Abstract B89: Multivariate modeling of biological, socioeconomic, and treatment level variables partially explains the differential breast cancer mortality between African Americans and white Americans in Michigan

Abstract

Abstract Introduction In this study, we examine whether biological, socioeconomic and treatment level factors can explain the observed 2-fold difference in 5-year breast cancer mortality between African Americans (16%) and Whites (8%) in Michigan. We aim to inform the development of robust targets of intervention to reduce this racial disparity. Materials & Methods Our study examines differences in biological factors (stage, grade, estrogen receptor status (ER), progesterone receptor status, socioeconomic factors (insurance type and median income quintile) and treatment level factors (surgery, radiation), as contributors to the racial difference in breast cancer specific mortality. Case data was extracted from the Michigan Cancer Surveillance Program. Subjects diagnosed in 2004 with breast cancer (n=6648) and followed for 5 years, without missing values for key predictor variables were utilized in our analysis. Selecting these specific cases did not greatly alter the racial or age distribution of cases. Results Bivariate analysis showed that African Americans were more than twice as likely to be diagnosed at a metastatic stage than their White counterparts. This was the most prominent predictor variable with an odds ratio (OR) of 60.349 for metastatic cancer relative to in situ disease (p<0.0001). ER negativity (ER−) and PR negativity (PR−) were independent predictors of breast cancer mortality with ORs of 1.694 (p<0.05) and of 1.707, (p<0.05) respectively. African Americans were more likely to be ER− (30.24%) and PR− (42.87%) relative to their White cohorts (ER-20.41%, p<0.05; PR-29.25%, p<0.05). Analysis of insurance type indicated that those who did not have insurance had 1.420 times the odds of dying as those who had private insurance (p<0.05). African Americans had a greater percentage uninsured (18.89%) versus Whites (13.85%, p<0.05). Among treatment variables, receiving surgery was associated with a 73% reduction (p<0.05) in the odds of mortality over not receiving surgery. African Americans were more than twice as likely to not have surgery as their White counterparts (AA: 11.70%, Whites: 4.81%, p<0.05). Finally, receiving radiation therapy was associated with a 14% reduction in the odds of breast cancer associated mortality (p<0.05). African Americans were significantly more likely to not receive radiation therapy than their White cohorts (AA: 66.28% and White: 57.04%, p<0.05). Multivariate analysis indicated that race ceased to be a significant predictor of mortality even after adjusting for biologic, socioeconomic and treatment level variables. The adjusted odds ratio for mortality among African Americans was 1.115 (95% Confidence Interval: 0.848 to 1.466). The unadjusted OR for this variable was 2.11, (p<0.05). Thus, ∼90% of the increased risk for mortality associated with being African American was explained by the predictor variables utilized in our multivariate model. Discussion & Conclusions We determined that race ceases to be an independent predictor of breast cancer mortality when key biological, socioeconomic and treatment level predictor variables are accounted for. Careful examination of the dataset reveals that African Americans are diagnosed at a more advanced stage and grade. Interventions that promote more aggressive breast cancer screening of African Americans would be recommended to increase the likelihood of diagnosis at an earlier stage and grade. However, African Americans also tend to have more aggressive tumor biology reflected in their greater prevalence of ER− and PR− tumors; therefore, part of the racial difference in mortality may be biological. Conversely, socioeconomic factors such as insurance and median income quintile also contributed significantly to the disparity in breast cancer mortality. More African Americans were uninsured, which was an independent predictor of mortality in our adjusted multivariate model. Finally, our analysis showed that surgical and radiation treatments were each associated with lower odds of mortality. African Americans were less likely to either receive surgical or radiation treatment for all stages of breast cancer except metastatic disease. This difference in surgical practice was not due to difference in tumor stage, but rather reflected racial differences in treatment choices. Thus, addressing factors that affect treatment or treatment access for African American patients may be an important target of public health intervention. Citation Information: Cancer Prev Res 2011;4(10 Suppl):B89.