[Racial disparity in 10-year breast cancer survival: a mediation analysis using potential responses approach].

The study's objectives were to investigate the association between race/color and 10-year survival in women with breast cancer and the role of staging as mediator. This was a hospital cohort with 481 women with invasive breast cancer diagnosed from 2003 to 2005. Comparisons were made between white and black women as to sociodemographic characteristics and staging, using the chi-square test, and 10-year survival, using Kaplan-Meier and Cox regression methods. For the race/color variable, direct and indirect effects were estimated, mediated by staging, with adjustment for the socioeconomic status of the woman's area of residence and age, using the potential responses (counterfactual) model and Cox multiple regression. Black women living in low-income census tracts were more likely to use the public health care system and to be diagnosed at more advanced stages. Breast cancer-specific 10-year survival was 64.3% (95%CI: 60.0; 68.9), with a significant difference between whites (69.5%; 95%CI: 64.8; 74.6) and blacks (44; 95%CI: 35.2%; 55.1). In the multivariate models adjusted for income and age, black women had worse prognosis (HR = 2.09; 95%CI: 1.76; 2.51), and the proportion mediated by staging was 40% (95%CI: 37; 42). There is racial disparity in 10-year breast cancer survival in Brazilian women, mediated mainly by more advanced staging at diagnosis in black women. This highlights the need to expand both the coverage and the quality of breast cancer screening and to facilitate access to early diagnosis and treatment in order to reduce racial inequality.