Abstract B161: Longitudinal surveillance mammography phenotypes among older breast cancer survivors

Abstract

Clinical guidelines recommend ongoing surveillance mammography for breast cancer survivors. However, little is known about longitudinal surveillance patterns among older survivors. We aimed to identify distinct surveillance mammography trajectories over five rounds of follow-up and evaluate demographic, clinical, and socioeconomic factors associated with trajectory phenotype. Using SEER-Medicare data, we included women aged ≥65 years diagnosed with non-metastatic breast cancer in 2013–2014 who survived at least 5 years after completing primary treatment, defined as 1 year after diagnosis, and remained continuously enrolled during follow-up. Surveillance mammography was identified using HCPCS/CPT codes for screening and diagnostic mammography. Screening mammograms within 14 days of a prior mammogram and diagnostic mammograms within 3 months of a prior mammogram were considered part of a diagnostic episode and excluded from surveillance. The first mammogram occurring ≥1 year after diagnosis was baseline. Subsequent surveillance intervals were iteratively classified as on-time (3 to <18 months), delayed (18 to <24 months), or missed (no mammogram within 24 months). For missed intervals, a pseudo-date 12 months after the prior mammogram was assigned to maintain longitudinal sequencing and allow up to five surveillance rounds. Ward hierarchical clustering identified surveillance phenotypes. Multinomial logistic regression estimated associations between adherence phenotypes and individual, tumor, and area-level characteristics. Neighborhood deprivation was measured using the Area Deprivation Index (ADI), a composite measure of area-level socioeconomic disadvantage based on income, education, employment, and housing characteristics, and categorized into tertiles. Among 24,548 eligible breast cancer survivors, three surveillance adherence phenotypes were identified: optimal (consistently optimal adherence; n=8,874; 36.2%), suboptimal (delayed and declining adherence; n=6,957; 28.3%), and limited (early discontinuation; n=8,717; 35.5%). Women aged ≥85 years (vs. 65–69 years) had substantially higher odds of belonging to the limited (adjusted odds ratio [aOR]=7.61; 95% CI, 6.38–9.08) and suboptimal (aOR=2.59; 2.13–3.15) phenotypes. African American women were more likely to belong to the limited (aOR=1.88; 1.67–2.11) and suboptimal (aOR=1.32; 1.15–1.52) phenotypes, as were Hispanic women (limited: aOR=1.57; 1.37–1.80; suboptimal: aOR=1.35; 1.18–1.55). Women residing in the most socioeconomically deprived neighborhoods also had higher odds of belonging to the limited (aOR=1.31; 1.16–1.49) and suboptimal (aOR=1.17; 1.04–1.32) phenotypes. Nearly two-thirds of older survivors had suboptimal or limited longitudinal adherence to surveillance mammography. Older age, African American race, Hispanic ethnicity, and neighborhood deprivation were associated with limited surveillance adherence, highlighting populations that may benefit from targeted interventions to improve long-term surveillance mammography adherence. Luchang Cui, Swarnavo Sarkar, Eunji Choi, Yiwey Shieh. Longitudinal surveillance mammography phenotypes among older breast cancer survivors [abstract]. In: Proceedings of the 19th AACR Conference on the Science of Cancer Health Disparities; 2026 Oct 18-21; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2026;35(10 Suppl):Abstract nr B161.