Social Inequities Impact Infant Mortality due to Congenital Heart Disease

Abstract

Objectives:

To evaluate how educational, economic and racial residential segregation may impact congenital heart disease infant mortality (CHD-IM).

Study Design:

This is a population-based U.S. ecological study.

Methods:

This study evaluated linked live birth-infant death files from the National Center for Health Statistics for live births from 2006 to 2018 with cause of death attributed to CHD. Maternal race and education data was obtained from the live birth-infant death files, and income data was obtained from the American Community Survey. A spatial social polarization measure termed the Index of Concentration at the Extremes (ICE) was calculated and split by quintiles for maternal education, household income, and race for all U.S. counties (n=3,142). The lowest quintile represents counties with highest concentration of disadvantaged groups (income< $25K, non-Hispanic Black, no high school degree). Proximity to a pediatric cardiac center was also analyzed in a categorical manner based on whether each county was in a metropolitan area with a U.S. News and World Report top 50 ranked pediatric cardiac center (PCC), a lower ranked PCC, or not proximal to any PCC.

Results:

Between 2006 and 2018, 17,489 infant deaths were due to CHD, an unadjusted CHD-IM of 0.33 deaths/1,000 live births. The risk of CHD-IM was 1.5 times greater among those in the lowest ICE-education quintile (0.41[0.39–0.44] vs 0.28 deaths/1,000 live births[0.27–0.29], p<0.0001) and the lowest ICE-income quintile (0.44[0.41–0.47] vs 0.29[0.28–0.30], p<0.0001) in comparison to those in the highest quintiles. CHD-IM increases with higher ICE-race value (counties with a higher concentration of non-Hispanic White mothers). However after adjusting for proximity to a U.S. News and World Report top-50 ranked pediatric cardiac center in the multivariable models, CHD-IM decreases with higher ICE-race value.

Conclusions:

Counties with the highest concentration of lower educated mothers and the highest concentration of low income households were associated with higher rates of CHD-IM. Mortality as a function of race is more complicated, and requires further investigation.

Introduction

Congenital heart disease (CHD) is the leading cause of infant mortality from birth defects in the United States (U.S.).¹ Important advancements have been made in the detection and treatment of CHD over time, leading to improved survival of infants with CHD. However, significant infant mortality persists, with prior studies reporting considerable socioeconomic and racial disparities.^2–7

The impact of individual patient management and treatment factors on infant mortality secondary to congenital heart disease (CHD-IM) has been well-documented. However, the role of social determinants of health in patient outcomes is increasingly being recognized.⁸ Prior studies have demonstrated that non-Hispanic Black patients and Hispanic patients have worse post-operative outcomes following congenital heart surgery when compared to non-Hispanic White patients.^3,5 Another study showed that patients with CHD from the lowest income neighborhoods have greater odds of mortality and longer lengths of stay compared to patients from high income neighborhoods.² In addition, patients from the lowest socio-economic tertile who underwent stage 1 single ventricle palliation had more complications compared to patients in the highest socio-economic tertile.⁹

Population health disparities are increasingly mapped by geographic regions.¹⁰ Children’s health is affected by not just their individual family circumstances, but also the neighborhoods in which they grow up.^11,12 However, there is limited prior research investigating social inequities that increase the risk of CHD-IM in one county while protecting against it in another. A recently published study demonstrated significant geographic variation in CHD-IM with lower mortality risk correlating with closer proximity to a top-50 ranked pediatric cardiac center, emphasizing the importance of focusing on care delivery in areas with higher CHD-IM.¹³ Epidemiologic studies of infant mortality have noted the importance of understanding spatial and social polarization in socioeconomic factors to advance health equity and improve disparities in infant mortality outcomes.^10,14,15 A validated public health measure known as the Index of Concentration at the Extremes (ICE), used in this study, enables quantifying the economic and racial segregation that have shaped the geographic separation of socially defined groups.^10,14 ICE allows for the evaluation of how a county’s concentration of “privilege” and “disadvantage” may influence outcomes when compared to other counties.¹⁶ ICE has previously been used to study inequities across many areas of health outcomes, including life expectancy, infant mortality, pregnancy mortality, hypertension and cancer.^17–21

The purpose of our study is to assess the impact of geographic social inequities on CHD-IM. The study aims to identify how concentration of “privilege” and “disadvantage” in the areas of income, education, and race affect CHD-IM. We also aimed to assess whether geographic social inequities could be mitigated by being in closer proximity to a U.S. News and World Report top 50 ranked pediatric cardiac center (T50-PCC).

Methods

Study Design

A population-based U.S. ecological study was performed using linked live birth-infant death files from the National Center for Health Statistics for live births from 2006 to 2018. All deaths of infants (<365 days of age) with cause of death attributed to CHD on the death certificate were included. The outcome of death secondary to CHD is based on the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes Q20·0 through Q28·9. CHD-IM, evaluated by county, was calculated by dividing the number of infant deaths by the number of live births in a year. All-cause infant mortality was also evaluated by county to evaluate if CHD-IM findings were specific to CHD or were generalizable to all-cause infant mortality. Infant mortality, maternal education and maternal race/ethnicity data were collected from the National Center for Health Statistics.²² Household income data were obtained from the American Community Survey conducted by the U.S. Census Bureau.²³ American Community Survey 5-year annual average (2013–2018) household incomes by county were used.²³ The American Community Survey publishes 1-year and 5-year average household income data, and for the purposes of this study, the average income data (not equivalized) over the most recent years in our study was used (2013–2018).

This study used Index of Concentration at the Extremes (ICE), a validated spatial social polarization measure, which has been utilized to evaluate geographic social inequities.^10,16 This measure quantifies extremes of privilege and disadvantage of social groups within a county in a single metric, to highlight inequities between counties.^16,24 The rationale for simultaneously including both privilege and disadvantage is based on evidence that suggests a higher concentration of privilege within a county provides health benefits to all residents, and that this would not be reflected by examining disadvantage alone.^16,24 ICE was selected over other commonly used population measures of social and economic inequity (e.g. Gini coefficient and Index of Dissimilarity) because of its ability to be informative on a neighborhood or county level.¹⁰ In addition, ICE avoids the well-known issue of multi-collinearity that occurs when attempting to simultaneously analyze opposite measures of both poverty and wealth or percent Black and percent White in a population.^10,25 ICE was calculated separately for maternal education, income, and race, for all U.S. counties (n=3,142).

ICE is calculated as¹⁶:

$${\text{ICE}}_{\text{c}}=\left(P_c-D_c\right)/T_c$$

where $Pc$ is the number of live births to mothers (or households) in the “privileged” group, $Dc$ is the number of live births to mothers (or households) in the “disadvantaged” group, and $Tc$ is the total number of live births to mothers (or households) in a county $\left(c\right)$ with known data (Table 1). For example, when calculating ICE-education, $Pc$ is the number of live births to mothers with a college degree in a county, $Dc$ is the number of live births to mothers with no high school degree, and $Tc$ is the total number of live births to mothers in a county with known education data. An ICE value of 1 indicates that every mother within a county is in the “privileged” group, whereas a value of −1 indicates that every mother is in the “disadvantaged” group. The continuous ICE variables for race, income, and education were split into quintiles with approximately equal number of counties, with the fifth quintile (“5”) representing counties with the highest concentration of individuals in the “privileged” category.

Table 1.

Index of Concentration at the Extremes (ICE) Categories for Privileged and Disadvantaged

	Privileged (P)	Disadvantaged (D)
Education	College educated	No high school degree
Income	≥$100,000	<$25,000
Race	Non-Hispanic White	Non-Hispanic Black

For ICE-income, an annual household income of ≥ $100,000 was defined as “privileged,” and < $25,000 as “disadvantaged,” based on US Census data estimates of the 20^th and 80^th percentiles of national household income in 2013 and as used in prior studies employing the ICE measure.^17,18 The lowest ICE-income quintile (“1”) represents the counties with the highest concentration of households with annual income <$25,000 and lowest concentration of households with an annual income ≥$100,000; whereas the highest ICE quintile (“5”) demonstrates the counties with the highest concentration of households with higher annual income. For ICE-education, the group of mothers with a college degree was defined as “privileged,” and the group of mothers without a high school diploma was defined as “disadvantaged,” as categorized in prior studies employing the ICE measure for education.^15,18

For the purposes of this study, race was conceptualized as a social construct arising from inequitable race relations.¹⁰ Maternal race and ethnicity were divided into the categories of non-Hispanic White individuals, non-Hispanic Black individuals, Hispanic individuals, and Other (included non-Hispanic Asian Americans, non-Hispanic Hispanic Pacific Islanders, and non-Hispanic American Indians). Given the history of residential racial segregation in the U.S experienced by non-Hispanic Black individuals and the known racial inequities in Black infant mortality,^19,26 this study focused on CHD-IM in non-Hispanic Black infants in comparison to non-Hispanic White infants. Mother’s race in the National Center for Health Statistics dataset is from self-identified race reported on the infant’s birth certificate. For ICE-race, non-Hispanic White was defined as “privileged,” and non-Hispanic Black was defined as “disadvantaged.” Mothers who did not identify as either of these two categories were included as ICE-race value of zero (8.6% of mothers listed as “Other”, 23.6% as Hispanic, <1% with missing data for race). Covariates of proximity to a pediatric cardiac center were categorical based on whether each county was in a metropolitan area with a U.S. News and World Report top 50 ranked pediatric cardiac center (T50-PCC), a lower ranked PCC, or not proximal to any PCC.²⁷ The cutoff of 50 centers was chosen because U.S. News and World Report only ranks 50 centers.¹³ For the purpose of this study, a metropolitan area was defined as contiguous counties with a high amount of social and economic integration that include at least one urbanized area with a population over 50,000 individuals.¹³ Race covariate was defined as the fraction of mothers who identified as Non-Hispanic White, Non-Hispanic Black, Hispanic, or Other.

Statistical Analysis

Observed CHD-IM, ICE-race, -income, and -education were calculated for each county. Approximately equal numbers of counties were grouped independently into quintiles by ICE-race, -income, and -education. Average observed CHD-IM with upper and lower confidence intervals were calculated for each quintile. Because county CHD-IM for each quintile was not normally distributed, a non-parametric test was used to determine the significance of the variation in observed CHD-IM with quintile.

A total of 208 live births and 2 CHD deaths to mothers with missing race/ethnicity or maternal county of residence were excluded from the analysis. In many cases, maternal education data for individual counties was systematically missing, as some states did not report in some years. The available maternal education data for each county was treated as a sample over the entire study period. The average standard error in ICE-education due to sampling was estimated to be 0.005, which was considered sufficiently small to proceed with ICE-education as a covariate.

Negative binomial univariable models were used to smooth CHD-IM by quintile, to calculate mortality ratios between quintiles, and to estimate linear and quadratic trends with ICE-race, - income, and -education as continuous variables. CHD-IM was calculated by quintile by fitting the data to a negative binomial model with a linear predictor consisting of only the ICE quintile without an intercept. Mortality ratios were calculated by fitting the data to the same model with an intercept, leading to quintile 5 being the reference for our study. Negative binomial multivariable models were used to evaluate the effects of ICE-race, -income, and -education as continuous variables and to adjust for other covariates. The negative binomial models were used due to the highly skewed nature of CHD-IM counts by county. The majority of counties have 0 counts of CHD-IM in any given year. We observed the distribution of the number of infant deaths by county to be over-dispersed for a standard Poisson distribution. The negative binomial multivariable models had a better fit for the data than a standard or generalized Poisson model. Using the same methods, subgroup analyses were performed individually for non-Hispanic Black and non-Hispanic White. The three multivariable models were (1) ICE-race with proximity, (2) ICE-income with proximity and race, and (3) ICE-education with proximity and race. Because CHD-IM appeared to vary non-monotonically with ICE-race, the linear predictor in the multivariable models for ICE-race included both linear and quadratic terms for ICE-race.

Analyses were performed using SAS for Windows, Version 9.4.

As this was not human subjects research, no institutional review board approval was obtained. There was no funding for this study.

Results

There were 17,489 infants who died due to congenital heart disease between 2006–2018 in the U.S., translating to an unadjusted CHD-IM rate of 0.33 deaths/1,000 live births. Total births and deaths attributed to CHD divided by sex, race, maternal education and proximity to a T50-PCC are displayed in Table 2. CHD-IM accounted for 5.4% of all-cause infant deaths during this time period.

Table 2.

Characteristics of study population including births, deaths and CHD infant mortality rate from 2006–2018

Observed	Births	%	Deaths	%	Mortality (deaths/1,000 live births)			P-value
					Value	LCL	UCL
All	52,357,666	-	17,489	-	0.33	0. 33	0. 34
Sex								<0.0001
Female	25,566,690	48.8	7,914	45.3	0.31	0.30	0.32
Male	26,790,976	51.2	9,575	54.7	0.36	0.35	0.36
Race								<0.0001
Non-Hispanic White	27,884,452	53.3	8,487	48.5	0.30	0.30	0.31
Non-Hispanic Black	7,619,293	14.6	3,304	18.9	0.43	0.42	0.45
Hispanic	12,356,198	23.6	4,182	23.9	0.34	0.33	0.35
Other	4,497,723	8.6	1,515	8.7	0.34	0.32	0.35
Mother Education								<0.0001
8th grade or less	1,837,537	3.5	846	4.8	0.46	0.43	0.49
9th through 12th grade with no
diploma	5,440,426	10.4	2,169	12.4	0.40	0.38	0.42
High school graduate or GED
completed	10,810,099	20.6	3,969	22.7	0.37	0.36	0.38
Some college credit, no degree	8,719,876	16.7	2,915	16.7	0.33	0.32	0.35
Associate degree	3,266,294	6.2	1,007	5.8	0.31	0.29	0.33
Bachelor’s degree	7,887,229	15.1	1,959	11.2	0.25	0.24	0.26
Master’s degree	3,423,635	6.5	703	4.0	0.21	0.19	0.22
Doctorate or Professional Degree	976,147	1.9	179	1.0	0.18	0.16	0.21
Unknown	540,358	1.0	396	2.3	0.73	0.66	0.81
Missing data	9,456,065	18.1	3,345	19.1	0.35	0.34	0.37
Proximity								<0.0001
Not Proximal to Top 50 PCC	27,868,470	53.2	10,260	58.7	0.37	0.36	0.38
Proximal to a Top 50 PCC	24,489,196	46.8	7,229	41.3	0.30	0.29	0.30

Abbreviations: LCL= lower confidence limit, UCL= upper confidence limit, GED= general educational development, Top 50 PCC = U.S. News and World Report Top 50 Pediatric Cardiac Center

Table 3 displays the breakdown of ICE-race, ICE-education, and ICE-income by quintile. ICE-race quintile 1 had a mean of 0.05 (range −1.00 to 0.34), and the ICE-race quintile 5 mean was 0.96 (range 0.94 to 1.0). ICE-education quintile 1 mean was −0.16 (range −1.00 to −0.09) and quintile 5 had a mean of 0.31 (range 0.20–0.74). ICE-income quintile 1 mean was −0.25 (range - 1.0 to −0.18) and quintile 5 had a mean of 0.18 (range 0.06 to 0.61). The U.S. maps of the geographic breakdown of ICE-education, ICE-income, and ICE-race for this study are available (Online supplement Figure 1).

Table 3.

Univariable Negative Binomial Model: Observed mortality (deaths/1,000 live births) by ICE-race, education and income quintile

Quintile	ICE-race					ICE-education					ICE-income
	ICE			CHD-IM (95% CI)	Mortality Ratio (95% CI)	ICE			CHD-IM (95% CI)	Mortality Ratio (95% CI)	ICE			CHD-IM (95% CI)	Mortality Ratio (95% CI)
	Mean	Min	Max			Mean	Min	Max			Mean	Min	Max
1	0.05	−1.00	0.34	0.36 (0.35, 0.38)	0.84 (0.77–0.91)	−0.16	−1.00	−0.09	0.41 (0.39, 0.44)	1.46 (1.37, 1.56)	−0.25	−1.00	−0.18	0.44 (0.41, 0.47)	1.50 (1.39, 1.61)
2	0.51	0.34	0.67	0.32 (0.31, 0.34)	0.75 (0.69–0.81)	−0.04	−0.09	0.001	0.41 (0.39, 0.43)	1.45 (1.36, 1.54)	−0.13	−0.18	−0.09	0.38 (0.36, 0.40)	1.30 (1.22, 1.39)
3	0.76	0.67	0.85	0.34 (0.32, 0.36)	0.79 (0.72–0.86)	0.05	0.001	0.09	0.37 (0.35, 0.39)	1.30 (1.23, 1.38)	−0.06	−0.09	−0.02	0.40 (0.38, 0.42)	1.36 (1.28, 1.44)
4	0.90	0.85	0.94	0.37 (0.35, 0.40)	0.86 (0.78–0.95)	0.14	0.09	0.20	0.35 (0.33, 0.36)	1.22 (1.15, 1.29)	0.02	−0.02	0.06	0.36 (0.34, 0.37)	1.22 (1.15, 1.29)
5	0.96	0.94	1.00	0.43 (0.40, 0.46)	1 (Ref)	0.31	0.20	0.74	0.28 (0.27, 0.29)	1 (Ref)	0.18	0.06	0.61	0.29 (0.28, 0.30)	1 (Ref)

Abbreviations: ICE=Index of concentrations at the extremes, CHD-IM= infant mortality due to congenital heart disease, CI= confidence interval, Ref=reference value

Counties with the highest concentration of lower educated mothers (ICE-education quintile 1) had a CHD-IM rate of 0.41/1,000 live births (95% CI 0.39–0.44/1,000), 1.5 times higher than counties with the highest concentration of highly educated mothers (CHD-IM rate of 0.28/1,000; 95% CI 0.27–0.29/1,000). Infants of mothers who lived in counties with the highest relative concentration of low income households (ICE-income quintile 1) also had a 1.5 times risk of CHD-IM (CHD-IM rate of 0.44/1,000; 95% CI 0.41–0.47/1,000) compared to those who lived in counties with the highest concentration of high income households (CHD-IM rate of 0.29/1,000; 95% CI 0.28–0.30/1,000). The p-value was <0.001 for variation in mortality with quintile (Table 3, Figure 1).

Figure 1.

Observed congenital heart disease infant mortality by county ICE quintile (top); Observed all-cause infant mortality by county ICE quintile (bottom)

In the univariable model, CHD-IM increased with higher ICE-race value (p<0.0001). The CHD-IM for counties with highest concentration of non-Hispanic White mothers (ICE-race quintile 5) was 0.43/1,000 live births (95% CI 0.40–0.46/1,000), 1.2 times higher risk than ICE-race quintile 1, counties with the highest concentration of non-Hispanic Black mothers (CHD-IM rate 0.36/1,000; 95% CI 0.35–0.38/1,000). When a subgroup analysis of only Non-Hispanic Black live births was performed, there was not a significant decrease in CHD-IM with increasing ICE-income and ICE-education quintiles (p=0.05 respectively). The subgroup analysis for Non-Hispanic White live births demonstrated the same trend as the overall, non-stratified findings, with counties with the highest concentration of low income households and lower educated mothers having higher risk for CHD-IM (Figure 2; p<0.0001 for both).

Figure 2.

Observed congenital heart disease infant mortality by county ICE quintile stratified by race (Non-Hispanic Black -Top; Non-Hispanic White – Below)

In the multivariable models (Supplement table 1), ICE-education and ICE-income remained significantly associated with CHD-IM after adjusting for race and proximity to a T50-PCC (p<0.0001 for both). In the multivariable model when adjusted for proximity to a T50-PCC, CHD-IM decreased with higher ICE-race value (p=0.0005).

Overall infant mortality (not limited to CHD) was also evaluated using the same ICE measures, and demonstrated similar trends to CHD-IM for education and income (Figure 1b). However, there was not a clear trend for ICE-race by quintile; the counties with the lowest and highest concentration of non-Hispanic Black mothers had the highest rates of all-cause infant mortality.

Discussion

The findings from this study add to the growing body of literature on how social inequities affect outcomes in CHD. Live birth-infant death files from infants with CHD were analyzed to evaluate the effect a geographic area (county) may have on a child’s likelihood to survive their first year of life. U.S. counties with the highest concentration of education and income disadvantage were associated with higher rates of CHD-IM. These CHD-IM findings were independent of proximity to a T50-PCC. Interestingly, these same trends were seen when evaluating all-cause infant mortality, suggesting that the factors contributing to ICE-income and education impact infant mortality regardless of etiology. Mortality as a function of race is more complicated.

ICE has been validated as a promising measure of structural racism and has been utilized for analysis in several adult and pediatric studies.^{10,14,15,19,20} To our knowledge, no prior studies have used the ICE measure to evaluate the potential effect of racial and economic segregation on CHD-IM. The findings from our study are consistent with prior studies, that have demonstrated higher CHD-IM in areas of low socioeconomic status.^2,9,13,28,29 For example, previous analysis of the Pediatric Heart Network’s Single Ventricle Reconstruction trial data showed that living in a census block 5.4–13% below the poverty level was associated with higher interstage mortality.³⁰ Our findings may reflect that for families living in counties with lower income, it may be less feasible financially to travel to regionalized pediatric cardiac centers with better clinical outcomes. However, the persistence of this finding even after adjusting for proximity to a T50-PCC, suggests there are likely additional factors to consider. Differences on a county level in rates of prenatal diagnosis, prematurity and termination may also affect CHD-IM outcomes.^2,4,31 Our study also demonstrated the potential impact of maternal education on CHD-IM. There are limited prior studies evaluating how maternal education may affect CHD outcomes. One large meta-analysis also demonstrated that low maternal education was associated with increased mortality in the first year of life.³² Using birth defects surveillance data across four states, Kucik et. al demonstrated that for infants with CHD, being born to a mother with less than a high school education was associated with poorer survival.⁴ Low maternal education may be a surrogate indicator for poverty in a county, which is known to be associated with higher CHD-IM. It has also been suggested that higher levels of education leads to better health literacy, allowing for improved understanding of complicated medical information around CHD and the selection of higher-quality pediatric cardiac centers.⁴ In addition, mothers that live in more highly educated counties may have more resources for learning information about CHD and the care of medically complex children.

Interestingly, the association of lower CHD-IM in counties with a higher concentration of higher income households and in counties with more educated mothers was only demonstrated in the subgroup analysis of non-Hispanic White infants, but not non-Hispanic Black infants. This suggests that non-Hispanic Black infants are not necessarily benefiting from the protective effects of living in counties with a higher concentration of high income households and/or more highly educated mothers. Further research is needed to understand why this might be.

There are several prior studies demonstrating non-Hispanic Black infants with CHD having worse outcomes.^5,9,32 Our study adds nuance to the understanding of the impact of race on CHD-IM. Race alone does not fully explain the disparities in CHD-IM; there are complicating factors at play including proximity to a T50-PCC. This may shed light on the issues of access to quality congenital heart care, with most large congenital cardiac centers residing in urban centers. When evaluating CHD-IM by geographic region and ICE-race, this study suggests that there are a portion of rural counties in the U.S. with a high concentration of non-Hispanic White individuals who experience high CHD-IM (Online supplement Figure 1). These counties may lack access to care in that they are not proximal to a T50-PCC. A prior study of neonates with hypoplastic left heart syndrome in Texas found an 18.6% increase in neonatal mortality for those ≥90 minutes from a cardiac surgical center compared to those <10 minutes away.³³ However, when proximity to a T50-PCC is adjusted for, the opposite was seen; there was increased CHD-IM in counties with a higher concentration of non-Hispanic Black mothers. This suggests that, for infants born to non-Hispanic Black mothers, there are likely other social determinants of health and access issues beyond proximity to a T50-PCC that impact CHD-IM. Future studies are needed to understand whether it is the proximity to a T50-PCC or the effect of living close to a large metropolitan area that mediates these findings. A previous study that evaluated geographic variation in CHD-IM found that proximity to a T50-PCC correlated with lower CHD-IM, but that proximity to a lower tier PCC did not.¹³ This suggests that living in proximity specifically to a T50-PCC may confer benefits in CHD-IM, rather than the general effect of living in a metropolitan area with access to hospital facilities.¹³ Another study found that Black children with CHD had a higher chance of being admitted to a high-mortality hospital for a cardiac procedure compared to other children with CHD.³ Racial inequities in geographic referral patterns for pediatric cardiac care may also exist, and require further study.

This study has several limitations. Several of the limitations are characteristic of vital statistics and census data, including an inability to analyze more granular data related to clinical characteristics (e.g., CHD complexity, genetic syndromes, comorbidities) and infant mortality, and incomplete ascertainment or misallocation of county birth or death. Race and ethnicity was based on birth certificates, which may not always be accurate. Death certificates may misassign cause of death to CHD rather than as a contributor to the cause of death. In addition, the study was unable to evaluate the incidence of CHD by county. This study focused primarily on disparities among non-Hispanic Black individuals, and in-depth analysis of other races/ethnicities was not included. For example, for counties where the majority of residents are Hispanic, the ICE-race score would be near zero. Previous studies have shown inequities in CHD-IM among those of Hispanic descent as well.^3,6 It will be important for future studies to evaluate health inequities across all races and ethnicities. While the ICE measure highlights inequities between counties, it has limitations in capturing social inequities within a county. ICE also does not account for mothers’ length of exposure to a neighborhood, and there may be other unobserved variables that influence CHD-IM outcomes that are not accounted for in this study. The American Community Survey dataset included average household income by county between 2013–2018, in 2018 inflation-adjusted dollars. We acknowledge that this average household income has the potential to change significantly over the 12-year time period of our study and that there are limitations to using non-equivalized income. Lastly, the US News and World Report ranking of T50-PCC has its own limitations.

Conclusions:

In conclusion, CHD-IM is associated with counties with the highest concentration of lower educated mothers and the highest concentration of low-income households. Mortality as a function of race is more complicated, and requires further investigation. Understanding maternal education, income and racial characteristics by county may help to inform local and national policy efforts to improve outcomes in infants with CHD. Our findings suggest that policy prescriptions should be sensitive to racial and geographic contexts. Future studies are needed to determine which interventions are effective in addressing the social inequities identified in this study.

Supplementary Material

3

Supplement Figure 1. Maps of ICE-income, ICE-education and ICE-race by quintile

Supplement Table 1. Multivariable Models: ICE-race (quadratic) with proximity to a Top 50 Pediatric Cardiac Center, ICE-income with proximity and race, ICE-education with proximity and race