LONG BEACH, Calif--There is no controversy in medical oncology that G-CSF decreases the risk of infection associated with neutropenia in patients receiving cytotoxic chemotherapy. But does it save money? To answer that question, John Glaspy, MD, MPH, and his colleagues at UCLA School of Medicine used a decision analysis model to determine the economic impact of G-CSF given as supportive care.
LONG BEACH, Calif--There is no controversy in medical oncologythat G-CSF decreases the risk of infection associated with neutropeniain patients receiving cytotoxic chemotherapy. But does it savemoney? To answer that question, John Glaspy, MD, MPH, and hiscolleagues at UCLA School of Medicine used a decision analysismodel to determine the economic impact of G-CSF given as supportivecare.
The economic study was based on data from the licensing trialfor G-CSF (filgrastim, Neupogen). This trial involved patientswith small-cell lung cancer who were randomized to receive thesame chemotherapy regimen with either placebo or G-CSF, with theincidence of febrile neutropenia as the end point.
The researchers collected hospital bills from patients in thestudy at three sites--UCLA, Duke, and Pittsburgh. To calculatedrug cost, the mean duration of G-CSF therapy was multiplied bythe average wholesale price of the agent at that time, weightedfor the size of the patient, since G-CSF is prescribed on a perkilogram basis.
"We added on some extra money for the cost of doing the extrablood counts and of buying the needles and syringes, and cameup with a proposed cost for G-CSF therapy of $2,302 per patientper cycle of chemotherapy administered," Dr. Glaspy saidat the 1995 Quality of Life symposium sponsored by St. Mary MedicalCenter, Cancer Care Center.
To determine the cost of an episode of febrile neutropenia requiringhospitalization, the researchers relied on the Medicare database,which showed that nationally, these patients stay in the hospitalabout 8 days, with hospital charges of around $1,000 a day, fora total of about $8,000.
The G-CSF patients had fewer hospitalizations on the first cyclethan did the placebo group, and in addition, those admissionsthat did occur lasted about half as long, resulting in about a50% reduction in hospital charges.
Then, to "cut to the chase," as Dr. Glaspy put it, theinvestigators plugged into the model the amount spent per patientper cycle for G-CSF and the amount charged for the care of febrileneutropenia in the G-CSF and placebo groups. This showed thathospitals were charging about one tenth as much per patient percycle for patients getting G-CSF as for those on placebo, fora savings of about $1,200 per chemotherapy cycle.
Even though all patients, whether they developed febrile neutropeniaor not, were charged $2,302 for G-CSF, "it is still cheaperfor the health-care system to place all patients on G-CSF thanon placebo."
Unfortunately, Dr. Glaspy said, "you can't really use chargesas an accurate reflection of what something costs the health-caresystem any more." To convert from a charge model to a costmodel, the researchers put all the charge data through each individualhospital's cost-to-charge ratio.
"And when we did that, in this study, instead of saving $1,250per cycle, G-CSF ended up adding $140 to the cost of each cycleof chemotherapy," he said.
Dr. Glaspy and his colleagues then calculated the "break-evenpoint" for the use of G-CSF as supportive care. They foundthat in most health-care systems, this point will be reached whenbetween five and 10 doses of the 5 mcg/kg dose of G-CSF are beinggiven to prevent one patient day of hospitalization.
As the patient's risk of febrile neutropenia without G-CSF fallsbelow 20%, the number of doses of G-CSF required to offset onehospital day begins to rise rapidly, he said, so that small errorsin febrile neutropenia risk assessment can have large implicationsin terms of cost offsets. At higher risk (more than 30%), theeffect on cost offsets is much less.
Thus, for the oncologist in clinical practice, the responsibleuse of G-CSF involves identification of the high-risk patientsin whom it is most cost effective. When the risk is 30%, he said,G-CSF is clearly cost effective. But even if the physician's estimatesare off and growth factor is given to some patients with onlya 10% risk, "it is still better than waiting until the patientdevelops fever and using G-CSF as treatment," he said.
Growth factors used in conjunction with antibiotics to treat patientsonce they develop febrile neutropenia has been the focus of arandomized Australian trial, he said. This study showed a clearimpact of G-CSF on the duration of fever, but the impact on durationof hospitalization, which is the cost end point, did not achievestatistical significance.
"It's possible that a lower dose would have been just aseffective, and then it would be cost effective, but we just don'thave that study," he said. At UCLA, Dr. Glaspy stressed,patients are not given growth factors for uncomplicated febrileneutropenia "because we don't consider it to be a cost-effectiveapplication."
A charge-based study published in association with the use ofGM-CSF given to patients receiving high-dose chemotherapy/autologousbone marrow transplantation (ABMT) showed that the charges arelower when GM-CSF is given after bone marrow infusion.
"GM-CSF clearly more than paid for itself. It saved $22,700in charges, which, even if that converts to $11,000 in true costs,is still a reasonable use of medical resources," Dr. Glaspysaid. But this, of course, does not answer the question of whetherhigh-dose chemotherapy/ABMT should be done at all in certain patients,he said. "We still don't know whether it saves lives, anduntil we do, we really can't talk about the cost effectivenessof growth factors in this setting."
If the clinical efficacy of ABMT is proven, Dr. Glaspy said, theend points of the cost analysis of growth factor use will notbe money saved but rather survival (costs per year of life saved).
In his presentation (see above), Dr. John Glaspy discussed possiblechanges in scheduling and dosing that could lead to more cost-effectiveuse of growth factors.
"Some clinical trials suggest that growth factors can bestarted a day or two later and stopped a day or two sooner withoutcompromising efficacy. If that's true with G-CSF, then we willbe able to decrease the number of doses per chemotherapy cycle,making it much more cost effective," he said.
He emphasized that clinical trials designed to answer these questions,although they may not seem too interesting, "are criticallyimportant to making our practice more cost effective and makingthe drug available to more people."
He also speculated that the optimal dosing of G-CSF has not reallybeen determined. "We still don't know today how importantit is to dose CSFs in adults based on weight, and use of smallerdoses may be just as efficacious," he said.