Association between prenatal ambient particulate matter and childhood asthma is modified by community safety and child sex

Abstract

Studies document independent effects of prenatal air pollution exposure and social environmental factors, including neighborhood safety, on childhood asthma development with documented sex-specific effects. Further research examining these factors jointly is needed. We examined associations between prenatal residence-level daily fine particulate matter (PM_2.5) exposure and child asthma, considering effect modification by a validated Neighborhood Sentiment and Safety Index (NSSI) and child sex. Participants were mothers and full-term (> 37 weeks gestation) singleton-born children from two Boston-area pregnancy cohorts. The Asthma Coalition on Community, Environment and Social Stress (ACCESS) project enrolled 955 pregnant women between August 2002 and July 2009.The Programming of Intergenerational Stress Mechanisms (PRISM) study recruited 390 pregnant women from March 2011 to December 2013. Bayesian distributed lag interaction models (BDLIMs) were implemented to estimate associations between child asthma incidence and daily average maternal PM_2.5 exposure across gestation. Effect modification by NSSI and child sex was examined using a BDLIM comparing models with and without effect modification. Women were primarily minorities (29% black, 47% Hispanic) reporting less than a high school education (54%). Children were followed 15.1 ± 3 years; 204 (17%) developed asthma. In the overall sample (n = 1,178), increased PM_2.5 exposure between 21 and 27 weeks gestation was associated with increased odds of asthma in children born to women in the high NSSI group (NSSI ≥ 75th percentile representing safer neighborhoods). Both boys and girls were at higher risk of asthma when considering joint effects. These data add to a growing literature highlighting the need to consider both chemical toxins and psychosocial factors operating in communities to better elucidate which factors are driving respiratory health effects. A singular focus on changes to mitigate air pollution may not have high impact on improving respiratory health, particularly in historically under-resourced areas where effects may be more highly driven by social determinants.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-19639-7.

Introduction

The increased burden of asthma in ethnic-minority children living in lower-socioeconomic status (SES) urban communities is linked independently to differential exposure to chemical toxins as well as psychosocial factors contributing to increased stress^1,2. Outdoor air pollution is one of the most ubiquitous environmental determinants of respiratory health, particularly in urban settings^1,3,4. Exigencies of inner-city living also contribute to psychosocial risk factors conferring increased asthma morbidity^5–7. Community violence constitutes a key social determinant of health with a reduced sense of safety being a primary aspect of neighborhood disadvantage linked with poorer health^5,8. In the United States (US), community crime and violence is a unique stressor existing in many high-risk urban neighborhoods. Our group and others have linked perceived crime levels and/or fear of crime with greater psychological distress^9–11. Increased air pollution exposure and consequent health effects are more highly prevalent in communities that also experience elevated social stressors like higher crime due to historical discriminatory policies driving disinvestment that contributes to both past and contemporary exposures (e.g., redlining)^12–16.

Asthma risk begins in utero during rapid lung morphogenesis when the developing fetus is particularly vulnerable to toxins due to immature immune, neuroendocrine and antioxidant defenses. Studies link prenatal air pollution^17–22 and perceived crime levels and/or fear of crime^6,23,24 with adverse respiratory outcomes including asthma and reduced lung function in children. Air pollutants and stressors including violence impact a range of pathways contributing to asthma^6,25. Exposure to psychological stress in vulnerable developmental periods can shift homeostatic functioning of interrelated stress response systems (e.g., immune, autonomic, neuroendocrine, oxidative stress) and associated regulatory processes in pregnant women that play a role in the development of respiratory disorders². Pollutant-induced shifts in similar physiological states and their interacting systems are also thought to underlie respiratory disease programming²⁵. Given that psychosocial stress and ambient air pollutants can increase asthma risk through related but not completely overlapping mechanisms, they may act both independently and synergistically in jointly exposed pregnant women to impact outcomes. We previously demonstrated independent effects of prenatal community violence and traffic-related air pollution exposures on wheezing in urban children, although sample size limited power to examine joint effects²⁶. In addition, sex-specific effects of both ambient pollution and social stressors on prenatal programming of respiratory outcomes have been documented^17,27,28. In order to identify those at greatest risk, research should consider complex interactions among all of these factors.

We examined joint effects of prenatal ambient fine particulate matter with an aerodynamic diameter of 2.5 μm (PM_2.5) and neighborhood safety in relation to childhood asthma incidence in an urban sample in the northeastern United States (US). We hypothesized that children born to women living in less safe neighborhoods and/or women exposed to higher PM_2.5 would more likely develop asthma. We further considered effect modification of these associations based on child sex. Given emerging evidence suggesting variability in whether the chemical pollutant, social pollutant, or combined exposure to both will have joint effects on child asthma may vary across geographic region being considered, we did not have a specific hypothesis as to which pattern would emerge in these analyses^29,30.

Materials and methods

Study population

Participants were mothers and full-term (≥ 37 weeks gestation) singleton-born children from two Boston-area pregnancy cohort studies led by our group designed for harmonization by using parallel enrollment procedures and standardized assessment measures.

The Asthma Coalition on Community, Environment and Social Stress (ACCESS) project enrolled 955 English- or Spanish-speaking pregnant women ≥ 18 years of age receiving prenatal care at Brigham and Women’s Hospital (BWH), Boston Medical Center (BMC), and the East Boston Community Health Center (EBCHC) at 24.3 ± 9.9 weeks’ gestation between August 2002 and July 2009 who gave birth to a live-born infant and continued follow-up.

The Programming of Intergenerational Stress Mechanisms (PRISM) study recruited 390 English- or Spanish-speaking women ≥ 18 years of age receiving prenatal care from the Beth Israel Deaconess Medical Center (BIDMC) and the EBCHC at 21.4 ± 9.1 weeks’ gestation from March 2011 to December 2013 who gave birth to a live-born infant and continued follow-up. Women reporting intake of ≥ 7 alcoholic drinks/wk before pregnancy or any alcohol after pregnancy recognition, and children with congenital abnormalities that could impact participation were excluded. For both studies, procedures were approved by the ethics committee at the Icahn School of Medicine at Mount Sinai Institutional Review Boards (IRBs) (approval #’s for PRISM 12–00875 A and ACCESS 12–00661). Mothers provided written consent in their primary language (English or Spanish) and children older than 7 years of age provided written assent. Analyses included mother–child dyads from the ACCESS (n = 830) and PRISM (n = 348) studies with data on PM_2.5 exposure starting in 2003 (first year available as detailed below), neighborhood conditions, and children’s asthma. Characteristics for those included in analyses were not statistically significantly different when compared to overall cohort participants based on race/ethnicity, education, and maternal age.

Prenatal air pollution

Predicted daily ambient PM_2.5 concentration during pregnancy was derived based on each participant’s residential address using a validated model assessing temporally- and spatially-resolved PM_2.5^31,32. This method utilizes the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite-derived Aerosol Optical Depth (AOD) from the Multi-Angle Implementation of Atmospheric Correction (MAIAC) algorithm. To predict daily residence-level PM_2.5 concentration, we applied a hybrid approach using remote sensing data with traditional land use regression (LUR) predictors and advanced machine learning methods to yield residence-specific estimates of PM_2.5. The model calibrates the relationship between AOD and ground-level monitor-measured PM_2.5 with a grid cell of 1 × 1 km. Estimates were generated using an Extreme Gradient Boosting (XGB) and Inverse Distance Weighting (IDW) Synthesis: XIS. We used satellite AOD and a parsimonious set of additional predictors to make predictions at participant’s residential addresses recorded throughout pregnancy, capturing near-roadway gradients and allowing the estimation of residence-level exposures. PM_2.5 estimates generated via this machine-learning model were available starting from 2003 to 2021. For children born 37–40 weeks gestation, we used postnatal PM_2.5 exposure estimates so analyses included a 40-week PM_2.5 exposure period for all subjects.

Neighborhood sentiment and safety index

We derived the validated Neighborhood Sentiment and Safety Index (NSSI)³³ using data from three administrative sources: (1) Environmental Systems Research Institute’s (ESRI) Crime Indices, (2) Data Axle’s Historical Business Data, and (3) the U.S. Census and American Community Survey (ACS), to characterize neighborhood context linked in prior research to neighborhood safety, sentiment, and fear of crime^33–36. ESRI Crime Indices includes nine standardized indicators: personal crime, murder, rape, robbery, assault, property crime, burglary, larceny, and motor vehicle theft. Data sources 2 and 3 were drawn from 2010 files. Census tract-level crime data from 2016 available for Massachusetts’ fourteen counties, excluding Berkshire and Franklin Counties, were used; 2020 ESRI Crime Indices were used to address limited 2016 data (Supplementary Fig. 1 depicts participants with imputed data). We fit a penalized generalized additive model using 2020 crime data and other variables that are inputs into the NSSI, predicting missing 2016 data following variable selection described by Marra et al.³⁷. The 10-fold cross-validation adjusted R² for all crime data prediction models ranged between 0.89 and 0.94.

All variables were standardized using z-scores to minimize differences in scale, and a step-wise Principal Component Analysis (PCA) was implemented to derive the census tract-level NSSI comprised of 33 variables characterizing neighborhood safety and correlated social conditions as previously detailed³³. The NSSI score ranged from − 1.60 to 0.52 with higher values indicating better neighborhood safety. The NSSI was dichotomized for analyses (< 75th percentile = low, ≥ 75th percentile = high).

Childhood asthma outcome

As detailed in prior publications^4,38mothers were asked about their child’s respiratory health at follow-up visits through telephone and face-to-face interviews at approximately 4-month intervals for the first 30 months, then annually afterwards. Children’s asthma status was determined based on maternal report of any of the following: “Has your child ever had asthma [since birth]?”, “Has a doctor or other healthcare provider ever said that your baby had asthma”, “Does your baby currently take medicine for his/her asthma?” or “Has your baby been hospitalized overnight for the asthma [since birth]?”.

Distinguishing transient early wheezing from persistent asthma is a challenge in early childhood. To address misclassification, the primary asthma outcome was assessed at school age and diagnoses made solely before the age of 2 without subsequent evidence of asthma-related symptoms or medication use were excluded.

Covariates

Maternal age, race, education, smoking and pre-pregnancy height and weight were ascertained by questionnaire administered in pregnancy; child’s sex and date of birth were obtained by medical record review. Maternal atopy was defined by self-reported doctor-diagnosed asthma, eczema, and/or hay fever. Prenatal daily temperature was derived using MODIS satellite surface temperature measurements³⁹. Multiple imputation generating 10 iterations was implemented using the “mice” package in R^40,41 to account for missingness on maternal education (5.3%) and atopy (8.0%).

Statistical analysis

We estimated the association between daily average maternal prenatal PM_2.5 exposure and child asthma incidence using Bayesian distributed lag interaction models (BDLIM). BDLIM extends the traditional constrained distributed lag framework developed for time series and prenatal exposure studies to estimate critical windows during pregnancy and allows one to account for effect modification, in our case, by NSSI and child sex^42,43. The logistic BDLIM for child () who is in subgroup (e.g. for stratification by sex is for female and for male) is

where is a fixed subgroup-specific intercept, is the regression coefficient characterizing the subgroup-specific association between weighted pollution exposure and children’s asthma status, is the weighted exposure, and is the covariate regression term. The s identity subgroups-specific critical windows of susceptibility while identifies the subgroups-specific within-window effect. When weights are constant, this is equivalent to using pregnancy average exposure. However, when the weights vary over time (t) the model assigns greater relative weight to some periods. These time periods with greater weight are identified as the critical windows. We ran the model separately evaluating effect modification by sex (female/male), by NSSI (Low-Mid/High NSSI), and by a four-level group variable defined by cross-classifying sex and NSSI.

The approach quantifies the likelihood of each pattern of effect modification in the relationship between prenatal PM_2.5 exposure and children’s asthma differentiated by a particular effect modifier. These patterns range from having different critical windows and associations between exposure and asthma status, to sharing the same windows and associations across subgroups that is best supported by the data. Hence, we are able to identify if the groups have the same or different critical windows and identify if the groups have the same or different within window effects. We also conducted sensitivity analyses by adding maternal smoking during pregnancy to the list of confounders () in the model. All Bayesian models are were run for 100,000 Markov chain Monte Carlo (MCMC) iterations with 50% burn-in and thinning the chain by retaining every fifth sample⁴⁴. Models were adjusted for maternal age, education, atopy, and prenatal average temperature (as well as child sex when not considered as an effect modifier). Race/ethnicity was not included in models as it was not associated with child asthma in univariate analyses (p-value > 0.20). Moreover, as race/ethnicity is related to our exposures of interest (Supplemental Figs. 2 and 3), we conceptualized race as a social construct reflective of living in historically segregated communities shaped by discriminatory practices (redlining and disinvestment) contributing to differential environmental conditions by race/ethnicity⁴⁵. To address the temporal exclusivity of the two cohorts and the downward trend in pollution levels, we adjusted for cohort in the main model. Additionally, to confirm the effect of birth year, we conducted a sensitivity analysis using a linear term for birth year binned into 4-year intervals since 2003. The results are shown in Supplementary Figs. 5 and 6. Analyses were conducted using the “regimes” package in R (v4.0.2, Vienna, Austria)⁴². The methods applied in this study were in accordance with the approved regulatory guidelines.

Results

PM_2.5 levels averaged over pregnancy and NSSI scores at participant addresses are mapped in Fig. 1; PM_2.5 levels were only moderately correlated with NSSI scores (Spearman correlation = 0.2). Table 1 summarizes characteristics in the sample overall and by child sex. Women were primarily minorities (28.7% black, 47.1% Hispanic) with 53.7% reporting less than a high school education. Children were followed 15.1 ± 3 years; 204 (17.3%) developed asthma, with a higher proportion among boys vs. girls (19.7% vs. 14.8%).

Fig. 1.

Visualization of PRISM and ACCESS participant locations. Panel (A) shows participant locations with corresponding average PM_2.5 levels over pregnancy (ug/m³). Panel (B) displays locations coded by NSSI values ranging from − 1.6 to 0.52, indicating neighborhood sentiment and safety. *At this mapping scale, no participants can be identified by mapped location. **Figure 1 is an original map generated by the authorship team in ArcGIS Pro 2.7.2. https://www.esri.com/en-us/arcgis/products/arcgis-pro/overview.

Table 1.

Summary of sample characteristics.

	All		Girls		Boys
	N	(%)	N	(%)	N	(%)
Participants	1178		575	48.8	603	51.2
Maternal education
<=12years	632	53.7	325	56.5	307	50.9
> 12years	546	46.3	250	43.5	296	49.1
Maternal atopy*
Yes	722	61.3	362	63	243	40.3
No	456	38.7	213	37	360	59.7
Cohort
ACCESS	830	70.5	407	71	423	70.1
PRISM	348	29.5	168	29	180	29.9
Maternal race/ethnicity
White/Non-Hispanic White	201	17.7	96	17.2	105	18.2
Black/Hispanic Black	326	28.7	152	27.2	174	30.1
Non-Black Hispanic	535	47.1	274	49.1	261	45.1
Others	74	6.5	36	6.5	38	6.6
Maternal Obesity**
Yes	324	30.4	162	31.6	162	29.3
No	741	69.6	351	68.4	390	70.7
Maternal smoking***
Yes	362	30.7	181	32.3	181	31
No	782	66.4	380	67.7	402	69
Child Asthma
Yes	204	17.3	85	14.8	119	19.7
No	974	82.7	490	85.2	484	80.3
NSSI
High	306	26	134	23.3	172	28.5
Low	872	74	441	76.7	431	71.5
	Mean	SD
Maternal age at enrollment	28.13	6.04	28.11	6.08	28.16	6.01
Child age	15.09	2.96	15.06	2.85	15.14	3.07
NSSI****	− 0.17	0.27	− 0.17	0.27	− 0.16	0.28
Average pregnancy temp	11.23	2.5	11.45	2.5	11	2.5
Average pregnancy PM2.5	9.9	1.74	9.95	1.71	9.81	1.76

*Maternal atopy defined as mother ever reporting diagnosis of asthma, hay fever, or eczema.

**Maternal obesity defined by pre-pregnancy BMI above 30.

***Maternal smoking defined as mom reporting ever smoking cigarettes.

****Neighborhood sentiment and safety index for Massachusetts - derived via PCA and includes components race, crime, education, income, population size, community and legal services.

Figure 2 (panel A) shows time-weighted cumulative associations between prenatal PM_2.5 exposures and child asthma modified by low and high NSSI. Significant associations between prenatal PM_2.5 exposure and increased odds of developing asthma was only evident in the higher NSSI group. The estimated cumulative odds ratio (OR) of asthma was 2.25 (1.003–5.27) per 1 unit increase in PM_2.5 for those with NSSI ≥ 75th percentile, and 1.09 (0.83–1.61) for the group living in communities with a NSSI < 75th percentile. The derived posterior probability was 0.63 supporting that this was the best-fitting pattern of effect modification. Figure 2 (panel B) shows the modifying effects of NSSI on the time-varying association between PM_2.5 and child asthma. The BDLIM identified a significant association between increased PM_2.5 exposure later in pregnancy (21–27 weeks gestation) and increased odds of asthma in children born to mothers who also resided in a community with a higher NSSI.

Fig. 2.

Effect of PM_2.5 on asthma outcomes stratified by low and high NSSI groups. Panel A: Time-weighted cumulative associations between prenatal PM_2.5 exposures and child asthma modified by high NSSI (above the 75th percentile). Models were adjusted for maternal age, education, atopy, child sex, and prenatal average temperature. Panel B: The time-varying association modified by NSSI between PM_2.5 and child asthma risk. The x-axis shows the gestational age in weeks. The y-axis shows the OR of child asthma per 1 ug/m3 increase of prenatal PM_2.5 exposure. The solid line represents the predicted OR, horizontal thin line showing OR = 1, and the gray area indicates the 95% confidence interval. A significant sensitive exposure window is identified for the weeks where the estimated pointwise 95% CI (shaded area) does not include one.

Additional effect modification by child sex was examined using a BDLIM comparing the model statistics across different modeling constraints (Fig. 3). Based on the posterior model probability (Model probability = 0.99) and deviance information criterion, the model indicated NSSI and sex differences in both the sensitive windows and the magnitude of the effect estimates for the relationship between PM_2.5 and childhood asthma. Again, associations between gestational exposure to PM_2.5 and increased asthma incidence was only evident in children born to women living in neighborhoods with a higher NSSI. While associations were somewhat stronger among boys, the relationship between increased PM_2.5 exposure and childhood asthma risk was evident among both boys and girls born to women living in communities with NSSI scores at ≥ 75th percentile. For boys born to mothers living in communities with a high NSSI, the estimated cumulative odds ratio (OR) of PM_2.5 exposure across pregnancy was 2.26 (0.92–6.80) per 1 unit increase, and 1.61 (0.64–4.64) for girls born to mothers living in neighborhoods having a high NSSI. Models suggested a different critical window of vulnerability to PM_2.5 during gestation for boys compared to girls showing a peak association in boys around 30 weeks gestation, while the peak for girls was around 21 weeks. In sensitivity analyses further adjusting for maternal smoking during pregnancy, these findings were not substantively changed (Supplementary Fig. 4). As mentioned, to confirm the effect of birth year, we conducted a sensitivity analysis using a linear term for birth year binned into 4-year intervals since 2003. The results are shown in Supplementary Figs. 5 and 6, these findings were not substantively changed.

Fig. 3.

Efect of PM_2.5 on asthma outcomes stratified by low and high NSSI groups as well as child sex. Panel A: Cumulative effect of PM2.5 on asthma outcomes by low and high NSSI and child sex. Models were adjusted for maternal age, education, atopy, and prenatal average temperature (as well as child sex when not considered as an effect modifier). Panel B: Time-varying association between PM2.5 and child asthma by NSSI level and child sex.  The x-axis shows the gestational age in weeks. The y-axis shows the odds ratio (OR) of child asthma per 1 ug/m³ increase of prenatal PM_2.5 exposure. The solid line depicts the predicted OR, with a slender horizontal line indicating the no-effect threshold (OR = 1), and the shaded region representing the 95% confidence interval. A significant sensitive exposure window is identified for the weeks where the estimated pointwise 95% CI (shaded area) does not include one.

Discussion

These findings underscore the need to consider social conditions of communities in which we live as well as chemical toxins encountered in our environment. Considering psychosocial factors and physical pollutants together can help identify joint effects as well as identify which may be particularly driving effects. Herein we considered prenatal exposure to ambient outdoor PM2.5 and an index of neighborhood safety in relation to child asthma risk. Exposure to crime and resident sense of safety was conceptualized as a chronic community-level stressor. Notably, these data corroborate prior research documenting increased risk of childhood asthma related to gestational exposure to fine particulate matter. Also consistent with previous findings, the especially vulnerable exposure window occurred in the second half of pregnancy^20,46. However, analyses did not support synergistic effects of less optimal neighborhood exposures (i.e., higher ambient PM_2.5 and living in a less safe neighborhood). Rather, effects of PM_2.5 were only evident among children born to mothers living in neighborhoods with a higher safety index during pregnancy. There was no association between increased PM2.5 exposure over pregnancy with asthma risk in children born to mothers living in less safe neighborhoods. This pattern suggests that neighborhood disadvantage which has been independently associated with poorer respiratory health outcomes, with exposure to crime and resident sense of safety being a key social determinant of this relationship, may be driving asthma risk rather than ambient PM2.5 in these communities. While associations were somewhat stronger among boys, both sexes were at increased risk of asthma when accounting for both PM_2.5 and community safety-related conditions.

To date, few studies have assessed interactions between stress and ambient air pollutants and respiratory health, particularly in the prenatal period, with findings on interactions being variable. Studies have previously found that children exposed to both high postnatal nitrogen dioxide and high lifetime violence or stress exposure were at greatest risk for asthma^30,47. Another study of asthmatic children aged 9–18 years found that the association between higher family stress and asthma exacerbations was stronger when NO₂ levels were more modest²⁹. Specifically, they found that higher chronic stress was associated with increases over time in asthma symptoms and decreases over time in lung function (i.e., peak expiratory flow rates) as pollution levels decreased. Similar to Chen and colleagues²⁹our findings suggest a threshold effect for one exposure at differing levels of the other. In this study we observed that the association between higher prenatal PM_2.5 exposure and asthma was stronger in children born to mothers with a higher NSSI reflecting a safer and more accessible environment. One possible explanation is that the effect of a more disadvantaged neighborhood context on child asthma might be of significant magnitude to lead to a ‘saturation effect’ such that the additional risk conferred by prenatal PM_2.5 exposure beyond these effects could not be detected.

The handful of studies examining sex-specific effects of prenatal exposure to air pollution and children’s asthma risk demonstrate increased risk among boys^{4,18,19,48,49}. Relatedly, studies examining effects of air pollution exposures in early childhood on later wheeze and asthma risk have also largely demonstrated increased risk among boys⁵⁰ although some report effects among girls^51,52. In these analyses we found that both boys and girls born to women living in more desirable neighborhoods based on safety were at increased risk for asthma when exposed to PM_2.5 during pregnancy.

Studies examining sex-specific effects of prenatal stress and asthma risk are also mixed. We previously demonstrated associations between increased maternal-reported frequency of negative life events during pregnancy in relation to asthma onset in children⁵³ finding that prenatal stress was only associated with increased odds of asthma in boys. This is further supported by data from the Swedish Medical Birth Register including ~ 500,000 mother-child dyads finding that boys born to mothers who had experienced a loss of someone close (i.e., bereavement stress) during the second trimester were particularly at risk for asthma⁵⁴. Two studies provide notable contrasts to these findings. Brunwasser et al.⁵⁵ examined associations between increased maternal-reported negative life events during pregnancy and nasal inflammatory markers from nasal washings obtained during acute respiratory infections which are a major cause of early life wheezing and later asthma risk. They found that boys born to mothers reporting greater number of negative life events during pregnancy had reduced nasal inflammation, lower infection severity, and lower likelihood of hospitalization for respiratory infection over the subsequent 2 years, while no associations were observed in girls. In another study, researchers found that maternal-report of ice storm-related stress during pregnancy was associated with significantly increased rates of wheezing, asthma, and inhaled corticosteroid use in girls but not boys⁵⁶. In the data presented here, both boys and girls were at heightened risk when joint exposure to air pollution and neighborhood safety was considered, likely reflecting complex impacts on underlying sex-specific mechanisms when co-exposure is examined. While further studies are needed, there is sufficient evidence of sex-dependent associations for both air pollutants and psychosocial stress to recommend that researchers consider the potential role of child sex in asthma studies and at a minimum report study outcomes stratified by child biological sex so we can continue to disentangle these relationships.

We contend researchers need to broaden our environmental lens and jointly consider the physical and social factors driving respiratory health in studies so that we can better elucidate which factors are driving effects⁵⁷. This will allow the field to more effectively tailor interventions. This is further exemplified in a recently published randomized control trial designed to examine whether moving from high-poverty neighborhoods to more advantaged communities was associated with reduced morbidity among children with asthma and to explore potential mediating factors⁵⁸. This study enrolled 123 children aged 5–17 years with persistent asthma whose families participated in the Baltimore Regional Housing Partnership housing mobility program assessing asthma exacerbations and symptoms, measured exposure to indoor air pollution and allergens, and stress, including social cohesion, neighborhood safety, and urban stress, as well as psychological functioning in children and their caregivers before and after the move. Both maximum symptom days in the past 2 weeks significantly improved after moving with an adjusted difference of -2.37 days (95% CI, -3.14 to -1.59; P < .001) and a reduction in asthma exacerbations was also significant post move. While there was a significant decrease in exposure to indoor pollution and allergens post move, these changes did not mediate improved outcomes. Rather, measures of stress, including neighborhood safety, were estimated to mediate between 29% and 35% of the association between moving and reduced asthma exacerbations. This study suggests that programs to counter housing discrimination, largely stemming from historical policies such as redlining, can reduce childhood asthma morbidity through enhancement of the social environment.

The current study has notable strengths including the large (n = 1,178) and diverse urban sample with reasonable heterogeneity of air pollution and neighborhood conditions to be able to examine joint exposures as well as sex specific effects. Use of advanced statistical modeling enhanced our ability to assess the impact of timing of PM_2.5 exposure effects as well as increased power to detect higher order interactions. We also note some limitations. Child asthma was based on caregiver report which may result in misclassification. However we used well-established standardized questionnaires to obtain information including diagnosis by a healthcare provider which is a widely used approach in larger scale epidemiological studies. Moreover, we excluded diagnoses made solely before the age of 2 without subsequent evidence of asthma-related symptoms or medication use to avoid inclusion of early wheezing respiratory illnesses as done in prior work to further minimize the likelihood of misclassification³⁸. We did not consider indoor air quality. However, research suggests that variations in indoor source particles are largely uncorrelated with variations in outdoor source particles⁵⁹ and therefore unlikely to confound observed associations between ambient levels and health. Moreover, adjustment in a sensitivity analysis for prenatal smoking, a major contributor to indoor air quality, did not substantively change our findings (see Supplementary Fig. 4). There may be other environmental or social influences that covary with PM_2.5 that were not considered in this study.

These data add to a growing literature highlighting the need to consider both physical and psychosocial factors operating in our communities to impact respiratory health. Taken together, growing evidence suggests that a singular focus on changes to mitigate ambient air pollution exposure may not have high impact on improving respiratory health, particularly in historically under-resourced areas where effects may be more highly driven by social determinants.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

LH is responsible for analyzing data, interpreting results and creating all tables. MM oversaw spatial modeling and created all mapping figures. LH and MM are responsible for initial manuscript writing and preparation. MM, RJW, RF and IK were responsible for creating the NSSI. RJW led funding supporting data acquisition, hypothesis generation, and BC provided additional supervision of analyses and interpretation of results. All authors edited manuscript drafts and reviewed the final submitted manuscript.

Funding

The Asthma Coalition on Community, Environment, and Social Stress (ACCESS) project has been funded by the National Institutes of Health (NIH) grants R01ES010932, U01HL072494, R01MD006086, R01HL080674, R01HL097197, and UG3OD023337. The Programming of Intergenerational Stress Mechanisms (PRISM) has been funded by NIH grants R01HL09560, R01HL114396, and UG3/UH3OD023337. Biostatistics support was provided through R01ES028811, UL1TR004419, P30ES023515, and P30ES000002.

Data availability

The data analyzed in this paper were collected as part of the PRogramming of Intergenerational Stress Mechanisms (PRISM) pregnancy cohort study at the Icahn School of Medicine at Mount Sinai and the Asthma Coalition on Community, Environment and Social Stress (ACCESS) cohort study including participants from Brigham & Women’s Hospital (BWH) and Boston Medical Center (BMC). Due to human subjects confidentiality, data are not public available; however, a limited dataset may be obtained from the corresponding author upon reasonable request. All data included in the derivation of the NSSI exists within data repositories that are accessible without cost or with a subscription or purchase agreement.

Declarations

Competing interests

The authors declare no competing interests.