Early-Life Exposure to Air Pollution and Childhood Asthma Cumulative Incidence in the ECHO CREW Consortium

Key Points Question Is there an association between early life exposure to air pollution and the risk of asthma by early and middle childhood, and is this association modified by individual and community-level characteristics? Findings In this cohort study of 5279 children, mean fine particulate matter (PM2.5) and mean nitrogen dioxide (NO2) air pollution during the first 3 years of life were associated both with asthma incidence by early and by middle childhood, after adjusting for individual-level characteristics. The association of ambient pollution (PM2.5 or NO2) with incident asthma was modified by community-level and individual-level socioeconomic circumstances, including maternal education and race. Meaning These findings suggest that exposure to PM2.5 or NO2 air pollution during early childhood may play a role in the development of childhood asthma, with higher risk among minoritized families living in densely populated communities characterized by fewer opportunities and resources and multiple environmental coexposures.


Introduction
Air pollution is a near ubiquitous exposure and the largest environmental contributor to disease and premature death in the world, including for children. 1,2Exposure to air pollution has been consistently associated with respiratory morbidity, including wheeze and exacerbation of asthma in children.Several reviews [3][4][5][6] on air pollution, asthma, and respiratory symptoms in children concluded that outdoor traffic pollution contributes to the development of childhood asthma.
However, community-level and individual-level contextual factors that increase not only exposure, but also susceptibility to air pollution-related childhood asthma effects, including the developmental stage and age of onset of asthma, remain poorly understood.Most prior cohort studies with individual-level information lacked the geographic, racial and ethnic, and socioeconomic diversity to explore the modifying role of community-level contextual factors and the association between air pollution exposure and asthma development.
The Environmental Influences on Child Health Outcomes (ECHO) Children's Respiratory and Environmental Workgroup (CREW) 7 consortium, a US nationwide birth cohorts network, is wellpositioned to address these questions given the multiple decades of recruitment and follow-up of birth cohorts, the geographic and demographic diversity of study participants, 8 the wide distribution of state-of-the-art spatio-temporally-resolved fine particulate matter (PM 2.5 ) and nitrogen dioxide (NO 2 ) estimates, and the diverse individual and area-level socioeconomic and built environment factors that vary across and within cohorts.
Previously, we found in the CREW consortium that Black and Hispanic children and children who resided in census tracts with higher rates of household poverty and population density were at increased risk for developing childhood asthma. 9Building upon these findings, we hypothesized that (1) early life exposure to PM 2.5 and NO 2 is associated with increased asthma risk by early and middle childhood, and (2) not only individual-level, but also community-level factors (eg, living in areas with higher level of poverty, and less opportunity) are associated with increased child susceptibility to pollution effects on asthma risk.

Study Population
Our study population included 8 of the 12 longitudinal birth cohorts participating in the Children's Respiratory and Environmental Workgroup (CREW), representing a diverse sample of children and their families residing throughout the US in urban, suburban, and rural environments (eMethods 1 and eTable 1 in Supplement 1).We excluded cohorts with young children without harmonized asthma outcomes, and we excluded our oldest cohort for which early-life air pollution estimates were not available.
Eligibility criteria, study recruitment, and other methods, have been previously described. 7rth years of cohort participants spanned from 1987 to 2007, and participants were followed up to age 11.All cohorts had institutional review board approvals from each participating cohort's

Health Outcomes
We defined asthma as caregiver report of physician-diagnosed asthma.We ascertained the outcomes through survival analysis as asthma incidence by less than 5 years of age (early childhood) and by less than 12 years of age (middle childhood); and, secondarily, through logistic regression as ever asthma through age 4 years and through age 11 years.As a sensitivity analysis, we also considered ever asthma before age 4 years with any wheeze reported after age 3 years, to ascertain whether associations were consistent for those children diagnosed before age 4 years who had an indication of persistent (rather than transient) asthma.We obtained daily estimates of PM 2.5 and NO 2 from 2000 to 2016 at a 1 km 2 resolution from previously validated prediction models [10][11][12] and monthly PM 2.5 predictions at the 6 km 2 grid for the years 1988-2007 from a previously published model 12 to estimate PM 2.5 and NO 2 exposures (eMethods 2 in Supplement 1).

Individual Characteristics
The 1 km 2 estimates were linked to the home addresses for each CREW participant using our Decentralized Geomarker Assessment for Multi-site Studies (DeGAUSS) software 13,14 approach (eMethods 2 in Supplement 1).For the cohorts with person-time before 2000 when the 1 km 2 estimates were not available, we created cohort-specific calibration factors based on monthly mean values, using all available overlap data to obtain calibrated annual, monthly, and prenatal 6 km 2 exposure estimates for periods before 2000 (eMethods 2 in Supplement 1).
Annual mean values for each year of life beginning at the birth date through age 1 year and up to age 5 years were calculated.In addition, we calculated mean values from birth through age 1 year, 1 to 2 years, 1 to 3 years, and 1 to 4 years.As NO 2 estimates were only available from 2000 and later, we were not able to assign early life NO 2 exposures to 2 cohorts (Children's Asthma Study, Epidemiology of Home Allergens and Asthma Study), whose children enrolled at birth in the 1990s.

Neighborhood-Level Characteristics
We obtained US census data for percentage population with low income, percentage Black population, population density, median household income, and percentage low-income families.We also obtained the Child Opportunity Index (COI), 15,16

JAMA Network Open | Environmental Health
Early-Life Exposure to Air Pollution and Childhood Asthma Cumulative Incidence

Statistical Analysis Primary Analysis
We analyzed the association between exposures and asthma incidence with a Cox proportional hazard model, adjusting for potential confounders including mother's education, child's race and ethnicity, sex, smoking during pregnancy, parental history of asthma and an indicator variable for cohort.In secondary analysis, we examined the association between exposure and ever asthma using logistic regression models, adjusting for the same confounders as in the survival analysis.
We examined effect modification by child's race and ethnicity and sex, mother's education, and neighborhood socioeconomic factors, including an interaction term between each pollutant and each modifier in separate models.We then computed the association of pollution in each category of the modifier.For continuous modifiers (neighborhood socioeconomic factors), we computed the association of pollution at the 10th (Low) and 90th (High) percentiles of the modifier.We reported the results as hazard ratios (HRs) for the survival analysis and odds ratios (ORs) for the logistic regression with 95% CI for a 1 IQR increase in each pollutant.P < .05 was considered as the significance level.Statistical analysis was performed using R version 4.

Sensitivity Analyses
We repeated the analysis using asthma incidence by less than 5 years of age with any wheeze reported after age 3 years.To account for potential residual spatial correlation, we repeated the analysis using mixed-effect models adding random intercepts for census tract, in addition to the indicator variable for cohort.We then applied a multinomial regression, where the outcome was a categorical variable defined as asthma by age 4 years and first asthma diagnosis between ages 5 and 11 years.

Results
Among  Table 2 and eFigure 5 in Supplement 1 present the results of the association between early life air pollution exposure and asthma incidence by younger than 5 and younger than 12 years of age using survival analysis and logistic regression.We found stronger associations with both outcomes and pollutants averaged over the first 2 and 3 years of life compared with other pollution averages (eFigure 5 in Supplement 1).Specifically, we found that 1 IQR increase in mean NO 2 (   When we examined effect modification of air pollution averaged over the first 3 years of life by neighborhood factors (Table 3; eFigures 6 and 7 in Supplement 1), we found that for an IQR increase in PM 2.5 , children who resided in areas with lower education and health and environment opportunity had higher asthma incidence in the first 4 and 11 years of life.For an IQR increase in NO 2 , children who resided in areas with higher proportion of Black population and more urban areas had higher asthma incidence through the first 4 and 11 years of life.(Table 3) In sensitivity analyses, the results were similar when we restricted analyses to those with documented wheeze symptoms during middle childhood following an asthma diagnosis by age 4 years (eFigure 8 in Supplement 1).In mixed-effect models with a random intercept for census tract in addition to cohort, the results did not change (eFigure 9 in Supplement 1).The multinomial regression results (eFigure 10 in Supplement 1) showed similar associations between PM 2.5 and NO 2 and asthma by age 4 years.The associations were weaker with asthma between ages 5 and 11, mostly for PM 2.5 .

Discussion
In this multicohort study, we found that exposure to PM 2.5 and NO 2 during the first 3 years of life were associated with increased asthma incidence by early (<5 years) and middle (<12 years) childhood.Individual-level characteristics, including Black race and lower maternal educational attainment, and community-level factors, including lower health and environment child opportunity indices, population density, and neighborhoods with higher proportion of Black population, were associated with increased magnitude of the association between air pollution exposure and risk of childhood asthma.
Early childhood is a period of heightened concern, as higher exposures may lead to altered trajectories of airway and immune system development, with decreased lung function and asthma pathogenesis. 18,19The Tucson Children's Respiratory Study found, with suggestion of a larger response in Black children, that higher childhood NO 2 was associated with lower CC16, a biomarker in which its decrease has been associated with oxidative stress and reduced lung function.PM 2.5 and NO 2 may influence asthma development not only through oxidative stress leading to inflammation (eg, IL-6), but also through altered immune development, increased IgE-mediated allergic sensitization, and Th17-associated responses. 20,21 examining multiple periods of exposure and health outcomes, 5 this analysis adds to the growing epidemiologic evidence that early-life air pollution exposures are associated with the onset of childhood asthma.In the Southern California Children's Health Study (CHS) 22 traffic-related pollution exposures during childhood, including NO 2 at school 23 and homes, 22 were associated with increased asthma incidence.A follow-up study using 3 waves of the CHS Study 24 found that decreases in NO 2 and PM 2.5 were significantly associated with lower asthma incidence.In Boston, 25 first year of life and lifetime exposure to PM 2.5 were associated with increased risk of pediatric asthma during early childhood (3-5 years of age).A previous analysis in the Cincinnati Childhood   26 found that exposure during the first 2 years of life, but not exposure during later childhood, was associated with asthma development by age 7 years.
Another birth cohort study 27 found increased risk of having an asthma diagnosis at age 13 years with NO 2 postnatal exposure in the first year of life.Our study is consistent with these prior studies and suggests that the first 1 to 3 years of life are the most susceptible period for air pollution exposure to promote asthma development.
Our previous analysis 9 of asthma incidence in the CREW consortium found that adverse neighborhood characteristics and Black race were associated with increased childhood asthma incidence.We interpreted race to signify measured and unmeasured exposures from racism.Here, we leveraged the geographic, racial, and socioeconomic diversity of participant families and their communities to examine factors modifying associations between air pollution and asthma.
We found that associations of air pollution exposure with asthma were elevated among Black children and children born to mothers without a high school diploma.These results are consistent with prior studies showing that socioeconomic position (SEP) and race are key drivers of elevated environmental exposures, including air pollution, and race and SEP, each of which can be independently or synergistically associated with elevated physiologic stress leading to inflammation that increases susceptibility to disease, including asthma. 28,29In addition, Black children are more likely to be exposed to adverse childhood experiences, poor housing quality and indoor environments, and have less access to healthy food and greenspace. 30These factors, likely in combination, may inequitably affect Black children such that the negative health effects due to exposure to air pollution are heightened.
Similar to individual-level race and SEP, we also found that neighborhood measures, including the percentage of Black population and lower health and environment COI, also were significant modifiers.The COI domain of health and environment represents the quality of neighborhoods in terms of housing, access to food, built environment, and exposure to pollution and heat. 15Previous studies have shown disparities in asthma prevalence and morbidity across communities related to violence, poor housing, elevated environmental exposures, lack of access to health care. 30The COI has been previously associated with population-level asthma morbidity in the US. 31 A Swedish study using administrative data 32 found that PM 2.5 during the first 3 years of life increased asthma risk and that pollution-asthma associations were stronger in areas with lower education.
Our study has several strengths.To our knowledge, this is the first multicohort study of air pollution and asthma in the US that focused on the independent and interacting environmental and social influences on the age and child life-stage of asthma onset, demonstrating heightening of earlylife pollution effects by adverse community-level exposures.We report findings that have a consistent pattern and are robust to different methods of statistically testing associations. 33e added value of this study includes cohorts that were specifically focused on measuring outcomes related to childhood asthma, with multiple decades of recruitment, varying in terms of the population selection criteria (general and high risk), and with socioeconomic, racial and ethnic, geographic and temporal diversity, and with a wide distribution of state-of-the-art modeling for spatio-temporally-resolved estimation of pollutants.
In Europe, a multicity study 34 did not find significant association between air pollution exposure and childhood asthma.These cohorts were recruited in the late 1990s, and air pollution exposure was determined with land use regression models.In CREW, collecting and harmonizing data from each cohort enabled us to examine early and middle childhood onset of asthma as well as examine whether the associations differed by both individual-level and area-level factors in urban US settings.
Given the longitudinal nature of the study, we were able to examine incident cases instead of a prevalence.The use of the DeGAUSS approach, 13,14 a software containerization platform, enabled individual sites to geocode their birth addresses and then assign tract-level and exposure variables to those geocodes, so that all individual study sites had identically constructed data sets.Examining modification of air pollution on asthma by COI and SVI (indexes used in previous studies 35,36 on body mass index and cardiometabolic risk) enabled us to characterize how composite, as well as individual metrics of community resources and risk were associated with worse pollution effects on childhood asthma.

Limitations
This study had limitations.While the census variables are available every 10 years and have been merged to the nearest year of birth, SVI was available for the years 2000 and 2010 and COI only for 2010; therefore, some misclassification is possible.In these analyses, we examined NO 2 and PM 2.5 , although other pollutants common in urban environments like ozone and other traffic-related air pollution may also contribute to asthma and interact with NO 2 and PM 2.5 .Similarly, we do not have information about the contribution of pollution from indoor environments.While our analysis focused on early life exposures, future studies could test whether exposures in later life add cumulatively to lifetime risk of asthma. 25

Conclusions
In this cohort study of children from a highly diverse US population, PM 2.5 and NO 2 averaged over the first 3 years of life were associated with increased asthma incidence by early and middle childhood.
The associations of PM 2.5 and NO 2 were greater among families living in urban US communities with fewer opportunities and resources with multiple environmental coexposures.Air pollution continues to be a global burden with serious consequences on childhood health.Reducing asthma risk in the US requires regulation and reduction of air pollution combined with creation of greater environmental, educational, and health equity at a community level.

4 . 1 (
R Project for Statistical Computing) and SAS version 9.4 (SAS Institute) from February 2022 to December 2023.
eFigure 3 in Supplement 1 shows the flowchart of the analytic data set.eTables 2, 3, 4, and 5 in Supplement 1 present the characteristics when asthma, PM 2.5 , and NO 2 are missing.The distribution of the outcomes and the individual characteristics varied widely across cohorts, demonstrating racial and socioeconomic diversity among participants.The Figure shows the distributions of PM 2.5 and NO 2 as well as the neighborhood-level variables; all of the variables present substantial variability across and within cohorts.eFigure 4 in Supplement 1 shows the correlation among the neighborhood-level variables and the pollutants.PM 2.5 was not correlated with any variables, whereas NO 2 correlated with population density (r = 0.63).The census variables were positively correlated with SVI (eg, correlation between total SVI and percentage of Black population: r = 0.63) and negatively correlated with COI (eg, correlation between total COI and percentage of Black population: r = −0.64).

Table 1 .
Child and Caregiver Demographic Characteristics and Distribution of Respiratory Health Outcomes Among CREW Participants Asthma Study; CCAAPS, Cincinnati Childhood Allergy and Air Pollution Study; CCCEH, Columbia Center for Children's Environmental Health Cohort; COAST, Childhood Origins of Asthma Study; CREW, Children's Respiratory and Environmental Workgroup; EHAAS, Epidemiology of Home Allergens and Asthma Study; IIS, Infant Immune Study; URECA, Urban Environment and Childhood Asthma Study; WHEALS, Wayne County Health Environment Allergy and Asthma Longitudinal Study.For the same 3-year average, a 1 IQR increase in mean PM 2.5(3.4μg/m 3 ) was associated with increased asthma incidence among children younger than 5 years (HR, 1.31 [95% CI, 1.04-1.66])and children younger than 12 years (HR, 1.23 [95% CI, 1.01-1.507]).NO 2 and PM 2.5 were also significantly associated with increased odds of asthma through age 4 years and through age 11 years.

Table 3
presents results of air pollution modified by individual characteristics and by a selection of neighborhood-level variables separately for asthma incidence through the first 4 and 11 years of life.We found that the associations were stronger when mothers had less than a high school diploma and among Black children; for example, there was a significantly higher association between PM 2.5 and asthma incidence by less than 5 years of age in Black children (HR, 1.60 [95% CI, 1.15-2.22])comparedwithWhite children (HR, 1.17 [95% CI, 0.90-1.52])(Table3).Figure.Distributions of the Exposures, US Census Variables of Percentage Black Population and Population Density, and Child Opportunity Index (COI) and Social Vulnerability Index (SVI) Across the Children's Respiratory and Environmental Workgroup (CREW) Cohorts The figures shows cohort-specific boxplots, where the box is used to represent the IQR, or the data between the first and third quartile, the line within the box represents the median, the whiskers extend from each quartile to the minimum and maximum values, and the points beyond the whiskers represents the outliers.Bal indicates Baltimore; Bos, Boston; CAS, Children's Asthma Study; CCAAPS, Cincinnati Childhood Allergy and Air Pollution Study; CCCEH, Columbia Center for Children's Environmental Health Cohort; COAST, Childhood Origins of Asthma Study; COI, Child Opportunity Index; EHAAS, Epidemiology of Home Allergens and Asthma Study; IIS, Infant Immune Study; NO 2 , nitrogen dioxide air pollution; NY, New York; PM 2.5 , fine particulate matter air pollution; StL, St Louis; SVI, Social Vulnerability Index; URECA, Urban Environment and Childhood Asthma Study; WHEALS, Wayne County Health Environment Allergy and Asthma Longitudinal Study.JAMA Network Open | Environmental Health Early-Life Exposure to Air Pollution and Childhood Asthma Cumulative Incidence JAMA Network Open.2024;7(2):e240535.doi:10.1001/jamanetworkopen.2024.0535(Reprinted) February 28, 2024 6/14 Downloaded from jamanetwork.comby guest on 03/10/2024

Table 2 .
Risk of Asthma Incidence and Odds of Asthma for PM 2.5 and NO 2 for Early Years of Life a The survival analysis assessed asthma incidence though the first 4 and 11 years of life.b The logistic regression assessed odds of asthma through age 4 and 11 years.

Table 3 .
PM 2.5 and NO 2 by Individual Characteristics and Selected Neighborhood Characteristics a (continued)

eTable 1 .
Overview of Participating CREW Cohorts eTable 2. Child's Demographic Characteristics and Caregiver Demographic Characteristics Among CREW Participants When Asthma Incidence Is Missing eTable 3. Child's Demographic Characteristics, Caregiver Demographic Characteristics and Distribution of Respiratory Health Outcomes Among CREW Participants When PM 2.5 Averaged Over Years 1-3 Is Missing eTable 4. Child's Demographic Characteristics, Caregiver Demographic Characteristics and Distribution of Respiratory Health Outcomes Among CREW Participants When NO 2 Averaged Over Years 1-3 Is Missing eFigure 1. Social Vulnerability Index (SVI) and Its Domains eFigure 2. Child Opportunity Index (COI) and Its Domains eFigure 3. Strobe Diagram of Analytical Cohort eFigure 4. Correlations Among PM 2.5 , NO 2 , U.S. Census Variables and COI and SVI eFigure 5. Odds Ratios (OR) of Asthma by Age 4 and 11 (A) and Hazard Ratios (HR) of Asthma Incidence (B) for PM 2.5 and NO 2 for the First Year of Life and Up to 4 Years and for the Averages of Years 1 and 2 and 1 Through 4 eFigure 6.Effect Modification of PM 2.5 by Neighborhood Characteristics eFigure 7. Effect Modification of NO 2.5 by Neighborhood Characteristics eFigure 8. Sensitivity Analysis: Odds Ratio of Asthma by Age 4 With Persistent Wheeze for One IQR Increase in Each Exposure Average eFigure 9. Odds Ratios (OR) of Asthma by Age 4 and Ages 5-11 for an IQR Increase in PM 2.5 and NO 2 for the First Year of Life and for the Averages of Year 1 and 2 and 1 Through 4 Using a Multinomial Regression eReferences