I am working on a web application that maps healthcare expenses by providers in a geographic area. Our prototype was previously using mocked data, and I am now trying to replace it with real data from the 2012 Medicare Provider Utilization data set.
Basically I feel like I have good data in front of me, but I'm not sure how to intelligently interpret it. Below are two (abbreviated) example lines, each of which corresponds with one provider's costs relating to one medical procedure.
avg_Medicare_allowed_amt avg_submitted_chrg_amt avg_Medicare_payment_amt
------------------------------------------------------------------------------
220.4492 | 517.3611 | 176.3583
112.8608 | 1181.0625 | 90.2879
The methodology PDF including more detailed variable descriptions is here, variable definitions starting under heading 5 on page 4.
Each number corresponds with a dollar amount. allowed_amt
is the amount that Medicare expects to pay that provider for that procedure, submitted_chrg
is the amount that a provider actually bills medicare, and payment_amt
is how much medicare actually reimburses. I also have unique IDs for each provider and procedure, as well as general identifying and geographical data.
What I want from this data is to be able to make an informed estimate of how much cost is actually incurred by the patient. In a less confusing world, this would just be (avg submitted charge) - (amount reimbursed)
with the leftovers being passed on to the patient, but the way I understand this, Medicare billing does not really work this way, and the amount billed to Medicare does not really directly relate with the amount that a provider expects to be paid.
Is there any way from this data that I could make an educated estimate about how much a patient could expect to pay for a given procedure?