Racial Disparities and Survival Outcomes in Breast Cancer

e18588 Background: Real-world studies comparing racial differences in overall survival (OS) outcomes between African American (AA) and White (W) patients (pts) with TNBC have yielded mixed findings. This exploratory study assessed differences between national and county-level OS estimates-comparing AA vs. W pts, and identifying counties with greater racial disparities for future research. Methods: This retrospective cohort study used the Surveillance Epidemiology and End Result (SEER) cancer registry to assess county-level OS among non-hispanic AA and W pts ≥18 years of age who were diagnosed with TNBC 2011-2016. Counties with low levels of racial diversity (< 30 AA or W TNBC pts) were excluded. Five-year OS was evaluated using Kaplan-Meier and multilevel parametric OS models with a Weibull distribution. By county, OS differences between AA and W pts were tested using linear hypothesis tests after model estimation. OS models were adjusted for age at diagnosis, year of diagnosis, marital status, insurance type, surgery type, stage, grade and metastasis at brain, bone, liver and lung. Results: Among 46,125 TNBC pts in SEER, 15,723 pts (33.4% AA, 66.6% W) residing in 8 States and 47 Counties met inclusion criteria. W pts were older (mean 60; ∓14 vs 57∓13 for AA) and more likely to be married (57% vs 33%), insured (49% vs 45%) and diagnosed at an earlier stage (stage I 37% vs 29%). Controlling for potential confounders, AA pts were more likely to experience 5-year mortality compared to W pts (Hazard Ratio [HR] = 1.16; 95%CI = 1.07, 1.26). Among 47 included counties, only 4 counties demonstrated a statistically significant 5-year OS racial disparities after adjusting for confounders. The difference in HRs between AA and W pts in these 4 counties ranged from 1.49 to 2.26 as depicted in the result table (p value < 0.05). Removing these 4 counties (n = 2,929) from the final analytic cohort yielded statistically non-significant estimates (HR = 1.06; 0.97,1.16). Conclusions: There is clinically meaningful OS variation between AA and W pts with TNBC across U.S. counties included in SEER. Measuring disparities at the national or state level may mask important OS disparities at the county level. Using a county-level analytic approach, we were able to identify localities that disproportionately contribute to OS disparities at the national level. Our findings will facilitate better targeting of local interventions to address TNBC racial disparities. Future analyses should also assess counties excluded (n = 13,032) for small population sizes and/or low levels of racial diversity, as well as states not represented in SEER.[Table: see text]