e13511 Background: Evidence suggests that centralization of care in high(er) volume facilities is an effective strategy to improve patient outcomes. However, particularly in the case of breast cancer, concerns about the potential impact on timely access to care have limited its promotion. Leveraging a natural experiment afforded by New York’s statewide implementation of a policy in 2009 restricting Medicaid reimbursement for breast cancer surgery provided at low-volume hospitals (<30 breast cancer surgeries annually), we examined the effect of centralization of care on days from breast cancer diagnosis to initial treatment. Methods: From a linked dataset merging NY state tumor registry with hospital discharge data, we identified 71,135 women diagnosed with Stage 1-3 incident breast cancer who underwent surgery as first course of treatment during the pre-policy (2004-08) and post-policy (2010-13) periods. Multivariable difference-in-difference models were used to estimate the policy effect on the probability of experiencing delayed care (>60 days) between diagnosis and initial surgical treatment by Medicaid beneficiaries before and after policy implementation relative to the experience of commercially or otherwise insured women unaffected by the policy change. Results: The cohort consisted of 8435 Medicaid beneficiaries (4070 pre-/4365 post-policy) and 62,700 non-Medicaid women (34,909 pre-/28,610 post-policy). The mean time to surgery pre- and post-policy was 34 (SD=37.2) and 45.5 (SD=41.3) days for Medicaid beneficiaries and 27.5 (SD=27.7) and 36 (SD=28.4) days for non-Medicaid patients. Multivariate estimates indicate that, regardless of insurance status, the more recently treated patients were more likely to wait >60 days to initial treatment (OR=1.5, 95% CI=1.43-1.59, p<0.001) suggesting a temporal trend of longer time to treatment unrelated to the centralization policy. Regardless of the policy period, Medicaid beneficiaries were more likely to experience delayed care than commercially insured women (OR=1.75; 95% CI=-1.59 to 1.92, p<0.001). Difference-in-difference estimates, however, indicate that the adjusted probability of delayed care among Medicaid patients (from 14.3% to 20.5%) was not significantly larger than that of patients not affected by the policy (from 8.8% to 12.6%). Controlling for time trends and policy effects, age, race, ethnicity, comorbidities, urbanicity, tumor size, nodal status and hormone receptor status, Black women (OR=1.74, 95%CI=1.63-1.85), those of other races (OR=1.15; 95%CI=1.01-1.24), and Hispanic women (OR=1.34; 95%CI=1.24-1.44) were more likely to experience delayed care than their white or non-Hispanic counterparts. Conclusions: Given these findings and growing evidence of better survival among breast cancer patients treated at high(er) volume hospitals, centralization of breast cancer care may be one effective approach to improving breast cancer outcomes.