Measuring geographic variation in quality and equity of breast cancer care.

Abstract

139 Background: Persistent higher breast cancer mortality among Black Americans is partly due to treatment disparities. There is a paucity of surveillance data quantifying racial disparities at local levels, and the relationship of geography to racial disparities in treatment is poorly understood. We sought to develop and apply a claims-based set of breast cancer quality measures, stratified by race, across regions of North Carolina. Methods: We created a retrospective cohort (n = 37,021) of stage I-III breast cancers from 2004-2017 using linked cancer registry and multi-payer claims data. We assessed 4 quality measures derived from ASCO/NCCN and Commission on Cancer including chemotherapy (CT) for hormone receptor negative disease, radiation (RT) after breast conserving surgery (BCS), RT for N2-3 disease, and adjuvant endocrine therapy (ET). Each measure was limited to the ages and tumor features eligible by the ASCO/NCCN/CoC guideline. We estimated age-adjusted overall performance (%) on each measure, and the Black-White gap, at the state level and across 9 geographic sub-regions. We used multivariable Poisson regression to estimate Risk Ratios (RR) comparing quality performance for Black vs. White patients overall and across regions. Exploratory analyses adjusted for patient- and neighborhood-level social determinants of health (SDOH). Results: Median age of the cohort was 68 years and 19% of patients were Black. At the state level, the % of eligible patients meeting quality measures ranged from 73% for adjuvant ET to 95% for RT after BCS. There was significant inter-region variation on all 4 measures. Black patients were less likely to receive adjuvant CT (RR = 0.92, CI 0.87-0.96), post-BCS RT (RR = 0.98, CI 0.97-0.99) and ET (RR = 0.95, CI 0.93-0.98). There was no racial disparity in RT for N2-3 disease. Both overall performance and the magnitude of disparity varied widely across regions and geographic patterns were not consistent across measures. For example, the Charlotte region had the top quality performance for RT for N+ (90%), but worst for ET (70%). Similarly, the Wake region had no racial disparities in CT or RT, but the largest disparity of any region for ET (RR = 0.90, CI 0.84-0.96). Adjustment for SDOH modestly attenuated the relationship between race and quality performance, but significant disparities persisted in the regions with the largest racial gaps. Conclusions: Racial disparities persist in breast cancer care delivery, but with non-uniformity of performance across measures and varying local patterns of disparities. Our findings suggest that cross-cutting interventions spanning the care continuum, and/or interventions adapted to local challenges, may be needed to effectively address racial disparities. The contributions of local resources, structural, and institutional barriers to these patterns remain to be explored. Community-engaged work to share and interpret these findings is ongoing.