Febrile neutropenia (FN) during adjuvant chemotherapy is associated with morbidity, mortality risk, and substantial cost. The optimal duration of filgrastim as primary FN prophylaxis in early stage breast cancer (EBC) patients is unknown. We compared the cost-effectiveness of 5, 7 or 10 days of filgrastim offered for primary FN prophylaxis.
We conducted a cost-utility analysis from the perspective of Canada’s health care system using a decision tree to evaluate the costs and quality-adjusted life years (QALYs) based on the results of the REaCT G study. In this study EBC patients who were to receive adjuvant chemotherapy were randomized to 5 versus 7 or 10 days of filgrastim as primary FN prophylaxis. Using EQ-5D-5L measures from patients at baseline, after one cycle and at the completion of chemotherapy, along with the cost of complications of febrile neutropenia and treatment-related hospitalization from the provincial database, we determined incremental cost-effectiveness ratio (ICER) for the 7 or 10 days of filgrastim compared to 5 days. Both deterministic and probabilistic sensitivity analyses were carried out.
From 464 patients on the REaCT G trial showed that 5 & 7 days compared to 10 days filgrastim were associated with lower costs and a slight decrease in QALYs. Compared to 5 days of filgrastim, 10 days was associated with an increase cost $3,763 and 0.019 QALYs with ICER of $198,052, and 7 days of filgrastim led to a rise in cost by $2,461 and QALYs by 0.004 with ICER of $615,250. The probabilistic sensitivity analysis showed that cost-effectiveness results were uncertain. At commonly willingness to pay threshold of $50,000/QALY, probabilities that 5, 7 and 10 days filgrastim are cost-effective were 48%, 35%, and 18%, respectively.
We conclude that shorter schedules of filgrastim provide cost savings with a slight decrease in QALY. The schedule of filgrastim prophylaxis that would be considered cost-effective depends on the willingness of the payer.
Mark Clemons.
REaCT Program and Cancer Care Ontario.
All authors have declared no conflicts of interest.