Neighborhood-Level Redlining and Lending Bias Are Associated with Breast Cancer Mortality in a Large and Diverse Metropolitan Area

Abstract

Background: Structural inequities have important implications for the health of marginalized groups. Neighborhood-level redlining and lending bias represent state-sponsored systems of segregation, potential drivers of adverse health outcomes. We sought to estimate the effect of redlining and lending bias on breast cancer mortality and explore differences by race.

Methods: Using Georgia Cancer Registry data, we included 4,943 non-Hispanic White (NHW) and 3,580 non-Hispanic Black (NHB) women with a first primary invasive breast cancer diagnosis in metro-Atlanta (2010-2014). Redlining and lending bias were derived for census tracts using the Home Mortgage Disclosure Act database. We calculated hazard ratios and 95% confidence intervals (CI) for the associations of redlining, lending bias on breast cancer mortality and estimated race-stratified associations.

Results: Overall, 20% of NHW and 80% of NHB women lived in redlined census tracts, and 60% of NHW and 26% of NHB women lived in census tracts with pronounced lending bias. Living in redlined census tracts was associated with a nearly 1.60-fold increase in breast cancer mortality (hazard ratio = 1.58; 95% CI, 1.37-1.82) while residing in areas with substantial lending bias reduced the hazard of breast cancer mortality (hazard ratio = 0.86; 95% CI, 0.75-0.99). Among NHB women living in redlined census tracts, we observed a slight increase in breast cancer mortality (hazard ratio = 1.13; 95% CI, 0.90-1.42); among NHW women the association was more pronounced (hazard ratio = 1.39; 95% CI, 1.09-1.78).

Conclusions: These findings underscore the role of ecologic measures of structural racism on cancer outcomes.

Impact: Place-based measures are important contributors to health outcomes, an important unexplored area that offers potential interventions to address disparities.

Figure 1:. Distribution of redlining and lending bias in the Metropolitan Atlanta area.

Figure 1 illustrates the distribution of (A) redlining, and (B) lending bias indices modeled using adaptive spatial filters in the Metropolitan Atlanta area (2010–2014).