Using geospatial methods to analyze associations of neighborhood change on breast cancer mortality in Maryland, 2000-2019

Purpose : To examine whether neighborhood change status (declining, stable, or upgrading) is associated with the risk of breast cancer (BC)-specific mortality. Background : Neighborhood-level social determinants of health are associated with BC mortality, but few studies have investigated how dynamic neighborhood conditions over time and geography (i.e., neighborhood trajectories) can modulate the risk of BC-specific mortality. Methods : We identified primary BC cases diagnosed among women age ≥18 in the Maryland Cancer Registry (NPCR) between 2005-2019 with valid census tract information at diagnosis (proxy for neighborhood). Observed and expected BC deaths were calculated by tract using 2000 standard female population and referent BC mortality rates for Maryland from 2018-2022 (age <50: 4.4 deaths/100k; age ≥50: 60.8 deaths/100k). Tracts were linked to a continuity-adjusted version of the Social Vulnerability Index (CA-SVI), which harmonized SVI component variables across years. Neighborhood change was defined as the difference in CA-SVI percentiles between consecutive 5-year periods (2000-2004, 2005-2009, 2010-2014, 2015-2019). Tracts were classified as stable (Low, Low-Medium, Medium-High, High CA-SVI)), upgrading (−25 to −49 or ≤−50 %ile; becoming moderately/substantially less disadvantaged), or declining (+25 to +49 or ≥+50 %ile; becoming moderately/substantially more disadvantaged). We evaluated the relationship between neighborhood change and BC mortality using Bayesian negative binomial space-time models that captured effects for geographic clustering and temporal trends. Models were adjusted for individual-level covariates (age, stage, grade, receipt of treatment [surgery, chemotherapy, radiation, hormone], and tumor characteristics [ER, PR, HER2 status]), starting/ending CA-SVI categories, and county-level rurality (2013 USDA Beale Rural-Urban Continuum Codes). Results : Among 60,388 BC cases, 8,946 died as a result of their diagnosis. Women residing in neighborhoods that modestly improved (+1 cat.) had a slightly elevated risk of BC mortality compared to those living in areas of stably low CA-SVI (Relative risk [RR] Inc =1.11, 95% Credible Interval [CrI] 1.01-1.22). Women living in neighborhoods that became more disadvantaged had between 11 to 24% higher risk of BC mortality (RR Decl1 cat. =1.11 [1.01-1.02], RR Decl2 cat. =1.24 [1.04-1.48]). Conclusion: Both sustained vulnerability and shifts in neighborhood conditions were associated with BC-specific mortality. Marked declines showed the highest risk for BC mortality, but sustained vulnerability and modest upgrades also indicated elevated risk. These results highlight the importance of neighborhood contexts on cancer outcomes, emphasizing the need for longitudinal research evaluating how neighborhood trajectories might affect place-based inequities.