Tuesday, October 26, 2010

Medicare Claims Records

WSJ with an interesting story about their use of Medicare claims data to identify unusual or odd patterns of payment by Medicare for physician services. In this case, they are saying that a particular physician billed Medicare for around $2 Million in one year. The file they analyzed, known as the Carrier Analytic File, records payment from Part B (physician services) of Medicare to physicians. Such files contain a 5% random sample of all such files, and allow one to get a sense of the trends in Medicare's payment for such care across the nation.

Probably 30 of the papers I have published are based on Medicare claims data such as what WSJ describes. If you want to describe the experience of the full Medicare program you would not only need to use the file noted in this story (Carrier Analytic File, Part B) but also inpatient hospital files, outpatient, skilled nursing facility (SNF), durable medical equipment (DME), home health and hospice files. Each file is in a separate database (though there are summary files) and it takes quite a lot of technical expertise to analyze such data. An ongoing project of mine currently has a data request under review at Center for Medicare and Medicaid Services to get the series of files I note above so that we can describe the full Medicare-financed health profile of a group of patients. The paper notes the cost of the files they wanted to be expensive ($100,000).....the cost of the data I am currently seeking is roughly $45,000.

The upshot of the WSJ article is that you can use such files to find odd patterns of care that may constitute fraud. Perhaps that is true. However, there are many other uses for such data, and it certainly would be a step in the right direction to expand the access (by reducing the cost) to Medicare claims records to researchers/analysts so long as safeguards are in place to protect the identity of patients. Identity of providers is also masked, at least in all the Medicare claims files I have worked with.

As a researcher and policy analyst, count me as a yes vote for easier access to Medicare data.


  1. Really interesting article. Thanks for the link. I agree that the WSJ analysis is simplistic but for non-researchers, not so bad either. (The carrier file should get the majority of payments made to an individual physician.) Incidentally, while you haven't gotten it, the NPI is part of the "research identifiable files" (at least on request). One would have to show justification to get it, and the WSJ's justification would probably not have gone over well.... http://www.resdac.org/ddvib/Files/ORD_NPI.htm

  2. We are still dealing with RESDAC, so I will look into NPI. It is a bit odd to get a federal research grant that then pays another portion of the federal govt for data.