While much has been written about child health and developmental disparities associated with poverty and/or race/ethnicity, much remains unknown about the mechanisms through which these disparities occur or the role that racism plays as a social determinant of health [1,2,3,4,5,6]. The concept of race and, thus, racism in the United States emerged as the nation emerged [7]. The European settlers were dependent on enslaved Africans to maintain the economic well-being [7]. In order to maintain their power differential, they created legal categories of people, primarily Black and Native American, who were deemed to be inferior in many ways to those who were white [7]. Although now discredited, scientific theories and studies were developed on the premise that race was a biological construct that was fixed in nature at birth [7]. From these basic assumptions, systems of white supremacy and privilege emerged while non-White groups, especially those of African descent were oppressed [7]. The literature refers to different types of discrimination and racism, but structural racism is most often studied in relation to adult and child health outcomes [5,6,7,8,9,10]. Structural racism refers to inequalities in multiple overlapping societal systems such as housing, education, employment, distribution of resources, health care access and quality, criminal justice, salaries, etc. [5, 6, 8,9,10]. Neighborhood quality or residential segregation has been one area that has been shown to be associated with disparities in mortality and morbidity across multiple systems or disorders [9, 11,12,13,14].

Previous work has focused on maternal, child, and environmental factors at the individual level [2, 15,16,17,18,19,20,21,22,23,24]. Recent literature suggests that the effects of racism on health and developmental outcomes can best be understood by examining a broader range of influences including structural racism and discrimination (SRD) [5, 8, 25,26,27].

Neighborhood and environmental quality differences have a long history in the U.S. that were exacerbated by the practice of “redlining” [14, 28,29,30]. “Redlining” was a practice that was first documented in the U.S. around 1930 and continued through 1968 [30]. The practice originated to inform loan officers, appraisers, and real estate professionals of the risk of providing loans to individuals in certain geographic areas [30]. Maps were created by the Home Owners’ Loan Corporation (HOLC) in major cities throughout the country [30]. The HOLC sent workers out to rank the neighborhoods “based on criteria related to the age and condition of housing, transportation access, closeness to amenities such as parks or disamenities like polluting industries, the economic class and employment status of residents, and their ethnic and racial composition” [30]. Once the assessment was completed, neighborhoods were color-coded on maps such that green was the “most desirable;” blue was “still desirable;” yellow was “definitely declining;” and red was “hazardous.” Thus, the term “redlining” emerged [29, 30]. The significance of this practice is that individuals in these areas were unable to get mortgage or small business loans, which resulted in the areas remaining poor, underserved, and largely occupied by those of minority status [30]. [30] These practices represent a systemic level of structural racism creating segregated and impoverished communities by restricting loans to homeowners and businesses based on the HOLC grading classifications of neighborhoods [14, 28, 30, 31]. Currently, many of the “redlined” areas remain economically and racially segregated [30, 32]. In a recent study including cities for which HOLC maps were available, Louisville, Kentucky was one of five cities that remain highly segregated and still have the lowest increase in income and housing values among previously “redlined” areas [30].

To date, two studies specifically mapped birth outcomes to areas of historic redlining [14, 31]. Two others used a similar construct of structural racism using the Index of Concentration at the Extremes (ICE) [33, 34]. Bishop-Royse et al. found that infant mortality rates in Chicago were related to structural racism as measured by the ICE after controlling for socio-economic marginalization, hardship, family support, and healthcare access [33]. Similarly, Chambers et al. found preterm birth and infant mortality rates in California were related to extreme income, race, and income + race concentrations and that Black women were more likely to live in zip codes with greater extreme income concentrations and moderate extreme race + income concentrations [34].

Health disparities have been associated with race and ethnicity [35,36,37,38,39,40,41,42,43,44], but little is known about the effects of racism [3, 45]. Williams and Mohammed [4] warn that more research is needed to ensure that inaccurate conclusions are not drawn for the relationship between race and health disparities. Conclusions from some studies that present racial/ethnic differences in outcomes across multiple domains have led to the misconception that the disparities have a biological basis due to race. However, it is important to note that race is a social construct. Research is needed to identify the root causes for these disparities [10, 11, 46, 47].

The Child Opportunity Index (COI) [48, 49] is another measure of neighborhood quality, which uses 29 indicators from census tract data to measure neighborhood resources and conditions relevant to children’s health and development [49]. Indicators are ranked from very low opportunity to very high opportunity (5 levels) for 3 subscales (education, health environment, and social and economic) as well as an overall score. Family home addresses are used to link to locally or nationally normed COI datasets. Based on previous literature, we hypothesized that low levels of neighborhood child opportunity would be associated with higher levels of risk for adverse child health outcomes.

Recent literature demonstrates the need to elucidate the effects of racism and its multiple levels on specific health outcomes [4, 9, 45]. Two recent papers found that present day risk for preterm birth was associated with “redlining” in New York City [14] even after controlling for maternal covariates such as age, educational attainment, and measures of current poverty, and was associated with multiple adverse birth outcomes in California [31]. More data are needed to examine the effects of environmental adversity on pregnant and parenting women as well as child outcomes beyond birth. The COI captures factors related to neighborhood disadvantage that are often related to racist practices and policies. Previous studies have found racial differences in the rates of and treatment for mental health disorders in adults and children [50,51,52] and developmental disabilities [35, 50, 53,54,55,56]. The purpose of the current study is to describe rates of mental health and developmental disability diagnoses in children by COI levels and race/ethnicity and to examine their associations after controlling for age and sex. As described above, data are not available that examine the effects of neighborhood disadvantage on children beyond infancy. Knowing that neighborhood quality is rooted in structural racism practices, we hypothesize that COI levels will be associated with increased mental health and developmental disability diagnoses in children.

Methods

Study Design and Data Source

In this cross-sectional analysis, we retrieved the sample cohort from the electronic health record for outpatient visits, which included emergency department and primary care visits, in 2022 from a large children’s health system in an urban setting (n = 181,887). Unique children < 18 years of age with valid addresses living in the KY-IN Louisville Metropolitan Statistical Area (CBSA 31140) were included (n = 115,738).

The cohort (n = 115,738) was predominantly non-Hispanic White (57.3%) followed by non-Hispanic Black (25.2%) with the remainder being Hispanic (9.3%), Other race/ethnicity (4.7%), or unknown (3.7%). The local population demographic is 66% non-Hispanic White, 22% non-Hispanic Black, 6% Hispanic, and 6% other races/ethnicities. Almost 18,000 children (15.5%) had a MH diagnosis and 7,905 (6.8%) had a DD (Table 1).

Table 1 Sample demographic data: Patients residing in the Louisville Metropolitan Statistical Area
Full size table

The primary independent variable was the most recent version of the overall metro-normed Child Opportunity Index 2.0 (COI 2.0) levels (from “very low” to “very high”) updated in 2015. The patients’ home addresses were geocoded and linked by census tract in the KY-IN Louisville Metropolitan Statistical Area (USA). Our outcomes were mental health diagnoses and developmental disorder diagnoses (See Online Resource Table S1 for diagnosis codes.). Diagnoses were retrieved from encounter diagnoses, past medical histories, problem lists, professional charge transaction diagnoses, and final coded diagnoses. If the child had multiple visits or multiple diagnoses, they were only included once for each category. Developmental diagnoses and substance related disorders were excluded from mental health diagnoses and mental health diagnoses and substance related disorders were excluded from developmental disability diagnoses. Children under 2 years of age were excluded from the mental health disorder diagnosis group but not from the developmental disability diagnosis group. Substance use disorders are, generally, reported in the literature separate from other mental health disorders although there is significant co-morbidity [57,58,59]. Additionally, substance use disorders are tracked for children 12–17 years of age only in national databases [60]. Race and ethnicity were extracted from the electronic medical record, which, at our institution, is not standardized. The variable combines race and ethnicity and allows only one option. Multivariable logistic regression models were performed to examine the association between the specific diagnosis groups and COI levels, controlling for sex and age.

The sample was first stratified by race, and a univariate analysis was conducted among each race group to determine whether the binary outcome (mental health: yes/no; developmental disorder: yes/no) was associated with the predictor (COI levels), age (continuous variable in years), and sex (male/female). In order to identify potential confounding variables or effect modifiers, the Wald Chi-Square p-value of univariate logistic regression was employed using the SAS's PROC LOGISTIC procedure. The independent variable was initially assessed to determine whether it was an effect modifier with the interaction term achieving bivariate significance at p < 0.05. If there was no significant interaction term, we considered it a confounder and, thus, we included age and sex in the multivariable model. Patients with missing data were not included in the logistic regression analysis but were included in Table 1. Missing data was minimal and missing at random.

The study was approved by the institutional review board at the University of Louisville. The study was also approved by the Research Office at the Norton Children’s Hospital. A data-sharing agreement is in place through an affiliation agreement between the two institutions. Data were stored and analyzed on a secure HIPAA-compliant network.

Results

Non-Hispanic White children make up 28.9% of the children living in the Very Low Opportunity level neighborhoods compared to 77.5% living in the Very High Opportunity level neighborhoods (Table 1). In contrast, 53.8% of the non-Hispanic (N–H) Black children live in Very Low Opportunity level neighborhoods compared to 7.6% living in the Very High Opportunity level neighborhoods.

Figure 1 depicts the overall prevalence rates of MH diagnoses & DD diagnoses by COI levels. In Fig. 2, the prevalence rates were displayed for N–H Black and N–H White children as they constituted the two largest groups in our sample. MH diagnosis was lower for N–H Black children than for N–H White children in each COI level. DD were diagnosed slightly less for N–H White children than N–H Black children.

Fig. 1
figure 1

Prevalence rate (%) of MHD & DD diagnosis by COI levels

Full size image
Fig. 2
figure 2

Prevalence rate (%) of MH diagnosis & DD diagnosis by COI levels: Comparison of Non-Hispanic Black and White Children. MH—mental health; DD = developmental disability

Full size image

In multivariable logistic regression, COI was associated with MH diagnosis for N–H White children only (Table 2) with an adjusted odds ratio (95% CI) of receiving a MH diagnosis for children living in a Very Low compared to a Very High COI area of 1.74 (1.62, 1.87; Table 2). Similarly, for N–H White children only, the adjusted odds ratio (95% CI) of receiving a DD diagnosis for children living in a Very Low compared to a Very High COI area was 1.69 (1.51, 1.88; Table 3). There were no significant associations between diagnoses and COI for children who were N–H Black, Hispanic, or Other race/ethnicity for MH or DD diagnoses.

Table 2 Multivariable logistic regression models examine the association between mental health disorders and COI levels, adjusting for sex and age by racial groups
Full size table
Table 3 Multivariable logistic regression models examine the association between developmental disorders and COI levels, adjusting for sex and age by racial groups
Full size table

Across all races/ethnicities except “other,” males had a higher adjusted odds ratio of receiving a mental health diagnosis and older children across all racial/ethnic groups had a higher adjusted odds ratio of receiving a mental health diagnosis. Similarly, males had a higher odds ratio of receiving a developmental disability diagnosis across all racial/ethnic groups and younger children were more likely to get the diagnosis. For both diagnostic categories, Hispanic male children had the highest adjusted odds ratio of receiving a diagnosis.

Discussion

The current findings present new information on the relationships among COI, race/ethnicity, and mental health and developmental disability diagnoses in an urban area in a southern state in the U.S. Contrary to our expectations, neighborhood disadvantage did not explain disparities in diagnosis for non-White children. More research is needed to determine the complex array of factors beyond neighborhood disadvantage that are operating to create differences for children of color. Our findings suggest that there are factors that impact the prevalence of or the diagnosis of mental health and developmental disorders across all levels of neighborhood quality.

As expected from previous literature, we also found that Black children are disproportionately represented in the very low opportunities neighborhoods compared to non-Hispanic White children (54% vs 29%, respectively). Likewise, White children make up 78% of the children living in very high opportunity neighborhoods, while only 8% are Black, 4% are Hispanic, 6% are listed as other, and 5% are of unknown race/ethnicity.

The effects of toxic stress on parents and children have been studied for more than a decade [60,61,62,64]. More recently, researchers are looking at chronic and cumulative stress in relation to health equity and differences in child development [64,65,66,67,69]. Neighborhood disadvantage, which is related to poverty, is only one stressor families may face. Since poverty and neighborhood disadvantage are more often associated with being of minority status, more attention is needed on the effects of racism-related cumulative stressors [46, 68,69,71].

One explanation for increased rates of developmental disabilities in Black children may be prenatal exposure to cumulative stressors, especially stressors caused by structural racism and discrimination (SRD) at multiple levels (individual, interpersonal, community, and societal) [26]. Similarly, exposure of both caregivers and children to racism-related cumulative stress may increase the prevalence of mental health disorders in children [71]. Research has shown that stress is associated with adverse outcomes for adults and children [71,72,74]. However, less is known about stress that is specifically caused by racism and discrimination [75]. Conversely, lower rates of mental health diagnoses in minoritized groups may result from caregivers’ lack of help-seeking behavior due to fear of stigmatization, lack of access to care, or provider bias in diagnosing [75,76,77,78,79,80,82].

More research is needed to identify root causes such as provider bias, caregivers’ medical mistrust, and other potential forms of interpersonal and structural racism. Additionally, the identification of protective factors may inform interventions. It is not enough to identify an issue, but the information must be used to inform interventions to improve health equity. Some protective factors that have previously been studied include racial socialization [82,83,84,86], racial identity [83, 86,87,89], parental social support [90], neighborhood cohesion [89,90,91,93], parenting self-efficacy [90, 94, 95], and spirituality/religiosity [96,97,98]. Developing interventions to promote these protective factors has potential to, at least partially, mitigate the effects of SRD. Additionally, identifying ways to minimize provider bias and enhance parent and child trust in the health care system has potential to reduce inequities. Strengthening community and school resources has the potential to increase access and minimize stigma [99, 100]. Health information exchanges could be strengthened by expanding the system to include school and community-based diagnosis and treatment [101], which could be used to provide case management and service utilization. Expanded public and private insurance coverage for family-centered mental health promotion interventions is needed [102, 103].

Although the study contributes new information to the literature, there are limitations. First, the collection of race and ethnicity within the electronic medical record is not ideal. The methods are not standardized, are not consistently parent-reported, and include a combined race/ethnicity variable with no option for selecting more than one group. Second, the COI does not include access to health care, which could also influence the receipt of a diagnosis. Third, data were collected from the EHR only, which would exclude diagnoses that may have been made at community mental health agencies or through the school system which may not be accessed equally among children from all levels of COI. However, the use of the problem list in addition to the visit diagnosis may have captured this. Fourth, the data for these analyses came from a single healthcare system, therefore, generalizability to other locations may be limited. Lastly, COI measures neighborhood quality from census data, which does not capture individual-level factors. Relatedly, measures of multi-level exposures to SRD are emergent [102,103,106]. More research is needed to examine the effects of the complex array of factors that may be associated with differences in the prevalence and/or diagnosis of mental health and developmental disabilities in childhood. The cumulative and intergenerational effects of SRD on health disparities have yet to be fully captured [106].

Despite the limitations, the study highlights important findings. If neighborhood differences do not explain racial/ethnic differences in diagnosis, what are the root causes? The effects of racism-related cumulative stressors on parents and children seem like a promising direction. Additionally, the study of protective factors to support positive outcomes in the face of such stressors is critical. Lastly, identifying programs, policies, and interventions to reduce childhood poverty and link children and families to affordable, quality community mental and physical health resources is needed. These programs must be developed with input from the communities they serve to ensure that care is provided in such a way that the families can build trusting relationships with the providers and that stigma toward the parents and the children can be avoided.

Summary

Neighborhood disadvantage has its roots in racist policies and practices and has been identified as a potential source of disparities in adult health and pregnancy outcomes. The current study aimed to identify patterns of childhood diagnosis of mental health and developmental disabilities by race/ethnicity and neighborhood characteristics as measured by the Child Opportunity Index (COI). COI was related to the rate of diagnosis for non-Hispanic White children only. Based on the findings, it appears that non-White children across COI levels may have other factors contributing to racial/ethnic differences in diagnosis rates. More research is needed to better understand the complex array of factors responsible for the disparate outcomes such as racism-related cumulative stressors.