Historical redlining and clustering of present-day breast cancer factors.

Abstract

Purpose: Historical redlining, a 1930s-era form of residential segregation and proxy of structural racism, has been associated with breast cancer risk, stage, and survival, but research is lacking on how known present-day breast cancer risk factors are related to historical redlining. We aimed to describe the clustering of present-day neighborhood-level breast cancer risk factors with historical redlining and evaluate geographic patterning across the US.

Methods: This ecologic study included US neighborhoods (census tracts) with Home Owners' Loan Corporation (HOLC) grades, defined as having a score in the Historic Redlining Score dataset; 2019 Population Level Analysis and Community EStimates (PLACES) data; and 2014-2016 Environmental Justice Index (EJI) data. Neighborhoods were defined as redlined if score ≥ 2.5. Prevalence quintiles of established adverse and protective breast cancer factors relating to behavior, environment, and socioeconomic status (SES) were used to classify neighborhoods as high-risk or not. Factor analysis grouped factors into domains. Overall and domain-specific scores were calculated for each neighborhood according to historical redlining status. Percent difference in score by historical redlining was used to assess differences in average scores, with Wilcoxon-Mann-Whitney test used to estimate significance. Kappa statistic was used to estimate concordance between historical redlining status and high-risk status. Heatmaps of scores were created to compare spatial clustering of high-risk factors to historical redlining.

Results: We identified two domains: (1) behavior + SES; (2) healthcare. Across the US, redlined neighborhoods had significantly more breast cancer factors than non-redlined (redlined neighborhoods = 5.41 average high-risk factors vs. non-redlined = 3.55 average high-risk factors; p < 0.0001). Domain-specific results were similar (percent difference for redlined vs. non-redlined: 39.1% higher for behavior + SES scale; 23.1% higher for healthcare scale). High-scoring neighborhoods tended to spatially overlap with D-grades, with heterogeneity by scale and region.

Conclusion: Breast cancer risk factors clustered together more in historically redlined neighborhoods compared to non-redlined neighborhoods. Our findings suggest there are regional differences for which breast cancer factors cluster by historical redlining, therefore interventions aimed at redlining-based cancer disparities need to be tailored to the community.