Quality of treatment selection for Medicare beneficiaries with cancer.

Abstract

140 Background: Because NCCN guidelines recommend multiple chemotherapy regimens, guideline-concordant treatments often differ substantially in clinical benefit. Variation in treatment quality – whether patients receive the most highly-recommended treatment option for their cancer – has not been assessed broadly across cancer types. Methods: Population-based cohort study using linked SEER-Medicare and NCCN Evidence Blocks (EB) data, 2015-2017. We reviewed historical NCCN Guidelines and ranked recommended treatments according to their EB scores. We then identified indications for initial systemic therapy in which 1) ≥1 treatment dominated (higher Efficacy AND Safety scores); 2) indication was identifiable in SEER-Medicare (eg, not biomarker-defined); 3) NCCN EB and ASCO Value Framework were in agreement regarding the optimal treatment, and 4) sufficient SEER-Medicare cases from initial EB publication date through 2017. For each indication, we identified the corresponding SEER-Medicare cohort. The primary outcome was initial systemic therapy received, categorized as Optimal (highest EB scores), Lower But Non-Inferior (LNI; lower overall EB scores but Efficacy or Safety = Optimal therapy), Inferior/Non-Recommended, Unclassifiable, or No Systemic Therapy. Exposures of interest were patient race/ethnicity, comorbidity, regional median income and poverty prevalence, Part D Low Income Subsidy (LIS) participation, and treatment at an NCI-designated cancer center. We fit multinomial logistic regression models adjusting for patient characteristics and cancer indication. Results: N=9,290 across 11 cancer indications; 18% (1,703) received Optimal therapy, 22% (2,041) LNI, 10% (904) Inferior/Non-Recommended, 7% (688) Unclassifiable, and 43% (3,954) No Systemic Therapy. Compared to non-LIS participants, LIS participants trended towards lower likelihood of receiving Optimal therapy vs. LNI (OR = 0.87, 95%CI 0.72-1.04), Inferior/Non-Recommended (OR = 0.79, 95%CI 0.62-1.01), Unclassifiable (OR = 0.74, 0.57-0.95), or No Systemic Therapy (OR = 0.54, 95%CI 0.46-0.64). Patients with ≥3 comorbidities (vs. 0) were less likely to receive Optimal therapy vs. LNI (OR = 0.78, 95%CI 0.64-0.95), Inferior/Non-Recommended (OR = 0.77, 95%CI 0.59-1.01), Unclassifiable (OR = 0.49, 0.37-0.66), or No Systemic Therapy (OR = 0.66, 95%CI 0.55-0.79). Other exposures were not associated with Optimal therapy. Conclusions: After adjusting for comorbidity – which may be an appropriate consideration in therapy selection – LIS participants remained less likely to receive anticancer systemic therapy and, contingent on receiving therapy, less likely to receive Optimal therapy for their cancer type. These results suggest that despite the LIS program subsidy for oral/Part D drugs, LIS-participating cancer patients may still face financial barriers to receiving infused/Part B cancer drugs.