I expect this question to be well beneath the audience, but am hoping to get set out in the right direction to solve a problem. I am responsible for developing a tool that analyses health claims data over the course of a given contract for a specific provider and tells the user if/when prices changed for given procedures, and by what percent.

What sort of analytical methods/concepts should I be considering and which metrics are the most meaningful/useful for my purposes?

Edit for details: Each time a member goes to a provider we get an itemized invoice that bills any number of procedure codes; the data we are looking at is the aggregate of all those invoices over the 12 month period that precedes our query. For the most part the billed amount for a given procedure code will be the same with the exception of errors in claims processing. This feature of the data should make it easy to spot when changes in price occur since if code c0120 has been being billed at 294.00 over a 9 month period then jumps to 311.00 for the last 3 months of our analysis it tells us that a 6% price increase occurred. This feature of the data also means that once we eliminate any strange deviations from the mode which are likely explained by human error in claims processing, nearly any sustaining change should be viewed as a price change.

As you can imagine, the amount of data we have depends on the kind of facility (i.e. is it a Hospital with 50,000,000 in billed charges or a Mom & Pop shop with 1,500,000 in billed charges), but for the most part the contracts we're interested in looking at closely are going to be ones where the procedure codes will show up hundreds if not thousands of times in a given 12 month period.

  • 1
    $\begingroup$ Could you give a little more detail on your data? E.g. how much data you'll have, what time spans you're interested in, what the required thresholds for price changes might be? $\endgroup$ – naught101 Sep 22 '15 at 1:20
  • 1
    $\begingroup$ @naught101, sure thing. See above. $\endgroup$ – Acemanhattan Sep 22 '15 at 1:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.