OBJECTIVES: Cost of cancer care is high, with its large portion stemming from costly cancer treatments. Cancer treatments are often critical for cancer care, but some treatments add little to no clinical benefits. Using clinically unnecessary treatments results in adverse medical (treatment-related reactions) and financial (increased costs) impacts. It is not thus recommended by clinical guidelines. Our study examined use of not-recommended treatments in Medicare in the U.S., focusing on granulocyte-colony stimulating factors [GCSF] for patients receiving low-risk (<10% neutropenic fever) chemotherapy; and denosumab for patients with castration sensitive prostate cancer [CSPC]. Particularly, we analyzed whether disadvantaged populations, who usually have low use of recommended services, also have low utilization of not-recommended cancer treatments.

METHODS: We identified newly-diagnosed cancer patients after a one-year wash-out period. The GCSF cohort comprised Medicare beneficiaries with breast, lung, or colorectal cancer who received low-risk chemotherapy. The denosumab cohort comprised patients with CSPC and without indications for bone-modifying agents. We estimated logit models where the outcome was the receipt of not-recommended treatments. Explanatory variables of interest were racial/ethnic minority indicators, rural residence, and Medicare-Medicaid dual eligibility, which is a proxy for low-income.

RESULTS: Black beneficiaries had higher rates of GCSF use (9.4%) than White beneficiaries (8.3%; p<0.01). Dually eligible individuals had lower rates of GCSF use (7.5%) than Medicare-only beneficiaries (versus 8.7%; p<0.01). Among patients with CSPC, black beneficiaries were less likely to receive denosumab compared with Whites (27.2% versus 32.8%; p<0.01). Use of any not-recommended treatment was not significantly different between rural and urban residents.

CONCLUSIONS: Disadvantaged populations generally had similar or low rates of not-recommended cancer treatments. However, this finding was not consistent across all disadvantaged groups or by treatment. Our results suggest that targeted approaches depending on treatment would increase the effectiveness of efforts to reduce utilization of not-recommended cancer treatments.