The physician charges were actually a minimal part of the charges. $2100 for the neurosurgeon, $3500 for the radiation oncologist, $380 for the radiologist. The biggest charges were for the actual use of the machine. I had LINAC treatment with 4 passes and each pass was broken down on the bill and averaged $14500 per pass. The charge for use of the head frame itself was also on the bill.
Back to the original discussion, I was able to obtain the actual breakdown for the GK costs here in Ontario, Canada that are charged to our health plan (which is run by the province):
Gamma knife treatment day $2,121.16 includes MRI with Gadolium, CT Scan, Treatment planning and Gamma Knife treatment. Total AN related costs $5,250.60 includes all diagnosis related testing, including audiology and MRI's. Follow up MRI's, surgery and radiology consultation over a two year period and the treatment day itself.
Assuming that physician costs in N. America would be (relatively) comparable, it seems that the major difference is the cost for using the equipment. I suspect that the reason why costs for equipment usage in the States is so high, as everyone mentioned, is that the cost for capital cost recovery is factored in. i.e. different models for calculating the cost: In the States someone (a hospital, or clinic for example) makes the initial investment and pays for the capital cost for installation/initial commissioning etc., and then this cost is recovered through the span of many years, from the users.
In the other model, the cost is covered upfront (by taxpayers money), so there is no need to factor in the cost for capital cost recovery into the actual treatment. Plus the state-run health insurance plans can put a cap on what they pay for usage.
Recognizing that capital cost for such equipment must be covered, it is then a matter of paying upfront (by the taxpayers), vs. paying afterwards (by the users).
I don't want to stir any discussion about which model is better; the point is that cost must be recovered somehow, and there are different models for doing this. I am just trying to find an explanation for the big differences reported.
Marianna