RESUMEN
BACKGROUND: Breast cancer survival disparities by race are likely multifactorial. In a small pilot cohort, we demonstrated a statistical interaction between age and race. The purpose of this study was to validate earlier findings in a larger, more diverse cohort and to test the hypothesis that breast cancer survival is influenced by the dependent relationship of age and race. METHODS: We conducted a retrospective analysis of a multi-institutional breast cancer database for patients treated between 1999 and 2009. Study variables included age and disease stage at diagnosis, race, treatment (surgery, chemotherapy, radiotherapy, hormone therapy) and overall survival. Statistical analysis and regression models were performed by Stata software. RESULTS: A total of 9,249 patients were included in this study. African American, Hispanic, and Asian patients were more likely to present at a younger age with metastases. African American and Hispanic race were associated with increased mortality after adjusting for stage, age, and treatment. A 2-way interaction between age and race was identified in the Cox regression model (p < 0.001). To further define this interaction, a postestimation analysis was performed to determine the predicted relative hazard for each race with age fixed at 40, 50, 60, 70, and 80 years. At younger ages, the predicted relative hazard was significantly higher for both African American and Hispanic race. CONCLUSIONS: Despite adjusting for stage and treatment differences, African American and Hispanic race predicted poor survival. The effect of age and treatment on breast cancer survival differs across races. Additional research is needed to accurately determine the reasons for worsened survival.