Linking insurance claims across time to characterize treatment, monitoring, and end-of-life care in metastatic breast cancer.

Abstract

7063 Background: Treatment and monitoring options for metastatic breast cancer (MBC) are increasing, but little is known about patterns or predictors of their use. Insurance claims are potentially informative, but tracking patient care timelines using episodic claims data has been cumbersome. Methods: We used an advanced cohort engine, implementing a temporal query language, to link IBM Marketscan claims over time for > 125 million US individuals from 2007-2014. To select the most common MBC subtype (ER+, HER2-), our criteria were: ≥2 MBC ICD codes, ≥1 year of follow-up, and ≥1 CPT or medication code for endocrine therapy and 0 for HER2-targeted therapy. We defined aggressive MBC as ≥1 ICD code for visceral or central nervous system metastasis < 1 year after the first MBC code. Geographic area was defined by 9 Census-Bureau designated regions. End-of-record was used as a surrogate for death in a Cox regression. We used multivariate logistic regression to determine correlation of factors, including disease aggressiveness and geography, with treatment, monitoring, and end-of-life events. Results: 7,335 women met criteria for ER+, HER2- MBC, with median age 59. Nearly half (46%) had aggressive disease, which correlated with shorter survival (hazard ratio (HR) 1.5 [1.4-1.6], P < 0.001). Treatment: first-line was endocrine therapy for 64% versus chemotherapy for 36%. Monitoring: 79% were imaged mostly by CT versus 21% by PET-CT, with median between-scan interval of 81 days; 63% received CA 15-3 serum tumor markers. End-of-life: 10% had a hospice code, of whom 19% had an ICU stay and 34% chemotherapy in the prior 3 months. Correlates of care: Disease aggressiveness correlated with first-line chemotherapy (odds ratio (OR) 2.0 [1.8-2.2], P < 0.001), PET/CT (OR 1.6 [1.4-1.8], P < 0.001), more frequent scans (OR 2.3 [2.1-2.6], P < 0.001), and chemotherapy < 3 months pre-hospice (OR 1.2 [1.1-1.4], P < 0.001), but not with CA 15-3 monitoring or ICU stay. Disease aggressiveness did not vary by region (χ2 P = 0.8), but region was significantly associated with treatment, monitoring, and end-of-life care (P < 0.001). Conclusions: Approximately two-thirds of ER+, HER2- MBC patients receive first-line endocrine therapy and are monitored with CA 15-3 serum tumor markers; 10% had evidence of hospice use, a likely underestimate due to differential follow-up. MBC care patterns vary by geography while disease aggressiveness does not, suggesting that care is not optimally tailored to individuals. These insights from claims data can inform quality improvement for MBC care.