Historical Neighborhood Redlining and Contemporary Cardiometabolic Risk

Cardiovascular disease (CVD) and mortality have been linked with a multitude of socioeconomic (ie, poverty, lower income, lower rates of insured population), environmental (ie, limited access to greenery, healthy foods, health care, proximity to highways, industries, and land-fills), and social (ie, systemic racism, racial segregation) factors. Recent studies have elucidated that racial segregation has been linked with various health outcomes throughout the United States.¹ Disparities in CVD often accompany disparities in neighborhoods and racial segregation and plague the most vulnerable communities.

In the 1930s, in an effort to stabilize the housing market during the Great Depression, the federal government created an entity known as the Home Owners’ Loan Corporation (HOLC). To ameliorate default of loan repayments, assess foreclosure risks, and to enforce racial/ethnic segregation, the HOLC created maps of nearly 200 U.S. cities ratings based on racial/ethnic composition, housing conditions, and neighborhood environments. These rated areas were color-coded based on potential lending risk as A (“best” or green), B (“still desirable” or blue), C (“definitely declining” or yellow), and D (“hazardous” or red), with the latter deemed as “redlined” neighborhoods.² Although these housing practices were outlawed in the 1960s, subtle discriminatory practices have continued to perpetuate and shape current social (eg, widening disparities in socioeconomic status and residential segregation) and built environmental structures over the past century, widening health inequities.³

Historic redlining has been linked with several modern-day health inequities across major urban cities, including asthma, certain types of cancer, preterm birth, mental health, and other chronic diseases.³ It remains unclear whether historical redlining influences inequities in cardiometabolic risk factors and diseases. Accordingly, we sought to examine the link between historical redlining and contemporary cardiometabolic health risk factors and outcomes in the United States.

METHODS

This is a cross-sectional study of the mean prevalence of the 2019 cardiometabolic outcomes and risk factors at the census tract level according to HOLC grading.

EXPOSURE DEFINITION AND ASSESSMENT.

We obtained original HOLC-graded data from the Mapping Inequality Project digitized maps.² As previously described, we generated census tract–level scores by weighting the percentage of areas that are spatially consistent with neighborhoods of HOLC-graded areas.⁴ First, using QGIS version 3.16, we calculated the percentage intersection between each HOLC-graded neighborhood boundary and the 2020 census tract boundaries. We then calculated the total graded percentage intersection for each census tract and excluded census tracts with <20% total area of intersection. We then multiplied the fractions of the graded area by their corresponding HOLC numeric scores (1–4 corresponding to A-D) and generated a score on the continuous scale, which was rounded and transformed back into 1 of 4 categories: A (1), B (2), C (3), and D (4). We defined “redlined” neighborhoods as D-graded census tracts, and “non-redlined” neighborhoods were defined as A through C-graded census tracts.

OUTCOME DEFINITION AND ASSESSMENT.

We used the 2021 Centers for Disease Control and Prevention (CDC) PLACES database, which includes variables from the 2019 Behavioral Risk Factor Surveillance System, the 2010 U.S. Census population, and the American Community Survey estimates. PLACES reports the prevalence estimates of census tract–level health indicators by using a multilevel regression and poststratification approach.⁵ We also linked census tract–level exposure of particulate matter <2.5 μm and diesel particulate matter from the Environmental Protection Agency’s 2021 environmental justice tool as potential environmental confounders.⁶

Our outcome variables included the following: markers of health care access (cholesterol screening in the past 5 years, routine health care checkup in the past year, and lack of health insurance in adults aged 18–64), cardiometabolic risk factors (diabetes, current smoking, obesity, hypertension, and high cholesterol), and cardiometabolic outcomes (prevalence of coronary artery disease [CAD], stroke, and chronic kidney disease [CKD]). HOLC-graded census tracts were then linked with the prevalence of cardiometabolic indicators, followed by calculating the average of each indicator across census tracts in each HOLC grade. Demographic composition analysis was based on the 2010 Census in which minority population represents all people other than single-race non-Hispanic White individuals. To assess for regional variations in the impacts of redlining, we compared the trends of outcomes and risk factors across the 4 census regions (Midwest, Northeast, South, and West).

STATISTICAL ANALYSES.

For all cities, we first used analysis of variance to determine whether there were statistically significant differences in census tract–level prevalence for cardiometabolic outcomes across HOLC grades. We then performed sequentially adjusted generalized linear regression models, first adjusting for median age, followed by adjusting for census tract–level percentage of minority population, prevalence of diabetes, high cholesterol, hypertension, smoking, obesity, particulate matter <2.5 μm, and diesel particulate matter. All analyses were conducted using R version 4.1.2 and graphs were generated using GraphPad Prism version 8.0.

RESULTS

A total of 11,178 HOLC-graded census tracts were included, comprising 38,537,798 inhabitants. A-graded areas covered 797 (7.1%) census tracts, B-graded areas covered 2,173 (19.4%) census tracts, C-graded areas 4,691 (42.0%), and D-graded areas covered 3,517 (31.5%) census tracts. The percentage of Black individuals (13.2%, 21.7%, 23.3%, and 32.2%) and Hispanic individuals (8.5%, 17.1%, 25.9%, and 28.8%) increased across HOLC grades (A, B, C, and D, respectively).

Neighborhoods with better HOLC grades had higher cholesterol screening (A to D: 88.4%−84.0%, P < 0.001) and routine health visits (A to D: 78.2%−76.0%, P < 0.001) compared with neighborhoods with worse HOLC grades. Further, the prevalence of individuals lacking insurance aged 18–64 years almost doubled from A through D-graded areas (10.5%−21.4%, P < 0.001). We also observed overall increase in the prevalence of diabetes (9.2%−13.5%, P < 0.001), obesity (28.5%−35.3%, P < 0.001), hypertension (30.0%−33.8%, P < 0.001), and smokers (13.1%−20.6%, P < 0.001) in stepwise increments across the HOLC grading spectrum from grade A to D. However, we observed a relative decrease in the prevalence of patients with high cholesterol (31.3%−29.2%, P < 0.001) (Figure 1A).

FIGURE 1. Mean Prevalence of Cardiometabolic Health Indicators Across HOLC Grades.

(A) Mean prevalence of cardiometabolic health outcomes according to HOLC grade. (B) Mean prevalence of cardiometabolic health risk factors according to HOLC grade. Error bars represent SDs. P values correspond to the variance of means for the prevalence of health indicators across HOLC grades. Columns are color-coded based on HOLC grading: A (“best” or green), B (“still desirable” or blue), C (“definitely declining” or yellow), and D (“hazardous” or red). BP = blood pressure; CAD = coronary artery disease; CKD = chronic kidney disease; HOLC = Home Owners’ Loan Corporation.

Across HOLC grades A through D, we found statistically significant increases in the prevalence of CAD (5.3%−6.2%), stroke (2.9%−4.2%), and CKD (2.7%−3.6%) (P < 0.001) (Figure 1B). In the unadjusted model, there was a statistically significant positive association between historically redlined census tracts and prevalence of CAD (β = 0.57, P < 0.001), stroke (β = 0.72, P < 0.001), and CKD (β = 0.52, P < 0.001), which remained significant after adjusting for median age: CAD (β = 0.82, P < 0.001), stroke (β = 0.82, P < 0.001), and CKD (β = 0.57, P < 0.001). In the fully adjusted model, the associations were weakened for CAD (β = 0.02, P = 0.063), stroke (β = 0.03, P < 0.001), and CKD (β = 0.04, P < 0.001).

There were few regional differences in association between redlining and cardiometabolic indicators. For example, Southern and Northeastern states had the lowest and highest mean relative difference of outcome prevalence between HOLC grade D and grade A, respectively (CAD: 1.8% vs 41.5%, stroke: 23.3% vs 86.2%, and CKD: 18.5% vs 59.3%). In addition, the difference in prevalence of high cholesterol between HOLC grades was not statistically significant (P > 0.05) in the Southern and Midwestern states, whereas that of high blood pressure was not statistically significant in the Northeastern states. The remainder of differences were consistent across census regions.

DISCUSSION

To our knowledge, this is the first study examining the national relationships between redlining neighborhoods and CVDs. In a recent study, Mujahid et al⁷ used data from the Multi-Ethnic Study of Atherosclerosis to explore the effects of redlining on the cardiovascular health scores of individuals from 6 cities (949 census tracts), and demonstrated that Black adults living in historically redlined areas received a lower composite cardiovascular health score than Black adults living in A-graded (best) neighborhoods. Our findings support and extend this inequality nationally, and show that redlining not only affects CAD, stroke, and CKD, but increases risk of associated comorbidities and lack of access to appropriate medical care.

In the analysis of individual risk factors and indicators of CVD, we observed a relationship between HOLC grade and almost all comorbidities (ie, hypertension, diabetes, obesity). The reasons for these associations are speculative, but may be related to psychosocial stress,⁸ lack of access to care, environmental exposures, and unhealthy behaviors (eg, smoking).³

In addition, we show that redlined neighborhoods had decreased routine health care visits and higher rates of lack of insurance. Generally, residents of redlined neighborhoods, especially minorities, have lower access to public transportation, health care insurance, and healthy food choices putting them at risk for missed prevention and adverse health outcomes.³

The association between redlining and the prevalence of cardiometabolic conditions was attenuated after adjusting for comorbidities, risk factors, demographic composition, and environmental exposures. This further illustrates that redlining may affect cardiovascular outcomes via traditional and nontraditional risk factors, with possible regional variations. Future studies should focus on incorporating region-specific social and environmental exposures in the predictive assessment of cardiometabolic disease risk.

Although the association between redlining and health outcomes is multifactorial, disparities in environmental exposures and in socioeconomic attributes may help explain the poor health outcomes in redlined neighborhoods. Historically redlined neighborhoods are situated next to major sources of environmental pollution, making inhabitants more likely to experience detrimental health effects resulting from disproportionately higher exposures to ambient air pollution, less green space, and environmental toxicants such as phthalate and phenols.³ Furthermore, disinvestments in these neighborhoods left many minorities and low-income households residing in inadequate indoor environments and exposed to toxicants.³ Financial strain, dismantled communities, and racial discrimination experienced by many residents of redlined neighborhoods may lead to increased stress and associated adverse health outcomes. Stress affects health in complex and diverse pathways that may include increased stress hormones and adopting unhealthy lifestyles.^3,7 It is also postulated that the health consequences of racial discrimination and segregation can be persistent and intergenerational via epigenetic changes as an embodiment of racial inequalities.⁹ The combined effect of transgenerational social vulnerability and health consequences entailed by racial segregation create an unfavorable milieu for minorities living in redlined neighborhoods.

STUDY LIMITATIONS.

This study was based on modeled prevalence of health outcomes developed by the CDC based on self-report, which may be prone to estimation mischaracterization. Despite these limitations, the CDC’s PLACES data have been widely used in epidemiological studies and provide a valuable knowledge for generating hypotheses to be tested in future studies using empirical data. We also recognize that unmeasured confounders, such as behavioral and genetic factors, may have introduced confounding bias. Furthermore, the exposure definition of “redlining” has not yet been standardized across studies assessing redlining, which may be prone to misclassification bias.

CONCLUSIONS

Historical housing discriminatory practices are associated with modern-day cardiometabolic disease and risk factors. Future studies should examine micro-level neighborhood characteristics, which make redlined neighborhoods more susceptible to disease.

ABBREVIATIONS AND ACRONYMS

CAD

coronary artery disease

CDC

Centers for Disease Control and Prevention

CKD

chronic kidney disease

CVD

cardiovascular disease

HOLC

Home Owners’ Loan Corporation