Abstract C007: Multilevel analysis of person-, county-, and state-level contributors to triple-negative breast cancer diagnosis among women in the United States

Abstract

Abstract While multiple individual level factors have been identified to play a role in the etiology of breast cancer and triple-negative breast cancer, few studies have used mixed modeling techniques to explore the role that additional levels of predictors may play in triple-negative breast cancer. Since disparities persist even when known factors are accounted for, the scope of research must be expanded to examine factors that contribute to disparate outcomes at an ecologic level. Mixed modeling can perhaps better account for the spatial heterogeneity in these cases. Random intercept mixed models can account for different effects across areas, yet they assumes higher levels are independent, thus are limited in terms of handling spatial dependence. In this study, we explored the odds of triple-negative diagnosis, given breast cancer diagnosis, at the individual level, controlling for individual, county and state level variables. The intraclass correlations for county and state were 3.8% and 0.8%, respectively. When controlling for county and state level predictors, disparities by age, race and stage persisted. Non-Hispanic black women consistently had twice the odds of diagnosis with TNBC, women age 40 and under had 1.7 times the odds of diagnosis, and women diagnosed at late stage had 1.5 times the odds of diagnosis. County-level residential segregation and educational attainment variables were significant predictors of triple-negative diagnosis, while no state level policy variables were statistically significant predictors, after controlling statistically for random state intercepts that account for other omitted state variables. Residential isolation proved to be disadvantageous to diagnosis, while residential diversity and area educational attainment were protective. Future studies should continue to explore various environmental factors, physical and social, that contribute the variation in disparate rates of diagnosis. Citation Format: Lia C. Scott. Multilevel analysis of person-, county-, and state-level contributors to triple-negative breast cancer diagnosis among women in the United States [abstract]. In: Proceedings of the Eleventh AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2018 Nov 2-5; New Orleans, LA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(6 Suppl):Abstract nr C007.