Association of historic redlining and present-day health in Baltimore · Huang SJ, Sehgal NJ

Discussion

These results lend support to the current health impact of living in areas of Baltimore that were categorized as “hazardous” or “declining” by the HOLC in the 1930s, specifically with regards to life expectancy and age-differentiated mortality. To our knowledge, this is the first study to examine the association of redlining with age-differentiated mortality.

This study adds an additional health dimension of support to broader claims that past redlining still has impacts today. While Aaronson et al. [12] and Appel and Nickerson [21] focus on the current impact of redlining and yellowlining on residential segregation and household values, and Boone et al. [32] focus on the impact of redlining on greenspaces, this study lends support that there are associated ultimate impacts on life expectancy and mortality as well, at least in Baltimore. It also lends support to the hypothesis posed by Mendez et al. [31] that redlining has an impact on health outcomes. However, instead of using modern loan applications as a proxy for historical redlining [31], these results are based on the actual HOLC security maps. More broadly, these results provide support for the argument that neighborhood level characteristics [23,28–31] have an impact on health characteristics.

In addition to categorization, the HOLC maps themselves provide insight on the bias implicit in grade-assignment and how that historical social bias is associated with present day health effects. The descriptions in the HOLC maps note “detrimental influences” such as a “heavy concentration of foreigners” or characterising a preponderance of African American residents as an “infiltration”. At the time, HOLC grades suggested that whiter neighborhoods were “desirable” and colored neighborhoods “hazardous”. The gravity of these historical assignments is felt even today, as the HOLC maps were both a consequence of existing discriminatory lending practices and a cause of sustained disinvestment in neighborhoods.

Current associations of redlining with life expectancy

Even after controlling for racial composition and median household income, CSAs that were categorized as red or yellow had a 5-year shorter life expectancy than CSAs that had been categorized as green or blue. While the effect is small and we are not testing a full model for racial composition, racial composition as a control continues to have a significant association with life expectancy, which suggests that red/yellow categorization impact on life expectancy is not entirely due to the racial composition of CSAs. These results are consistent with Richardson et al. [40] which also found significant differences in life expectancy between the most and least redlined quartiles of census tracts nationwide.

Care must be taken in saying that the HOLC security maps by themselves caused the decline in life expectancy or increases in mortality at various age groups. Instead, as discussed by others [12,47], the HOLC security maps are likely a proxy for de jure and de facto government policy throughout the 20th century that systematically disenfranchised and harmed the largely Black residents of those neighborhoods [9]. These include—but aren’t limited to—lending practices, restrictive racial covenants, and political choices for municipal investment not just in the 1930s but throughout the 20th century and into today [47]. However, the initial state of racial discrimination in the policymaking of the 1930s enabled many of these other racist policies and still leaves a mark today. Aaronson et al. [12] found discontinuous city size-dependent differences in residential segregation by exploiting how HOLC maps were used in differently sized cities. This suggests that HOLC security maps may have been a driving force behind broader policies hostile to Black residents.

Maternal and child health outcomes

O’Campo et al. [29] provide strong support for the notion that low birthweight was dependent on macro level neighborhood health factors at the census tract level in Baltimore. Krieger et al. [25] likewise provide evidence for New York City. However, HOLC categories in Baltimore are not associated with low birthweight or term birth. It is possible that HOLC categories at the CSA level are not sensitive enough to variations in birthweight by census tract, or that the sample size is too small to detect such variations. It is also possible that while historical HOLC categories are related to certain neighborhood characteristics (such as the significantly associated increased prevalence of liquor stores), they might not be related to current neighborhood characteristics that are relevant for a safe birthweight. We did find associations with increased teen pregnancy and reduced prenatal care in the first trimester. While we did not assess measures of the quality of maternal care, others have noted that redlining and residential segregation may reduce access to appropriate healthcare services [23,25,29].

While we found no association with infant mortality, Lynch et al. [27] noted that current investment patterns in census tracts interact with redlining and population displacement of Black residents by white residents that may drive infant mortality more than historical redlining patterns alone. This may suggest similar patterns for other maternal and child health outcomes. While analysis including the interaction of current investment patterns with redlining is beyond the scope of this paper, it deserves further study in the future.

Age-differentiated increased mortality

The primary driver of increased mortality came from those in the 25–84 age range, especially in the 65–84 age range. There are multiple possible explanations, though the data do not capture cause of death. People living in redlined neighborhoods had less access to green spaces [32] and physical activity that might have an impact on the prevention and management of chronic conditions [6]. Multiple, lifelong encounters with hostile systems and racism can also lead to increased and premature mortality in adults [2,5].

Liquor stores

CSAs categorized as red had more per capita liquor stores than those categorized as green or blue, while CSAs categorized as yellow had fewer. While our results for red categorized CSAs are congruent with findings from Trangenstein et al. [33], we found the opposite effect for yellow categorized CSAs. However, Trangenstein et al. used census tracts rather than CSAs.