6000 Background: Reducing racial disparities in breast cancer survival has been a federal priority since the early 1990’s. We present a new method to assess disparities using sequential multivariate matching. We ask if racial disparities have increased or decreased over time and if so, what were potential reasons for such changes. Methods: We studied all women over 65 years of age in the Medicare fee for service system diagnosed with breast cancer between 1991 and 2005 who were treated in one of 12 SEER sites (the sites in SEER since 1991). There were 5,251 black patients (74% early stage (I-III), 9% late stage (IV) and 17% missing stage) and 72,695 white patients (81% early stage, 5% late stage and 14% missing stage). All black cases represented the focal group for all matches. Using multivariate matching and the propensity score, white controls were matched to blacks in steps: (1) White controls matched to black cases on age and year of diagnosis; (2) Age, year of diagnosis, and stage; (3): Age, year, stage, estrogen receptor status, grade, and 30 comorbidities. We then compare 5-year survival in the Pre and Post-Taxane periods (1991-1998, 1999-2005). Results: When whites were matched to blacks on age and diagnosis year, 5-year Kaplan-Meier survival was 69.2% vs. 56.7%, P < 0.0001. Matching additionally on stage, differences = 64.1% vs. 56.7%, P < 0.0001; Matching further on tumor characteristics and 30 comorbidities, the disparity reduced to 61.6% vs. 56.7%, P < 0.0001. Comparing trends over time, white-black differences in survival matched for age and year were 67.6% vs. 55.2% (P < 0.0001) in the pre-Taxane era (difference = 12.4%) and 71.2% vs. 58.7% (P < 0.0001) in the post Taxane era (difference = 12.5%); age and year matched paired racial differences were not different across eras (P = 0.389). Conclusions: While there may have been some improvements in overall survival, racial disparities in breast cancer survival have not improved, despite important policy initiatives and treatment advances. Adjusting for presentation at diagnosis does reduce differences in survival, but even these differences remain large and significant, suggesting that differences in both presentation and treatment given presentation are contributing to this disparity.
Effect of treatment and mammography detection on breast cancer survival over time: 1990-2007