Mediating Pathways between Neighborhood Structural Investment and Cardiometabolic Health Across U.S. Cities

Abstract

Background:

Epidemiologic studies have linked neighborhood socioeconomic conditions to health. However, few have examined neighborhood structural investment (NSI) influences on cardiometabolic risk markers across urban environments. This study investigated whether NSI varies by historic redlining, associations between NSI and the prevalence of obesity, diabetes, and coronary heart disease (CHD) and whether redlining’s effect on obesity, diabetes, and CHD prevalence are mediated by neighborhood structural investment.

Methods:

NSI was measured using a composite score based on census tract data from MapUSA, which included home value, rent, vacant houses, and structures older than 30 years (higher scores representing greater investment). Obesity, diabetes, and CHD prevalence estimates were from the 2024 CDC’s 500 Cities 2024 data. Redlining data from the Home Owners’ Loan Corporation (HOLC) scores from the University of Richmond’s Mapping Inequality Project were analyzed for seventeen U.S. cities. NSI was tested as a mediator of these associations. Models were adjusted for % Black, % of people 60 and older, % of families in poverty, % of people with a college degree, % unemployed, median household income, and length of residency.

Results:

Living in a formerly redlined neighborhood was associated with lower NSI longitudinally. NSI was associated with decreased obesity, diabetes, and CHD prevalence, but these associations varied by city. NSI mediated associations between redlining and health outcomes varying by city.

Conclusions:

This study suggests that historic redlining is associated with contemporary health outcomes via neighborhood structural investment. Such findings could be used to inform cardiometabolic health intervention designs.

Introduction

Obesity and Type II diabetes are complex diseases with multi-faceted etiologies that directly contribute to coronary heart disease (CHD) and other cardiovascular diseases (CVD).¹ Social determinants of health encompass environmental, socioeconomic, and psychosocial factors that influence health through limited access to healthcare and exposure to environmental stressors.^2-4 Studies have reported associations between the neighborhood built and social environment and CVD risk factors, including obesity and diabetes, and subsequent CVD events.⁵ Historically, racial residential segregation in the U.S. has shaped neighborhood environments through reinforcing disinvestment, institutional discrimination, and the strategic marketing of unhealthy substances (e.g., tobacco).⁶

Increasing attention has focused on neighborhood investment in historically disadvantaged neighborhoods to improve conditions and overall well-being.^7-9 Longitudinal studies have linked neighborhood investment to more favorable perceptions of access to healthy food,⁷ increases in food security,⁸ improvements in walkability,⁹ increased safety,⁸ decreased crime,¹⁰ and improved neighborhood satisfaction.⁸ Epidemiologic research supports links between neighborhood investment and health,^8,9 using various measures such as public and private investments,¹¹ measures of physical disorder via virtual neighborhood auditing,¹² neighborhood deprivation,¹³ vacancy rates, and the neighborhood inventory for environmental typology.¹⁴ These studies found that lower investment was associated with worse breast cancer outcomes¹² while more investment was positively associated with sleep.¹¹ The different conceptualizations of neighborhood investment pose challenges for comparing studies and evaluating longitudinal trends. Comparing neighborhood investment to traditional socioeconomic measures presents a unique opportunity to examine the joint effects of housing market dynamics and infrastructure on cardiometabolic health.

As an investment marker, the Home Owners’ Loan Corporation (HOLC) score or “redlining” remains linked to contemporary health measures. Prior studies show associations between redlining and a variety of outcomes, including cardiovascular events,^15,16 cancer,^17,18 preterm birth,¹⁹ childhood obesity,²⁰ mortality,²¹ asthma,²² and COVID-19.²³ HOLC scores have also been linked to unhealthy food environments,^24,25 reduced access to health-promoting resources,^26-28 and increased firearm assaults.^29-31

The Home Owners’ Loan Corporation was created in 1933 by the Home Owners’ Loan Act of 1933, which was created in response to the 1929 stock market crash, as an emergency relief program to protect mortgages that were close to or in foreclosure.³² Between 1935 and 1940, it produced residential security maps for 239 U.S. cities, grading neighborhoods from A (greenlined or the best neighborhoods) to D (redlined or the worst neighborhoods).^22,33 Redlined areas included high proportions of Black or Jewish populations and specific eastern/southern European immigrants.

Although nearly a century has passed, redlining’s legacy continues to shape neighborhood characteristics. Mechanisms linking HOLC scores to contemporary neighborhood characteristics include urban renewal programs, eminent domain/highway construction, and subprime lending in neighborhoods with a higher proportion of non-white groups.³⁴ The exodus of middle-class Black families to the suburbs, coupled with disinvestment in non-white urban communities, exacerbated neighborhood and material conditions for non-white populations in urban areas.^35-41 Existing research has noted HOLC scores are associated with higher poverty, vacancy, and mortgage defaults, and lower homeownership rates and home values.^42-46 A study of 14 U.S. cities found that non-redlined neighborhoods had higher educational attainment and lower rates of poverty and unemployment.⁴⁷

Despite these findings, the role of neighborhood investment as a mediating pathway between HOLC scores and health remains not well understood. Prior studies often rely on national samples, limiting the understanding of the impact of geographic variations and local policies.⁴⁸ More recently, data from the CDC’s 500 Cities project showed that in 13 of 14 cities, “Grade A” neighborhoods had better health outcomes (including the prevalence of coronary heart disease, smoking, diabetes, no leisure time physical activity, obesity, and routine checkups) than “Grade D” or redlined areas.⁴⁷ However, this study did not examine how neighborhood investment might mediate these associations.

The aims of this study seek to examine: 1) whether historically redlined neighborhoods predict differences in neighborhood investment longitudinally; 2) the link between neighborhood investment and cardiometabolic health across U.S. cities; and 3) whether neighborhood investment mediates associations between redlining and contemporary health indicators. Neighborhood investment was hypothesized to be negatively associated with obesity, diabetes, and CHD, with city-specific variations. Redlined neighborhoods were hypothesized to receive lower investments over time, and neighborhood investment to partially mediate the associations between redlining and adverse cardiometabolic outcomes.

Methods

Study Sample

This study collated publicly available data for specific geographic locations measured at the census tract level in the U.S. Sixteen cities across the U.S. were selected for this analysis, including Baltimore, Maryland; Boston, Massachusetts; Chicago, Illinois; Cleveland, Ohio; Detroit, Michigan; Indianapolis, Indiana; Kansas City, Missouri; Los Angeles, California; Milwaukee, Wisconsin, Minneapolis, Minnesota, New Orleans, Louisiana, New York, New York, Philadelphia, Pennsylvania, Pittsburgh, Pennsylvania, San Diego, California, and San Francisco, California. These cities were selected from each region because they had an adequate number of HOLC-graded census tracts (N > 100).

Measures

Using data from the MapUSA project, longitudinal NSI was created for all US census tracts for 1970, 1980, 1990, 2000, 2008-2012, and 2015-2019 separately. Specifically, median home value, median rent, percent vacant housing units, and percent of structures more than 30 years old were used to construct this variable. Median home value and median rent values were log-transformed. Percent vacant housing units and the percent of structures over 30 years old were subtracted from 100 and then z-standardized. The log-transformed (median home value and median rent) and z-standardized (% vacant housing units and % structures more than 30 years old) variables were added together to create a composite yearly score. To correspond with the health measures, investment scores based on estimates from the American Community Survey 2015-2019 were used where higher scores indicate higher neighborhood structural investment.

HOLC score data were downloaded from the Inter-University Consortium for Political and Social Research website through the University of Michigan based on calculations from the University of Richmond’s Mapping Inequality Project.^49,50 HOLC score was assigned to census tracts by overlaying the HOLC maps over census tracts and multiplying a weighting factor based on the amount of a census tract that fell within a particular HOLC category. Ultimately, the most desirable neighborhoods, with those shaded in green, had higher proportions of White populations in them, while the least desirable neighborhoods, shaded in red, had higher proportions of non-White populations. To gain adequate statistical power, green-, blue-, and yellowlined areas were compared in aggregate to redlined neighborhoods.

To evaluate cardiovascular health indicators, data on obesity, diabetes, and coronary heart disease prevalence from the 2024 Centers for Disease Control and Prevention 500 Cities database measured on the census tract level using 2022 Behavioral Risk Factor Surveillance System data were used. Obesity was defined as respondents 18 and older who have a body mass index greater than or equal to 30 kg/m² calculated from self-reported weight and height. Diagnosed diabetes was defined as the probability among adults who report being told by a doctor or other health professional that they have diabetes (non-gestational). Coronary heart disease was defined as respondents 18 and older who reported ever being told by a doctor, nurse, or other health professional that they had angina or coronary heart disease.

Covariates included % Black populations, % aged 60 and older, median household income, % unemployed, the % with a 4-year college degree or more, and housing tenure (the year a householder moved into unit), all measured based on American Community Survey estimates.

Statistical Analysis

Descriptive data for each city are presented as percentages, medians, and means. Multivariable linear regressions were used to examine associations between NSI and health outcomes, adjusting for % Black population, % aged 60 and older, median household income, % unemployed, % college degree or more, and housing tenure as covariates. Root Mean Square Error (RMSE) was used to assess the predictive ability of the regression models and compared with each decade of NSI and each health outcome. The RMSE allows us to find the decade of NSI that has the best performance and reduced prediction error.⁵¹ The differences in NSI using linear regression models with time since 1970, redlining status, and their interactions to predict NSI were also tested. The interaction term measures the difference in NSI slopes longitudinally. Structural equation models were used to examine the mediating effect of neighborhood structural investment between each city's independent and dependent variables, adjusting for covariates (Appendix Figure 1). All statistical analyses were performed using StataSE 17. This study did not require IRB approval as it utilized publicly available, de-identified data and did not involve human subjects.

Results

Among the sixteen cities, 1,938 census tracts were formerly redlined, and 3.191 were not redlined, consisting of green, blue, and yellow HOLC census tracts (Table 1). Detroit exhibited the highest obesity and diabetes prevalences, alongside having the highest concentration of non-Hispanic Black residents, as well as the highest unemployment rate. Detroit also had the lowest levels of neighborhood structural investment and the smallest percentage of individuals with a college degree compared to the other cities. Cleveland had the highest prevalence of coronary heart disease (CHD), coupled with the lowest median household income. Cleveland had the highest proportion of people who were 60 and older, while Milwaukee had the lowest proportion of people who were 60 years and older. San Francisco had the lowest prevalence of obesity, while Minneapolis had the lowest diabetes and CHD prevalence. San Francisco also had the highest levels of neighborhood structural investment, median household income, and proportion of individuals holding a college degree. Furthermore, San Francisco had the lowest concentration of non-Hispanic Black people and the lowest unemployment rate.

Table 1.

Sociodemographic and health characteristics by U.S. City: Data from the 500 Cities Project and American Community Survey

	Obesity Prevalence % (SE) min-max	Diabetes Prevalence % (SE) min-max	Coronary Heart Disease (CHD) Prevalence % (SE) min-max	Neighborhood Structural Investment	Aged 60+ % (SE) min-max	Non-Hispanic Black % (SE) min-max	Unemployed, % (SE) min-max	College Degree or Higher %, (SE) min-max	Median Household Income ($)	Non-Redlined Census Tracts* N (%)	Redlined Census Tracts N (%)	% of census tracts of total census tracts with available data
Baltimore	36.01 (8.16) 18-51	14.19 (5.28) 3.2-23.7	6.37 (1.88) 1.7-10.8	18.47 (2.61) 10.58-22.49	19.88 (6.98) 5.05-39.49	63.78 (33.32) 0.51-99.32	9.39 (6.47) 0-27.62	30.60 (23.96) 2.17-87.95	54,071.96 (29,443.60) 13,289-165,443	110 (67.48)	53 (32.52)	82%
Boston	25.08 (6.99) 15.20 – 39.60	9.29 (4.26) 2.10 – 22.30	4.98 (1.65) 1.00 – 10.20	20.77 (0.78) 18.92 – 22.97	16.31 (6.12) 1.18 – 42.61	23.49 (24.75) 0.00 – 82.93	6.62 (4.61) 0.00 – 25.54	49.34 (25.14) 6.36 – 95.55	74,850.66 (34,810.14) 16,226 – 163,938	83 (57.24)	62 (42.76)	84%
Chicago	32.14 (8.83) 17.00 – 48.90	11.92 (5.20) 2.30 – 29.50	5.49 (1.99) 1.20 – 15.50	18.70 (2.57) 8.60 – 23.74	17.74 (7.57) 2.65 – 53.41	35.46 (39.47) 0.00 – 100.00	9.87 (8.08) 0.00 – 44.06	36.44 (25.96) 0.95 – 94.18	60,030.55 (33,820.15) 11,146 – 188,697	445 (61.46%)	279 (38.54%)	91%
Cleveland	43.27 (7.01) 24.50 – 54.20	18.12 (5.57) 6.50 – 35.30	8.81 (2.16) 3.70 – 18.50	16.07 (2.21) 10.47 – 20.91	21.61 (8.25) 2.24 – 55.95	54.62 (35.81) 0.77 – 99.81	14.74 (7.82) 0.66 – 39.43	15.46 (12.61) 1.10 – 66.21	30,918.70 (12,182.83) 4,129-82,112	81 (50.00)	81 (50)	93%
Detroit	45.22 (4.14) 26.80 – 54.50	18.29 (3.40) 5.20 – 27.40	8.76 (1.76) 2.70 – 14.30	15.57 (2.40) 9.76-21.36	20.33 (7.81) 3.38 – 54.92	79.96 (26.10) 0.41 – 100.00	16.73 (8.85) 0-44.10	13.86 (13.44) 1.00 – 71.58	30,960.14 (13,325.26) 10,066 – 122,344	159 (62.60)	95 (37.40)	87%
Indianapolis	37.71 (6.95) 22.10 – 52.80	12.59 (4.31) 3.80 – 23.40	6.94 (1.76) 2.20 – 10.70	17.50 (2.58) 12.08 – 21.87	18.05 (6.43) 4.95 – 43.38	31.94 (26.21) 0.37 – 89.09	7.99 (5.86) 0.84 – 27.19	27.11 (20.82) 2.00 – 79.04	45,273.64 (21,089.74) 18,713 – 147,039	85 (62.96)	50 (37.04)	63%
Kansas City	40.58 (8.10) 22.70 – 52.40	15.02 (5.58) 4.60 – 26.60	7.12 (2.08) 2.50 – 12.50	16.64 (2.59) 11.77-22.78	17.69 (6.15) 5.01 – 31.33	38.32 (31.96) 0-92.63	6.67 (4.81) 0.00 – 20.10	23.69 (23.42) 1.76 – 82.08	41,347.45 (20,479.37) 12,321 – 127,250	34 (33.01%)	69 (66.99%)	48%
Los Angeles	27.54 (6.66) 14.60 – 41.40	11.41 (3.50) 2.90 – 23.80	5.41 (1.21) 1.40 – 11.30	20.05 (1.51) 10.45-23.44	17.10 (6.53) 1.55 – 54.74	9.72 (13.12) 0.00 – 78.06	6.52 (3.17) 0.00 – 18.08	33.65 (23.16) 1.01 – 84.72	63,189.86 (33,149.02) 16,586 – 250,001	383 (66.49)	193 (33.51)	58%
Milwaukee	39.33 (8.70) 20.80 – 54.40	12.24 (4.60) 3.00 – 21.70	5.17 (1.33) 1.80 – 8.80	17.37 (2.55) 11.02 – 21.46	14.85 (5.44) 1.21 – 41.61	40.60 (36.45) 0.00 – 100.00	7.80 (5.32) 0.47 – 28.60	23.95 (20.16) 0.51 – 81.88	41,422.37 (17,710.13) 7,917 – 113,375	95 (58.28%)	68 (41.72%)	77%
Minneapolis	26.74 (5.72) 18.30 – 41.80	7.48 (2.81) 2.20 – 15.80	4.30 (1.28) 1.20 – 10.00	18.72 (2.35) 10.09 – 22.80	15.49 (7.26) 1.78 – 37.40	19.92 (17.89) 0.20 – 71.41	5.42 (3.63) 0.00 – 16.30	47.02 (20.38) 7.44 – 84.23	65,509.95 (30,612.29) 18,750 – 169,063	86 (76.79)	26 (23.21)	97%
New Orleans	34.24 (9.00) 18.70 – 50.80	13.08 (5.75) 2.80 – 28.40	7.70 (2.50) 2.30 – 16.00	18.85 (2.23) 12.29 – 23.64	21.67 (6.93) 4.36 – 52.34	47.82 (32.02) 0.03 – 98.53	8.42 (6.16) 0.00 – 32.20	42.47 (23.82) 0.00 – 88.49	49,957.39 (30,594.65) 11,435 – 165,089	44 (37.29)	74 (62.71)	69%
New York	26.18 (6.81) 11.80 – 43.60	10.92 (3.09) 2.40 – 20.40	5.44 (1.22) 1.20 – 11.40	20.52 (1.92) 10.75 – 25.32	20.43 (7.21) 2.18 – 61.50	22.96 (28.71) 0.00 – 98.86	6.13 (3.73) 0.00 – 28.24	36.06 (19.86) 3.46 – 96.99	71,646.94 (31,596.31) 16,400 – 250,001	1,179 (66.69)	589 (33.31)	84%
Philadelphia	34.18 (8.09) 18.50 – 47.90	13.38 (5.00) 3.00 – 26.60	6.57 (1.87) 1.40 – 14.00	18.54 (2.40) 12.25 – 24.14	19.24 (7.76) 1.29 – 52.61	45.80 (34.30) 0.02 – 98.52	10.05 (6.16) 0.00 – 32.65	29.35 (22.98) 1.55 – 96.36	47,541.18 (25,163.04) 10,793 – 143,646	181 (55.69)	144 (44.31)	86%
Pittsburgh	33.08 (8.32) 18.8-53	11.70 (5.46) 1.7-26.8	6.77 (2.21) 0.9-14	17.80 (2.44) 10.29-22.88	21.24 (7.29) 0.19-38.91	27.79 (28.32) 0-97.53	6.75 (5.89) 0-33.62	40.66 (23.94) 0-92.30	50,392.58 (25,294.02) 12,269-169,375	76 (64.41)	42 (35.59)	93%
San Diego	24.91 (5.42) 13.9-36.3	9.89 (3.46) 3.7-16.6	4.97 (1.28) 2.2-8.3	20.18 (1.16) 16.56-22.95	17.57 (8.63) 4.73-47.28	8.64 (8.20) 0-37.01	6.32 (3.50) 0.67-16.17	38.08 (23.79) 2.82-82.97	70,496.68 (33,099.07) 23 28,462-229,583	66 (57.89)	48 (42.11)	40%
San Francisco	18.09 (3.56) 13.4-39	8.26 (2.68) 3.6-15.8	4.33 (0.99) 2.3-8.4	21.91 (1.52) 17.39-24.84	21.38 (5.99) 7.50-43.79	4.94 (7.85) 0-51.33	3.92 (1.98) 0-11.65	59.83 (20.47) 14.13-91.75	126,526.2 (38,318.16) 24,450-204,909	84 (56.38)	65 (43.62)	76%
Pooled Sample	30.54 (9.49) 11.80-54.50	11.98 (4.70) 2.10-35.30	5.88 (1.92) 1-18.50	19.28 (2.62) 8.60-25.32	19.12 (7.37) 1.18-61.50	30.69 (34.08) 0-100	7.97 (6.16) 0-44.10	34.25 (23.31) 0-96.99	61,959.48 (34,421) 4,129-250,001	3,1919 (62.21)	1,938 (37.79)	77%

SE: standard error; min: minimum; max: maximum

^*Non redlined census tracts consist of those which were green-, blue-, or yellowlined

-Obesity, Diabetes, and Coronary Heart Disease Prevalence were downloaded from the 2024 Centers for Disease Control and Prevention’s 500 Cities Database

-Neighborhood Structural Investment was measured using a composite score of census tract level data on median home value, median rent, % of vacant housing units, and % of structures more than 30 years old. Higher scores = more investment.

Line graphs of trends for each city show a positive trend for neighborhood structural investments for red and non-redlined neighborhoods (Figure 1). The interaction between redlining, NSI, and time shows that while NSI increased longitudinally for each city in the sample, historically redlined neighborhoods received less investment longitudinally compared to non-redlined neighborhoods (Appendix Table 1). The distribution of HOLC and NSI scores by city can be found in Appendix Figures 2-17.

Figure 1.

Longitudinal Trends in Neighborhood Structural Investment by City

Models with NSI in 2019, along with socio-demographic covariates, had the lowest RMSE for each health outcome. Therefore, models of health outcome prevalences that included the 2019 NSI scores, along with socio-demographic covariates, were reported. NSI was associated with a decreased obesity prevalence in Baltimore, Chicago, Cleveland, Detroit, Indianapolis, Kansas City, Milwaukee, New Orleans, New York City, Philadelphia, and San Francisco (Table 2). Neighborhood structural investment was also associated with a decreased diabetes prevalence in Baltimore, Chicago, Cleveland, Detroit, Indianapolis, Kansas City, Milwaukee, and Philadelphia, but positively associated with diabetes prevalence in Pittsburgh (Table 2). Increased neighborhood structural investment was significantly associated with a reduced prevalence of CHD in Baltimore, Chicago, Indianapolis, Kansas City, and Philadelphia (Table 2).

Table 2.

Regression Results of the Effects of Neighborhood Structural Investment on Obesity, Diabetes, and Coronary Heart Disease by U.S. City: Data from the 500 Cities Project and American Community Survey

City	Obesity			Diabetes			Coronary Heart Disease
	β	SE	p	β	SE	p	β	SE	p
Baltimore
Investment	−0.54	0.12	0.00	−0.41	0.09	0.00	−0.18	0.04	0.00
Boston
Investment	0.08	0.35	0.83	0.00	0.23	1.00	0.07	0.11	0.53
Chicago
Investment	−0.38	0.06	0.00	−0.15	0.03	0.00	−0.06	0.02	0.00
Cleveland
Investment	−0.44	0.14	0.00	−0.23	0.12	0.045	−0.10	0.06	0.13
Detroit
Investment	−0.34	0.08	0.00	−0.14	0.07	0.045	−0.08	0.04	0.05
Indianapolis
Investment	−0.90	0.10	0.00	−0.48	0.07	0.00	−0.20	0.04	0.00
Kansas City
Investment	−0.86	0.12	0.00	−0.64	0.08	0.00	−0.20	0.05	0.00
Los Angeles
Investment	−0.13	0.07	0.09	−0.00	0.04	0.97	−0.02	0.02	0.34
Milwaukee
Investment	−0.41	0.13	0.00	−0.22	0.08	0.01	−0.03	0.04	0.33
Minneapolis
Investment	−0.01	0.11	0.92	−0.05	0.05	0.35	−0.00	0.03	0.98
New Orleans
Investment	−0.31	0.14	0.03	−0.11	0.10	0.31	−0.04	0.06	0.54
New York City
Investment	−0.31	0.04	0.00	−0.03	0.02	0.09	0.00	0.01	0.89
Philadelphia
Investment	−0.66	0.11	0.00	−0.34	0.07	0.00	−0.12	0.03	0.00
Pittsburgh
Investment	0.11	0.16	0.48	0.24	0.10	0.01	0.10	0.05	0.06
San Diego
Investment	−0.16	0.28	0.56	−0.05	0.12	0.71	0.03	0.07	0.65
San Francisco
Investment	−0.26	0.13	0.04	−0.02	0.05	0.75	−0.03	0.04	0.40
Pooled Sample
Investment	−0.77	0.02	0.00	−0.11	0.01	0.00	−0.11	0.01	0.00

Each model has form: Health Prevalence = Intercept + β*(2019 NSI) + β₁*(% 60 and older) + β₂*(% non-Hispanic Black) + β₃*(% unemployed) + β₄*(% college degree or higher) + β₅*(household income) + β₆*(length of residency) + error

NSI was next texted as a mediator of associations between HOLC score and each cardiometabolic indicator. Living in a redlined neighborhood was positively associated with obesity prevalence (in Baltimore, Boston, Chicago, Indianapolis, Milwaukee, New York City, Philadelphia, and San Francisco), diabetes prevalence (in Baltimore, Boston, Chicago, Indianapolis, Los Angeles, and Philadelphia), and CHD prevalence (in Baltimore, Boston, Indianapolis, and Philadelphia) (Appendix Table 2, Path c).

Living in a redlined neighborhood was negatively associated with NSI in Baltimore, Boston, Chicago, Cleveland, Indianapolis, Kansas City, Milwaukee, New Orleans, New York City, Philadelphia, Pittsburgh, and San Francisco (Appendix Table 2, Path a). In turn, NSI was negatively associated with obesity prevalence in Baltimore, Chicago, Cleveland, Detroit, Indianapolis, Kansas City, Milwaukee, New Orleans, New York City, and Philadelphia (Appendix Table 2, Path b). NSI was negatively associated with diabetes prevalence in Baltimore, Chicago, Detroit, Kansas City, Milwaukee, Philadelphia, and Pittsburgh (Appendix Table 2, Path b). Further, NSI was associated with decreased CHD prevalence in Baltimore, Chicago, Detroit, Indianapolis, Kansas City, Philadelphia, Pittsburgh, and San Francisco (Appendix Table 2, Path b).

Ultimately, neighborhood structural investment mediated associations between HOLC score and obesity prevalence (in Baltimore, Chicago, Cleveland, Indianapolis, Milwaukee, New Orleans, New York City, and Philadelphia), diabetes prevalence (in Baltimore, Indianapolis, Milwaukee, Philadelphia, Pittsburgh), and CHD prevalence (in Baltimore, Indianapolis, and Philadelphia). This investigation observed no indirect effects among the other cities or health indicators (Figure 2).

Figure 2.

Indirect Associations between HOLC Score and Obesity, Diabetes, and Coronary Heart Disease (CHD) via Neighborhood Structural Investment: Data from the 500 Cities Project, University of Richmond’s Mapping Inequality Project, and the American Community Survey

Each model has form: Health Prevalence = Intercept + β*(2019 NSI) + β1*(% 60 and older) + β2*(% non-Hispanic Black) + β3*(% unemployed) + β4*(% college degree or higher) + β5*(household income) + β6*(length of residency) + error

Discussion

This study contributes to research about the lasting impact of historical redlining on cardiometabolic health within the U.S. Living in a formerly redlined neighborhood was associated with decreased neighborhood structural investment (NSI) across many cities between 1970 and 2019. Higher NSI was linked to reduced obesity prevalence in Baltimore, Chicago, Cleveland, Detroit, Indianapolis, Kansas City, Milwaukee, New Orleans, New York City, Philadelphia, and San Francisco. NSI was also associated with decreased diabetes prevalence in Baltimore, Chicago, Cleveland, Detroit, Indianapolis, Kansas City, Milwaukee, and Philadelphia, but increased prevalence in Pittsburgh. NSI was negatively associated with CHD prevalence in Baltimore, Chicago, Indianapolis, Kansas City, and Philadelphia.

Living in a formerly redlined neighborhood was associated with increased prevalence of obesity (in Baltimore, Boston, Chicago, Indianapolis, Milwaukee, New York City, Philadelphia, and San Francisco), diabetes (in Baltimore, Boston, Chicago, Indianapolis, Los Angeles, and Philadelphia), and CHD (in Baltimore, Boston, Indianapolis, and Philadelphia). NSI mediated associations between HOLC score and obesity prevalence (in Baltimore, Chicago, Cleveland, Indianapolis, Milwaukee, New Orleans, New York City, and Philadelphia), diabetes prevalence (in Baltimore, Indianapolis, Milwaukee, Philadelphia, and Pittsburgh), and CHD prevalence (in Baltimore, Indianapolis, and Philadelphia).

Historically redlined neighborhoods had lower NSI than non-redlined areas following the Fair Housing Act of 1968. While limited research has linked redlining and NSI, Rutan and Glass (2018), found that the highest poverty rates were concentrated in Grade C and D neighborhoods in Pittsburgh between 1970 and 2000.⁴³ While their focus was on more socioeconomic characteristics, this study explored structural aspects of the neighborhood; their findings align with those reported, where redlined areas had lower NSI in multiple cities. The NSI gap narrowed post-2000, likely due to gentrification. Cities like New York City, Los Angeles, Philadelphia, Baltimore, Chicago, San Francisco, and Pittsburgh experienced significant gentrification, which may have improved economic conditions and residential.⁵²

This study is the first to examine NSI as a mediator between redlining and cardiometabolic outcomes. Given the roughly 100-year gap between the creation of HOLC maps, understanding the specific pathways through which HOLC scores impact contemporary health outcomes is essential. This study highlights the impact of economic exclusion on health. In Pittsburgh, HOLC score had a negative indirect effect on diabetes prevalence through NSI, unlike other cities where the effect was positive. This difference could be due to Pittsburgh’s gentrification rates and other local investments, not captured by the NSI, that may differentially impact health in redlined areas.

While ecological in nature, this study aligns with existing research. Mujahid and colleagues using the Multi-ethnic Study of Atherosclerosis showed that non-Hispanic Black adults living in formerly redlined neighborhoods had lower cardiovascular health scores than those living in formerly greenlined areas.⁵³ Motairek and colleagues, using CDC PLACES data, also found associations between HOLC grade and various cardiometabolic indicators, including hypertension, diabetes, and obesity.⁵⁴

In several cities, neighborhood investment was negatively associated with obesity, diabetes, and CHD prevalence, supporting the hypothesis that investment in historically redlined communities can improve neighborhood conditions and public health. This specific link was tested and has shown that, within certain cities, neighborhoods receiving more investment after the Fair Housing Act of 1968 had a reduced burden of obesity and coronary heart disease. However, housing discrimination continued through restrictive covenants, exclusionary zoning, steering, and blockbusting. HOLC scores still correlate with lower home values and homeownership rates, which may be influenced by the dual effect of race and risk stigma placed on African Americans and other structurally marginalized communities. Since contemporary neighborhood investment, or its absence, may be guided by racial/ethnic discriminatory practices, limited neighborhood investment can exacerbate subsequent housing discrimination patterns, home value, and neighborhood wealth.

Recent research highlights the importance of considering both social and structural determinants of health. The links between historic redlining and contemporary health behaviors and outcomes beg the question of how HOLC scores have longitudinally influenced neighborhood contexts. The data suggests that neighborhood investment mediates associations between HOLC score with obesity, diabetes, and CHD prevalence, neighborhood investment may be essential in linking historic institutional racism to contemporary cardiometabolic health. Consistent with Fundamental Cause Theory, addressing this structural factor with policy or system-level changes may lessen cardiometabolic health disparities.

Associations between HOLC scores, obesity, diabetes, CHD, and neighborhood investment vary by city. These differences may reflect city-specific investment rates and the influx of younger populations with better health. By analyzing city-level data, this study offers insights that local policymakers can utilize to equitably invest in historically disinvested communities. Further, living in a redlined area increases the prevalence of contemporary cardiovascular, kidney, and metabolic health.

Strengths of this investigation include the novel use of a neighborhood structural investment measure as both a predictor and mediator of CKMH indicators across multiple U.S. cities. Using publicly available data also enhances an understanding of how local contexts influence associations between redlining and contemporary health.

Limitations

Despite these strengths, limitations include the cross-sectional design, which prohibits causal inferences and the adjustment for all relevant covariates. This study utilizes self-reported health data aggregated to the census tract, which may lead to an underestimation of associations. There may also be exposure misclassification which may misrepresent the variation in the impact of NSI on health. Further, as an ecological study, these results may be subject to ecological fallacy. Additionally, some cities had census tracts with missing data and were excluded from the sample, further impacting the generalizability of these findings.

Conclusions

In conclusion, these findings show that discriminatory housing practices, in the form of HOLC maps, are associated with current disparities in obesity, diabetes, and coronary heart disease prevalence across U.S. cities. Additional research is needed to explore mediating pathways linking historic discriminatory practices to contemporary health and wellness indicators. Specifically, more longitudinal or natural experiment studies with individual-level data would be relevant in helping to measure the temporal effects of neighborhood change on health outcomes.