Children’s health impacts from a proposed decarbonization policy in the transportation sector in the Eastern United States

Health impact assessments have estimated substantial health co-benefits of climate change mitigation strategies due to reductions in air pollution in the US; however, few studies have considered children’s health impacts and related equity implications. We estimated the potential health co-benefits to children related to improved air quality associated with various emissions cap and investment scenarios for the transportation sector in the Northeastern and Mid-Atlantic US proposed under the Transportation and Climate Initiative (TCI). We modeled changes in ambient fine particulate matter and nitrogen dioxide between 2022 and 2032 associated with on-road transportation sector emissions under nine hypothetical carbon dioxide (CO2) emissions cap and investment scenarios proposed under TCI using the Community Multiscale Air Quality (CMAQ) model version 5.2. We estimated potential health co-benefits for adverse birth and pediatric respiratory and neurodevelopmental outcomes using an expanded version of the Environmental Benefits Mapping and Analysis Program, known as BenMAPR. We also examined impacts on pediatric asthma exacerbations across racial/ethnic groups. We found that health benefits to children increased as the CO2 emission caps became more ambitious. The combination of the highest emissions cap (25%) and the investment scenario which prioritized public transit improvement (Diversified strategy) conferred the greatest children’s health benefits for the majority of health outcomes considered, resulting in approximately $82 million per year in economic savings. Assessment of the distribution of avoided pediatric asthma exacerbations showed benefits across all racial and ethnic groups, with a slightly greater reduction in cases for non-White populations. Decarbonization policies in the transportation sector in the Eastern US have the potential to provide important air quality and pediatric health co-benefits.


Introduction
Fossil fuel combustion from on-road transportation is a major source of urban air pollution and one of the largest contributors to anthropogenic greenhouse gas (GHG) emissions and climate change in the United States (US) (US EPA 2020).Air pollution from on-road transportation is among the leading drivers of pollutant-related mortality in the US (Caiazzo et al 2013) and globally (Anenberg et al 2019).
Epidemiological studies have documented evidence for increased mortality risk associated with long-term ambient exposure to combustion-related pollutants attributable to on-road vehicles, including particulate matter (PM), ground-level ozone (O 3 ), nitrogen dioxide (NO 2 ), and polycyclic aromatic hydrocarbons (Dockery et al 1993, Bell et al 2004, Jerrett et al 2009, Crouse et al 2015).A robust body of literature has also shown that prenatal and postnatal exposure to air pollution can injure the developing brain, cause lung damage, and increase risk for low birth weight, premature birth, new onset asthma, asthma exacerbations, and respiratory infections (Khreis et al 2017, Orellano et al 2017, Horne et al 2018, Landrigan et al 2019, Perera et al 2019).These adverse health outcomes often persist, affecting health and functioning throughout the life course.
Climate change mitigation strategies targeted at decreasing fossil fuel combustion can provide significant ancillary reductions in ambient air pollution and related public health benefits (Gao et al 2018).Many health impact assessments (HIA) have estimated potential health co-benefits of climate change mitigation strategies and air pollution regulations in the US and globally (Anenberg et al 2012, Orru et al 2013, Chart-asa and Gibson 2015, Likhvar et al 2015, Seposo et al 2018, Xie et al 2020).Few studies have highlighted children's health impacts (Perera et al 2020); and most have overlooked the sociodemographic and geographic distributions of benefits (Levy et al 2007).As such, children's health impacts and the equity implications for pediatric outcomes have not been adequately addressed (Wong et al 2004, Benmarhnia et al 2014, Perera et al 2019).
The Transportation and Climate Initiative (TCI), a proposed cap and invest program aimed at reducing CO 2 emissions from the transportation sector (TCI 2019), is one example of a climate change mitigation policy with potential for health co-benefits.As proposed, TCI would have placed a cap on CO 2 emissions from motor vehicle gasoline and on-road diesel combustion in 12 participating Northeast and Mid-Atlantic US states and the District of Columbia.Regulated fuel suppliers were to be required to purchase CO 2 emission allowances from auctions.Auction proceeds were to be invested in clean transportation programs and projects under states' discretion, with a minimum of 35% required to be invested in initiatives that benefited underserved and overburdened communities (TCI 2020).Although the program will not be put into effect in its present form, modeling air pollution reduction and related health impacts resulting from the proposed program provides a blueprint for assessing similar policies and offers relevant lessons for future decision-making around transportation and climate change mitigation.
We used an air pollution HIA framework similar to Buonocore et al (2023) to estimate the potential health co-benefits to children related to improved air quality associated with various CO 2 emissions cap and investment scenarios for the transportation sector in the Northeast US proposed under TCI.We focused on children's health and the distribution of health impacts across racial/ethnic groups in recognition of the unique susceptibility of the fetus, infant, and child to air pollution (Perera and Nadeau 2022) and of the disparities in exposure by race/ethnicity and income (Clark et al 2017, Tessum et al 2021).
This study builds on our initial HIA under nine illustrative TCI policy scenarios which focused on adult mortality and several respiratory health outcomes traditionally included in most HIA studies on regional air pollution (Arunachalam et al in preparation).As in a previously published assessment of TCI limited to New York City (NYC) (Coomes et al 2022), we have included health outcomes that are not commonly assessed: preterm birth (PTB), term low birthweight (TLBW), lower respiratory infections (LRI) in children, pediatric asthma incidence, and autism spectrum disorder (ASD).We also included the impacts on infant mortality, bronchitis, and lower respiratory symptoms (LRS) in children, as well as pediatric asthma hospital admissions (HA), emergency department (ED) visits, and asthma exacerbations generally not requiring medical intervention.

Selection of policy and reference scenarios
Twelve Northeast and Mid-Atlantic US states and the District of Columbia were included in this analysis: Connecticut, Delaware, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont, Virginia, and Washington DC (figure 1).
The nine TCI policy scenarios and a reference scenario ('Business as usual' [BAU] scenario for the year 2032) considered in this analysis were developed by the Georgetown Climate Center and all 13 jurisdictions participating at the time.The scenarios include combinations of three CO 2 caps of 20%, 22%, and 25% applied to three different investment strategies: Diversified, Hybrid and Max GHG.The Diversified strategy prioritized public transit improvement and active mobility, while the Max GHG scenario focused primarily on electrifying vehicles and switching to cleaner fuel.The Hybrid scenario was a blend of the Diversified and Max GHG strategies.All nine scenarios are summarized in table S1.The changes in air pollution concentrations under each of the nine scenarios were modeled between a baseline year, 2022, and the proposed year of full TCI implementation, 2032.We present the results for all nine scenarios in supplemental material but focused primarily on the results of the five most distinct combinations of caps and scenarios: 20% and 22% CO 2 caps under the Hybrid scenario; and a 25% CO 2 cap under the Diversified, Max, and Hybrid scenarios.
For each policy scenario, we compared the changes in vehicle miles traveled and other changes in activity against the reference scenario.These changes were used as inputs for our air quality and health impact models.

Modeling change in air quality
To assess air quality impacts under each TCI policy scenario, we estimated changes in on-road transportation sector emissions across the region for all precursor pollutants, including nitrogen oxides (NOx), volatile organic compounds, ammonia (NH 3 ), sulfur dioxide (SO 2 ), and primary PM 2.5 , and modeled the corresponding changes in ambient concentrations using the Community Multiscale Air Quality (CMAQ) model version 5.2 (Arunachalam et al in preparation).These changes in concentrations were modeled at a 12 kilometer (km) × 12 km horizontal grid resolution across the TCI region, with a focus on NO 2 and PM 2.5 given their robust associations with the children's health outcomes under study.

Modeling change in health outcomes
We implemented our health impact calculations using the Environmental Benefits Mapping and Analysis Program-R (BenMAPR), a model similar to BenMAP that runs in the R statistical programming language (Arter et al 2021, Raifman et al 2021).BenMAP is a publicly available software supported by the US EPA that has been widely used to estimate air pollution-related illnesses and deaths and quantify their economic value (US EPA 2014).BenMAP's default database includes concentrationresponse (C-R) functions-estimates of the relationship between changes in ambient pollution concentrations and incidences-for more commonly assessed health endpoints.
We expanded the library of C-R functions provided in BenMAP to include C-R functions from the epidemiological literature for infant mortality, pediatric asthma incidence, pediatric asthma exacerbations, pediatric asthma HA and ED visits, pediatric LRI and LRS, pediatric bronchitis, PTB, TLBW, and ASD, as described in Coomes et al (2022).These outcomes are causally or likely to be causally related to PM 2.5 and NO 2 exposure based on a large body of epidemiological and toxicological evidence.A systematic review of the epidemiological literature provided the C-R functions for many of these child health outcomes (Perera et al 2019).We subsequently reviewed more recent epidemiological studies to identify relevant C-R functions for additional health outcomes of interest and updated the earlier C-R functions as needed, including for infant mortality (from birth until 1 year of age); pediatric asthma exacerbations, hospitalizations and ED visits; pediatric LRI; and pediatric bronchitis (Coomes et al 2022).We used the existing C-R function in BenMAP for LRS.(See tables S2 and S3 for C-R functions and sources).We included 95% confidence intervals (CI).

Baseline incidence data
Baseline incidence rates for infant mortality, PTB, and TLBW were obtained from the US Centers for Disease Control and Prevention Wide-Ranging Online Database for Epidemiological Research (CDC WONDER).We used 5 year average infant mortality rates for the most recent years available at the time of our analysis (2014)(2015)(2016)(2017)(2018) in order to reduce the potential bias that a single outlier year could introduce, as well as to avoid overestimating baseline rates given the steady decline in infant mortality rates in the US over the last 15 years (Matthews and Driscoll 2019).PTB was defined as all births that occurred before the 37th week of pregnancy.TLBW was defined as all births that occurred on or after the 37th week of pregnancy where the birth weight was <2500 g.The incidence rates for all birth outcomes were obtained at the county-level, where available, and substituted with state-level rates otherwise.
We relied on the national asthma incidence rate reported in Winer et al (2012), as asthma incidence rates at a more granular level were not available across the region.State-level incidences for both asthma HA and ED visits for children aged 0-17 were obtained from the CDC Healthcare Cost and Utilization Project Ambulatory Healthcare Data (CDC 2018a).Because data for the incidence of non-medical asthma exacerbation were not available, baseline incidence for this outcome was estimated following the approach used in other HIAs for air pollution regulations.We utilized CDC data on statelevel asthma prevalence, where available, and used the national asthma prevalence otherwise (CDC 2017(CDC , 2018b)); we then applied the average daily rates of new 'wheeze' episodes experienced in a Southern Californian cohort of asthmatic children (Ostro et al 2001) to the CDC prevalence values to estimate the rate of non-medical asthma exacerbations, as described in Coomes et al (2022).
The LRI incidence rate was obtained from a cohort study of children aged 5-14 in Boston, MA (Lazarus et al 2001).We used the baseline incidence rates for acute bronchitis and LRS currently in BenMAP (Schwartz et al 1994, American Lung Association 2002).Finally, the incidence rate for ASD was drawn from a Southern California cohort of infants (Jo et al 2019).

Demographic data
We obtained demographic data from the 2014-2018 US Census 5-year American Community Survey (ACS), dates generally consistent with the baseline disease prevalence data.We extracted census tractlevel population data on the race and ethnicity of 5-to 17 year-olds.The racial groups included White, Black, Asian, American Indian or Alaska Native, Native Hawaiian or Pacific Islander, those identifying as more than one race, and those identifying as 'Other' race.The ethnic groups included were Hispanics of any race, compared to non-Hispanic Whites as these were the only ethnic identities with available agestratified data.Each census tract was assigned the air pollution data from the 12 km × 12 km grid cell it fell within.Across the entire gridded region, the median number of census tracts overlapping with each grid cell was three.An area-weighted average was used for census tracts that spanned multiple grid cells.All residents within each census tract, regardless of subpopulation, were assumed to be exposed to the same concentrations, and these values were used to estimate population-weighted averages by subgroup.

Economic valuation of health impacts
We estimated the economic value of the change in incidences of our health outcomes by multiplying the cost-per-case estimates drawn from the literature in Shea et al (2020) by the reduction in cases modeled in BenMAPR (table S4).The cost estimates mainly included direct costs (medical costs, special education costs, hospital costs), and in some cases indirect costs (lost productivity, social costs, lost lifetime earnings, crime and delinquency costs).The majority focused on the direct, short-term costs and did not reflect potential lifetime effects.As in previous studies (Perera et al 2020, Coomes et al 2022), the cost-percase estimate of PTB additionally includes the cost of reduction in IQ points, hence lifetime earnings lost, associated with this health outcome (Gould 2009).This resulted in a cost-per-case of PTB of $330 028.We reported all values in $2015 USD.For health outcomes where only one cost estimate was available, we used it to calculate the low, central, and high estimates of the monetary value of the health burden.In cases where some form of uncertainty in the cost of illness was available (i.e.low, medium, and high values, or 95% CI), we used the central cost of illness estimate and the central health outcome estimate to calculate a central estimate of the total value of the health burden.We calculated the low and high monetized values of the health impacts using the upper and lower bounds of health outcomes' CIs, along with the low and high values of the cost of illness estimate to calculate low and high total monetized values for these health outcomes.

Equity analysis
In order to understand how the scenarios impacted existing health inequities, we focused on asthma exacerbations in children aged 5-17, the outcome for which we had the most robust health incidence data and enough spatial heterogeneity to analyze differences between groups in avoided cases.Based on the expected changes in pollutant concentrations (PM 2.5 , and NO 2 ) under each TCI scenario compared to the 2032 BAU case, we estimated the number of avoided cases of asthma exacerbations and determined their distribution between the racial/ethnic subgroups.Due to data limitations, we assumed the same baseline level of asthma exacerbation incidence for children 5-17 across all the defined racial and ethnic subgroups within a given geographic area.
In order to further characterize how these changes impacted existing health inequities, we used the Atkinson Index, a relative inequality measure, to quantify how the changes in pollutant concentrations and associated asthma exacerbations would ameliorate or exacerbate inequalities across racial and ethnic groups (Atkinson 1970).The Atkinson Index measure can be decomposed to assess risk inequalities between the subgroups, with a between-group Atkinson Index that ranges from 0 to 1, with 0 representing complete equality and 1 representing maximum inequality.This measure has been used to compare equity Autism spectrum disorder 1.8 (1.7-2) 3.6 (3.3-4) 6.9 (6.2-7.5)7.2 (6.5-7.9)6.7 (6-7.3) Note: Estimates for preterm birth, lower respiratory infections, lower respiratory symptoms, acute bronchitis, and autism spectrum disorder are for PM2.5 alone.
of health benefits from other air pollution mitigation measures (Levy et al 2006(Levy et al , 2007)).

Main analysis
Taking our five selected TCI strategies, we estimated greater pediatric health benefits associated with increasing CO 2 caps, with avoided cases approximately doubling with each increase in cap ambition (i.e. from 20% to 22% to 25%), consistent with the magnitude of CO 2 emissions reductions (table 1 shows results under combined PM 2.5, NO 2 reductions).The combination of the highest CO 2 cap (25%) and the Diversified scenario, which proposed the greatest investment in mass transportation, yielded the most health benefits to children and infants for six out of 11 health outcomes.Under this cap/scenario combination, we estimated the largest number of avoided asthma exacerbations not requiring medical intervention (cases avoided [95% CI] = 56 000 [1300-110 000]).We estimated 2100 (95% CI = 780-3400) and 230 (95% CI = 99-320) avoided cases of LRS and LRI, respectively, and 231 (95% CI = 91-340) avoided cases of new onset asthma.We also estimated 24 (95% CI = 5.7-41) and 23 (95% CI = 4.1-40) avoided PTBs and TLBW births, respectively, under this cap/scenario combination.Estimates for avoided cases under all nine scenarios, and for PM 2.5 and NO 2 separately, are included in table S5.Under the 25% GHG cap and Diversified scenario, NO 2 reductions yielded more than 10-fold more avoided asthma exacerbations than PM 2.5 .For both pollutants, the greatest reduction in cases was seen in the more densely populated areas in the Eastern US, particularly the NYC and Boston metropolitan areas and the Philadelphia/Baltimore/Washington corridor (figure 2).
monetary values for avoided cases under all nine scenarios, and for PM 2.5 and NO 2 separately, are included in table S6.

Equity analysis
For all pollutants studied, each TCI scenario resulted in a decrease in asthma exacerbation cases (not requiring medical intervention) in children across all racial and ethnic groups when compared to the BAU scenario (tables 3, S7 and S8).Focusing on the results from combined reductions in PM 2.5 and NO 2 , we found the following reductions in asthma exacerbations under the 25% cap/Diversified scenario: 5.86% for children identifying as 'Other' , 5.63% for American Indian or Alaska Native, 5.59% for Asian, 5.55% for Black, and 5.53% for White children when compared to the BAU scenario.Additionally, we estimated 5.73% fewer asthma exacerbations among Hispanic children, and a 5.59% reduction among Non-Hispanic White children, compared to the reference scenario.As in the main analysis, reduction in NO 2 emissions resulted in greater reductions in asthma exacerbation cases across all racial and ethnic groups compared to reductions resulting from decreased PM 2.5 emissions.Across all TCI scenarios, the results using the Atkinson Index showed a decrease in racial and ethnic inequality in asthma exacerbations for children aged 5-17 when compared to the BAU scenario.Under the 25% CO 2 cap, the greatest improvements in equity of health outcomes across racial/ethnic groups related to NO 2 exposure were seen for the Diversified strategy, followed by the Hybrid scenario and, finally, the Max GHG scenario (figure S3).For PM 2.5 exposure, the Hybrid scenario showed the greatest improvements in equity of health outcomes across racial/ethnic groups, followed by the Diversified scenario (figure S3).We estimated increasing benefits across all racial/ethnic groups and scenarios as the CO 2 cap became more stringent (figure S1).
When comparing each scenario's ability to reduce children's asthma exacerbations and also reduce existing racial and ethnic inequities in these exacerbations, we found that for each cap and pollutant, the scenario that conferred the greatest health benefits generally also conferred the greatest improvements in health equity.As an example of this, figure S1 shows how the TCI scenarios compare in both the magnitude of health benefits achieved and in the reduction of inequities between racial identities as measured by the between-group Atkinson Index.When considering the decrease in NO 2 emissions for each cap, the Diversified scenario resulted in the greatest reduction in asthma exacerbations for this population as well as in racial health inequities in these exacerbations, followed by the Hybrid scenario.When comparing outcomes between Hispanic and Non-Hispanic White children, we found a similar pattern, though notably smaller reductions, when compared to our results for racial identities (figure S2).The difference between scenarios become more pronounced as the GHG cap increased.

Discussion
To our knowledge, this study includes the broadest suite of children's health outcomes currently available in the HIA literature, providing a comprehensive picture of the benefits to children of different transportation policy strategies in the Northeast and Mid-Atlantic US. Results from our model indicate substantial health benefits to children and associated economic savings by 2032 under the proposed TCI scenarios.Previously, the benefits to adult health were assessed in Arunachalam et al (in preparation), with a total of $6.2 billion dollars under the 25% Diversified scenario.The economic benefits of both acute and chronic effects in children are very modest compared to those from prevention of adult death, with $82 million of total benefits to children under the same scenario.However, there are substantial policy interests in child health outcomes specifically, and the monetized child health benefits are likely underestimated because most cost-per-case valuations consider immediate or short-term costs.
The benefits increased as the cap on CO 2 became more stringent, reducing CO 2 emissions by 1%, 3%, and 5%, respectively, below the reference case.Assessment of the distribution of the avoided asthma exacerbations indicated that all racial and ethnic groups benefited, with modestly greater reduction in cases for non-White populations and associated reductions in the between-group Atkinson index.The investment scenario that devoted the most resources to mass transportation tended to perform the best in terms of avoiding cases of illness and developmental impairment in children, as well as in reducing inequality.
Although the epidemiological evidence for the association between many adverse health outcomes and NO 2 is not as comprehensive as that for PM 2.5 , it is important to include NO 2 in a transportation related HIA given its close linkage with vehicle emissions.Where we were able to compare results for both PM 2.5 and NO 2 for the same outcome, our analysis indicated that NO 2 accounted for a majority of the total benefit, especially for childhood asthma.
This study demonstrates the importance of highlighting children's health outcomes in HIAs, an important gap in the field.Since the economic benefits of climate and air pollution policies are dominated by adult mortality, the important societal effects and potential life course impacts of children's health outcomes are not sufficiently considered in most HIAs.Moreover, the life course impacts of PM 2.5 or NO 2 exposure, including effects on future health, cognitive function, and lifetime earnings, are significant and are commonly underestimated.For example, developmental issues at birth and respiratory illnesses in early childhood often persist throughout childhood and into adulthood.This can cause lost days of school and work as well as additional medical costs for related illnesses, such as asthma exacerbations and cardiovascular and pulmonary diseases (Luu et al 2016).These lifetime effects are not reflected in the cost of illness values for child health outcomes currently available in the valuation literature.Additionally, this suggests that the value of the health benefits in our study may be underestimated.
Fossil fuel combustion is the predominant source of both GHGs and air pollutants; hence decarbonization policies achieve benefits for health in mitigating the impacts of climate change and reducing coemitted air pollutant emissions.The air quality and health co-benefits are greatest where large populations experience large air quality improvements and, unlike the global benefits of reducing GHG emissions, the benefits are immediate (Gallagher and Holloway 2020).
There were several key assumptions in this analysis of the hypothetical program.First, it was assumed that all 13 political entities (Connecticut, Delaware, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont, Virginia, and Washington DC) would participate.Also, the scenarios included in this analysis assumed that 83%-92% of the proceeds from auctions of carbon allowances would be reinvested into transportation.Broadly, as TCI is unlikely to be implemented in the form that we modeled, our analyses should be considered as providing guidance regarding the relative benefits of different types of transportation interventions rather than specific policy measures.
A limitation of this study is that the analysis of changes in air quality used to estimate health impacts was conducted at a 12 km × 12 km scale, which did not allow us to examine impacts at a neighborhood level, thereby limiting our equity analysis to differences that manifest at a regional scale.We focused our analysis on PM 2.5 and NO 2 and did not include additional transportation-related pollutants (e.g.O 3 , CO, black carbon) due to the lack of reliable C-R functions for all the children's health endpoints of interest in this paper.A previous analysis (Arter et al 2021) showed that O 3 has a comparable order of magnitude impact on adult mortality as PM 2.5 from transportation sources in this region, indicating potential downward bias from this omission.The lack of neighborhood-level baseline health incidence data across the region further limited the equity analysis, especially since baseline rates of asthma exacerbations stratified by racial and ethnic subgroups were not available across the region.While our analysis provided a general estimate of how the TCI scenarios could impact existing health inequities, there is a need for more geographically resolved data on race/ethnicity and socioeconomic groups, as well as more finely resolved air pollution and baseline health data to fully assess equity impacts.This type of analysis is feasible and has been implemented in NYC in another context where it has revealed substantial differences in air pollution exposures from neighborhood-scale characterization of emissions sources (Shukla et al 2022).Due to limitations in age-stratified and ethnicity-stratified data, we were only able to compare Hispanics of any race vs. non-Hispanic Whites (i.e.excluding non-Hispanic Blacks or non-Hispanic Asians) in our equity analysis of asthma exacerbations among children aged 5-17.
Strengths of the analysis included the use of a validated comprehensive model for air quality that includes detailed treatment of the pollutants studied here and the incorporation of updated C-R and cost functions for a broad array of child health outcomes.Although we only focused on a single outcome (child asthma exacerbations) to illustrate the equity implications of alternative scenarios, the approach can readily be applied to other outcomes.Our analysis highlights the usefulness of tools to assess equity in the distribution of potential benefits of climate and clean air policies, as has been suggested by previous studies (Levy et al 2007, 2009, Clark et al 2017, Cushing et al 2018).Finally, this analysis has relevance to the evaluation of health co-benefits for children of other policies aimed at mitigating climate change.The methods used in this analysis are applicable to other policies, and the lessons learned here are generalizable, including the likelihood that an investment in mass transit will yield greater health benefits and more improvements in equity by targeting emissions reductions in more densely populated areas with higher baseline air pollution.The analysis also demonstrated the value of assessing a wide spectrum of health outcomes starting at birth to highlight the magnitude of health benefits that can be gained for this vulnerable population through targeted emissions reduction measures.

Figure 1 .
Figure 1.Study area, 13 formerly-participating entities in the Northeast and Mid-Atlantic United States.

Table 2 .
Estimated monetary value of avoided pediatric health outcomes (95% CIs) associated with combined PM2.5 and NO2 reductions under 5 TCI scenarios.