Therapeutic Use of Granulocyte Colony-Stimulating Factors for Established Febrile Neutropenia: Effect on Costs from a Hospital Perspective

Background: The prophylactic use of granulocyte colony-stimulating factors (G-CSFs) reduces the severity and duration of neutropenia and reduces the incidence of febrile neutropenia after cancer chemotherapy. However, the use of G-CSFs, particularly filgrastim, to treat established neutropenia remains controversial. A recent meta-analysis of randomised controlled trials (RCTs) evaluating G-CSF treatment for established febrile neutropenia demonstrated a reduction in prolonged hospitalisations. Because more than one-third of patients in the analysis were hospitalised for at least 10 days, this finding has broad pharmacoeconomic and clinical significance. This analysis presents the potential cost implications of G-CSF treatment for established neutropenia among hospitalised patients. Methods: Direct medical costs ($US, year 2003 values) related to hospitalisation for established neutropenia were modelled using a hospital perspective and according to two treatment options: (i) no use of G-CSF during the neutropenic episode (control); and (ii) addition of daily G-CSF until neutrophil recovery. Within each option, we modelled the probability of a long stay (>=10 days) and patient survival. The model used three data sets: discharge data from a consortium of academic medical institutions, drug cost data (filgrastim) from Federal payers, and estimates of G-CSF efficacy derived from a meta-analysis of RCTs of treatment in patients with established febrile neutropenia. The lowest expected total cost was predicted for both treatment options; sensitivity analyses and Monte Carlo simulations were used to evaluate the robustness of the model. Results: The G-CSF arm produced the lowest expected cost, and predicted net estimated savings of $US1046 per neutropenic episode compared with the control strategy. G-CSF was less expensive than the control for most reasonable estimates of cost per day and all lengths of stay (LOS) >=10 days. G-CSF was the least costly strategy for 73.5% of 10_000 Monte Carlo iterations, while the no-G-CSF control strategy predicted savings in 26.5% of iterations. Conclusions: This pharmacoeconomic model suggests that therapeutic use of G-CSF should be considered for patients with established neutropenia in order to reduce overall hospital cost. G-CSF treatment may offer substantial potential savings for hospitalised patients with established neutropenia over a wide range of model assumptions. Therapeutic G-CSF use among patients hospitalised for established neutropenia may complement the recommended prophylactic use of these agents for the prevention of neutropenic episodes.