HSR19-079: Disparities in Accessing Screening Mammography: Opportunities for Improving Diagnostic Outcomes

Introduction: Screening mammography is a key component of secondary prevention programs targeting reductions in breast cancer mortality. The early detection of cancers facilitates treatment at a more curable, locoregionally limited stage. We describe characteristics and outcomes of women who had only one screening mammogram versus those who had annual or biennial screens. Methods: A cloud-based big data platform is being used to integrate and transform data from electronic medical records, radiology management systems, and tumor registries to create a learning health system. This analysis includes data from 227,834 women, aged 40–79 years, who underwent screening mammograms between January 2015 and June 2018 at 64 imaging facilities within 3 large, geographically diverse healthcare organizations. Patients with breast cancer history or implants were excluded. Women were defined as having one screen if they had >24 months of follow-up with evidence of only one screen and were defined as having more than one screen if they had 2 screens at least 9 months apart. Interval cancer was defined as a breast cancer in the 12 months following a negative baseline mammogram. The chi-square test was used to test for differences between cohorts. Results: Of 227,834 women, 18.8% (n=42,911) met criteria for one screen [1-screen] and 81.2% (n=184,923) for 2 screens [2-screens]. There were significant differences between the groups in age (40.4% 60-79 years in the 1-screen cohort vs 49.1% in 2-screens; P<.001), race (24.7% African American and 5.3% Asian in the 1-screen cohort and 18.5% and 3.6% in 2-screens; P<.001), and lifetime risk of breast cancer (6.9% were in the elevated risk category in the 1-screen cohort and 9.3% in 2-screens; P<.001). Recall rate for the 1-screen cohort was 16.6% compared to 7.7% for the second screen for the 2-screens (P<.001). The interval cancer rate was significantly higher (P<.001) for the 1-screen cohort (2.9 per 1000 screens) as compared to the second screen for the 2-screens (0.8 per 1000 screens). Conclusion: Women with evidence of only one screen during the 3.5-year study period tended to be younger and non-white. Although they had lower scores for lifetime risk of breast cancer, recall rates were twofold higher and interval cancer rates were threefold higher in the 1-screen cohort. Targeted initiatives are needed to improve adherence to screening in women at risk of noncompliance.