exacerbations and symptoms (60% and 20% per SD increase, respectively). The change in PM_2.5 exposure with moving was estimated to mediate a substantial share of the improvement in exacerbations and a smaller share of symptom improvements that were associated with moving. These findings support the notion that changes in air pollution exposure play an important role in the asthma morbidity benefits of a move to a better-resourced neighborhood, and provide further evidence that persistent neighborhood-level differences in outdoor air pollution are relevant for acute asthma outcomes (7).

It is notable that reducing PM_2.5 exposure was not an explicit goal of the housing mobility program and nor directly included in its designation of ‘opportunity’ neighborhoods, but moving was nonetheless effective in reducing pollution-attributable asthma morbidity. This highlights an environmental justice issue contributing to asthma disparities: lower-resourced neighborhoods, which often have disproportionate share of residents of color, also have a disproportionate share of urban pollution sources (13). The use of housing mobility as an intervention to reduce air pollution exposure for sensitive individuals and to address environmental injustice is not often considered within the paradigm of environmental regulations, which target emissions sources rather than exposed populations. Given the slow pace of emission reduction policies, near-term strategies like housing mobility assistance provide a valuable complement in accomplishing public health goals. From the housing policy perspective, housing mobility has been conceptualized as a tool for improving socioeconomic outcomes and, more recently, health outcomes, supporting clinician referral of patients to these programs. However, housing mobility has generally not been thought of as a tool to specifically counter inequities in air pollution exposure and reduce air pollution-related health disparities (21, 22). While we found that housing mobility was already successful in reducing air pollution exposure, it is possible that explicitly incorporating air pollution exposure into their definition of “opportunity neighborhoods” could amplify the documented health benefits of housing mobility programs.

Findings varied among components of PM_2.5, with BC most strongly associated with an increase in exacerbations and with OC and SO₄ showing positive but not statistically significant associations. The OR of 1.43 (95% CI 1.04–1.97) per standard deviation of BC (0.07 μg/m³), which is substantially smaller than the sd for PM_2.5 (0.73 μg/ m³), also suggests a greater increase in risk per unit mass. These three components were also associated with asthma symptoms. Because pollution sources vary in the components released, differences in the strengths of the associations may lend insight into which sources might be most harmful for asthma morbidity.

Limitations

This investigation builds on prior MAP work documenting improvements in psychosocial factors and housing conditions after moving.(4) We adjust for move status and include a sensitivity analysis controlling for stress to account for the fact that relocation altered multiple neighborhood-level factors that may be determinants of health. However, we cannot fully rule out confounding by changes in housing characteristics or indoor environment, which are associated with neighborhood pollution levels through the common factor of neighborhood poverty. Interpretation of mediation estimates requires the assumption of no unmeasured confounding between mediator and outcome. Our sensitivity analysis showed that the PM_2.5-exacerbation association remained robust after adjustment for stress, but the PM_2.5-symptoms association was attenuated and no longer statistically significant, reflecting uncertainty about the proportion of benefits mediated by air pollution. Whether this suggests that the mediating effect of PM_2.5 is confounded by stress or some other factor causally related to stress and PM_2.5 is difficult to parse because the mediators may act via shared pathways, so these findings should be interpreted cautiously. The potential interaction between asthma response to stress and air pollution adds further complexity, which we were not able to elucidate in this analysis but deserves continued examination.. Nonetheless, examining mediation provides meaningful insight into mechanisms, and the magnitude of PM_2.5 reduction observed with moving supports its likely contribution to improved asthma outcomes.

Although the causal relationship between air pollution exposure and asthma symptoms and exacerbations is well established, not all pollutant-outcome associations here were statistically significant, possibly reflecting limited statistical power given the modest sample size. Importantly, the absence of significance does not negate mediation by air pollution, as moving produced health-relevant reductions in PM_2.5 exposure. Mediation estimates should therefore be interpreted cautiously, with effects for total PM_2.5 on exacerbations being more robust than those for some components and symptoms and ACT score.

Other limitations, detailed in previous work (4), include that the analytic sample was modest (123 children) and drawn from a single housing mobility program in Baltimore, which may limit generalizability and did not provide power to perform stratified analyses or identify interaction effects. The counterfactual analysis required complete pre- and post-move address data, reducing the sample to 93 participants. This restriction may introduce selection bias if address completeness correlates with unmeasured factors.

We focused on PM_2.5 and its major components because these pollutants exhibit intra-urban spatial variation relevant to neighborhood-level disparities. Other pollutants (e.g., ozone, NO₂, SO₂) were not modeled; the causal relationship between PM_2.5 and asthma has been shown to exist independent of these pollutants (26), and neither ozone nor SO₂ is expected to show fine-scale patterning comparable to PM_2.5. Multi-pollutant models, particularly in mediation analysis, were beyond the scope of this study given sample size limitations and methodological complexity (27). Our use of annual average exposures reflects the study’s aim to capture “typical” neighborhood conditions as a structural determinant of health, rather than short-term changes in pollution response and related acute responses (28). Chronic exposure differences influence both long-term risk and frequency of acute manifestations, making annual metrics appropriate for this question (7). However, because we assigned annual average exposure to outcomes reported over shorter windows, some degree of nondifferential exposure misclassification is likely, which generally biases associations towards the null and may attenuate mediation estimates, but is not likely to introduce spurious associations in this case (29).

Implications

This work builds on previous findings of greater effects of PM_2.5 on asthma severity among children residing in historically redlined neighborhoods, potentially attributable in part to the prevalence of higher-toxicity sources (23). Although traffic-specific PM components have been previously investigated, this is the first study to examine the relationship between a range of components and acute asthma morbidity beyond acute care use (9, 24, 25). Interventions targeting BC, OC, and SO₄—diesel engines, marine fuels (e.g., for marine shipping, a major source in Baltimore) and other biomass combustion (e.g., waste incinerators, two of which are located in Baltimore City)—may yield larger reductions in asthma morbidity risk. Such interventions may include reducing emissions of these components, as well as considering neighborhoods with a high density of such sources as priority candidates for housing mobility assistance for those with asthma.