Results
The BNIA shapefiles included 55 CSA polygons for Baltimore (Figs 1 and 2). The Mapping Inequality shapefiles for Baltimore included 42 HOLC categorized polygons (Fig 1). A total of 156 polygons were created with an intersect overlay between CSA polygons and HOLC polygons (Fig 3). One Baltimore CSA polygon (#7, Cherry Hill) was dropped from the analysis since there were no intersections with the Mapping Inequality shapefiles (the area was not assigned any HOLC categories in the 1930s). 54 primary polygons represented the remaining CSAs with HOLC categories assigned.
Figure. 2013 Baltimore CSA neighborhoods (numbers refer to CSA names) in light purple overlaid on top of HOLC redlining map (red: “Hazardous”, yellow: “Declining”, blue: “Still desirable”, green: “Best”).HOLC redlining map republished from Nelson RK, Winling L, Marciano R, Connolly N, et al. Mapping Inequality. n.d. [cited 2017 Nov 6]. In: Nelson RK, Ayers EL, eds. American Panorama [Internet]. Richmond: University of Richmond Digital Scholarship Lab. Available from: https://dsl.richmond.edu/panorama/redlining/#loc=11/39.293/-76.808&city=baltimore-md under a CC BY license, with permission from University of Richmond Digital Scholarship Lab, original copyright 2016. (See the original article for the figure.)
Figure. 2013 Baltimore CSAs by life expectancy.Darker blue is longer life expectancy. Numbers refer to CSA names. (See the original article for the figure.)
Figure. 2013 Baltimore CSAs’ life expectancy overlaid on top of HOLC redlining map.HOLC redlining map republished from Nelson RK, Winling L, Marciano R, Connolly N, et al. Mapping Inequality. n.d. [cited 2017 Nov 6]. In: Nelson RK, Ayers EL, eds. American Panorama [Internet]. Richmond: University of Richmond Digital Scholarship Lab. Available from: https://dsl.richmond.edu/panorama/redlining/#loc=11/39.293/-76.808&city=baltimore-md under a CC BY license, with permission from University of Richmond Digital Scholarship Lab, original copyright 2016. (See the original article for the figure.)
There were 14 CSAs that were classified as red using the predominant polygon by area. 19 were classified as yellow, 17 as blue, and 4 as green. Red and yellow polygons tended to be at the city core, while green and blue polygons tended to be at the northern peripheries (Fig 1).
Summary statistics
CSA characteristics are fairly heterogeneous in Baltimore. The mean life expectancy at birth in the 54 CSAs was 73.3 years, with a standard deviation of 4.33 years, and ranging from a low of 66.0 to a high of 85.3. Mean of median CSA household income was $44,610 with a low of $13,887 to a high of $106,771. Mean African American proportion of the population was 61.2%, with a low of 2.44% to a high of 97.4% (Table 1).
Table. Summary statistics for 2013 Baltimore CSAs. (See the original article for the full table.)
OLS regression
CSAs with red classification were statistically significantly associated at the 95% confidence level with a mean 4.01 years reduction (95% CI: 1.47, 6.55) in CSA life expectancy and CSAs with a yellow classification were associated with a mean 5.36 years reduction (95% CI: 3.02, 7.69) in CSA life expectancy (Table 2). Once CSA proportion of African Americans and CSA median household income were added to the model as controls, the effect was slightly stronger in red CSAs with a mean reduction of 5.23 years (95% CI: 3.49, 6.98) and slightly weaker in yellow CSAs with a mean reduction of 4.93 years (95% CI: 3.22, 6.63). This model accounted for 73% of the variability of the life expectancy response data around the mean. The Variance Inflation Factor of this model was 2.17, so multicollinearity seems less likely to be an issue. In the fully adjusted model, yellow classification is also significantly associated with mortality between the ages of 15–24 (mean 4.65 additional deaths per 10,0000 people), while yellow and red classifications are also significantly associated with mortality between the ages of 25 and 44 (mean 10.54 (yellow) and 6.19 (red) additional deaths per 10,000 people), mortality between the ages of 45 and 64 (mean 30.99 (yellow) and 36.4 (red) additional deaths per 10,000 people), mortality between the ages of 65 and 84 (mean 74.65 (yellow) and 95.47 (red) additional deaths per 10,000 people) (Table 3).
Table. Multivariable ordinary least squares analysis of Baltimore HOLC and 2013 life expectancy. (See the original article for the full table.)
Table. Multivariable ordinary least squares analysis of Baltimore HOLC and 2013 mortality by age bins. (See the original article for the full table.)
HOLC classification was not significantly associated with infant mortality, mortality from ages 1–14, or mortality over the age of 85 (Table 3). Red or yellow classification was not significantly associated with the health outcomes of percent of births given at term, percent of births at satisfactory weight, or number of children testing positive for elevated blood lead levels (Table 4). Red and yellow classification was significantly associated with the teen birthrate (additional 23.61 (yellow) and 20.36 (red) births per 1000) (Table 4). The process outcome of percentage of births with prenatal care was not significant for red classification but was for yellow (5.06 percent fewer births) (Table 4). For CSA characteristics, red or yellow classification was associated with a significant increase of 0.87 (yellow) and 2.18 (red) additional liquor stores per 1000 residents but was not significantly associated with the rate of fast-food restaurants (Table 5).
Table. Multivariable ordinary least squares analysis of Baltimore HOLC and 2013 maternal and child health outcomes and processes. (See the original article for the full table.)
Table. Multivariable ordinary least squares analysis of Baltimore HOLC and 2013 neighborhood characteristics. (See the original article for the full table.)
For our sensitivity analysis, 8 CSAs were excluded for having less than 25% HOLC coverage (CSAs 4, 9, 12, 30, 38, 44, 45, and 50). Five excluded CSAs were categorized as yellow, and the remaining 3 were distributed evenly among red, blue, and green. Rerunning our first model while excluding these CSAs resulted in effect sizes with the same directionality and significance, and the effect size was mildly stronger with yellow CSAs associated with a mean 5.78 years reduction (95% CI: 3.01, 7.69) and red CSAs associated with a mean 4.44 years reduction (95% CI: 1.61, 7.27).
CC BY 4.0. "Association of historic redlining and present-day health in Baltimore" by Huang SJ, Sehgal NJ, PloS one (2022). https://doi.org/10.1371/journal.pone.0261028. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Peer-reviewed and open access; full text from PubMed Central (PMC8769359). Headings and layout adapted for reading; figures and tables are in the original.