Analysis of CMS Release of Medicare Provider Payments Begins!

Both aggregate and detailed data for Kentucky and Louisville available for download below.

Its out there!
The release last week by the Centers for Medicare and Medicaid Services (CMS) of services by and payments to physicians and other non-hospital providers reverberated as expected throughout the nation. Analysis of the massive database is, and will continue to expose the complexity, quirks, inequities, fraud, and sometimes just plain bizarreness in our current health care system. Some large media outlets such as the New York Times, the Wall Street Journal, and the Washington Post were allowed to organize and work with the data before it was released to the general public. Both these sites and perhaps others allow the public to look up individual providers, but comparisons of individuals or groups is cumbersome to impossible. Furthermore, neither of these two portals to the data includes all non-physician providers. In any case, a host of individual providers or professional groups are now scrambling to try to explain why they stick out like sore thumbs in terms of frequency of services, charges, or payments from Medicare.

Eye doctors (and others) under the magnifying glass.
For example, ophthalmologists point out that the reason they receive among the highest payments is that they frequently inject an extremely expensive drug into the eyeball. While this may be rational and honest defense, it is not a reasonable one if, as is reported, it is true that a spectacularly cheap alternative generic drug works just as well. Thus, the issue of how much money a physician or other provider is entitled to make off a drug they chose to administer themselves is certain to enter public debate.

It isn’t easy to use the data.
There are many reasons why analysis and use of the information presents challenges for public consumption or professional policy analysis alike. For one thing, the number of records itself is huge. The main Public Use File (PUF) itself has over 9 million individual records for almost one million different providers. Each provider can have multiple entries– one for every specific service provided to more than 10 beneficiaries. Thus, a radiologist for whom an x-ray for every body part may be a different billable service may perform and charge for more than 100 or more different categories of service, while a primary care physician working in their office may have only a handful of different service entries.

Each individual record contains up to 27 different items including names, addresses, number of times each of several hundred different kinds was billed, and various financial and statistical data for each type of service provided. The charge and payment entries include: (1) the charges to a patient for each service, (2) the amount Medicare approves for billing and thus what the patient is responsible for, and (3) the amount Medicare itself pays for the procedure. In this main database, the total aggregate amounts charged, approved, or paid are not provided. These latter are provided in a separate database in which each provider appears only once.

These latter aggregate amounts (which are the easiest to understand and which have been most discussed in the media) are presumably calculated from the data in the main file but I do not yet understand how this is done– nor can I exclude that CMS is using other information available to it to make those calculations.

CMS makes the entire detailed database downloadable, but warns the user (correctly) that the usual home- or small-business computer will choke and crash on the size alone. CMS tries to help by breaking the main file up into smaller pieces based on letter of the alphabet, but even these are huge containing close to 1 million entries. Breaking up the main file makes it much more cumbersome to compare different providers or groups. Professional users comfortable with using statistical or database programs will have an easier time. I am not yet one of them!

There is no more complex system in the world than medical reimbursement policy!
I cannot understate how complicated Medicare’s payment system is! The reimbursement for even a single service can vary depending on whether it was provided in a doctor’s office or a hospital outpatient setting, the professional degree of the provider, how many times the service was performed in a single day, what part of the country the service was provided, or a variety of other reasons. Even what constitutes a single service is misleading. For example, each mile an ambulance drives is considered a single “service.” When a physician administers a drug in their office, the number of services may be the number of milligrams in the dose rather than the number of times the drug is given, and so forth.

Links to Detailed and Aggregate databases for Kentucky.
I am trying to understand the facts, methods, and policies that underlie these Provider Payment files. Nonetheless, I look forward to doing so because it is the best way to understand how our system works. In the meantime, I make available below manageable chunks of both the main detailed Provider Payment file and the Aggregate files for all providers in Kentucky, and in Louisville. Although the potential user will need some familiarity with sorting and searching Excel spreadsheets, they will be able to make comparisons of different kinds of providers, different cities; and for individual providers, a breakdown of the frequency and cost of individual services. Thus was I able to confirm my suspicion that the principal reason rheumatologists are among those receiving the highest reimbursements from Medicare is in largest measure due to the expensive drugs some (but not all) administer in their offices.

The data is incomplete, non-comparable, and confusing.
It is also important to recognize that as valuable as this payment information is, there are major shortcomings that limit its usefulness and applicability. The American Medical Association and other professional organizations have seized upon these shortcomings. Only payments through Part B of Medicare’s traditional fee-for-service system are included. Patients in Medicare managed care, Medicaid, or those with other or no insurance are not included. (This is the price we pay for our fragmented system of medical care.) Therefore these files will not generally reflect the patient mix of a provider. Payments for durable medical equipment are not included. Services for which 10 or fewer were performed by an individual provider are excluded from all calculations. The data are not corrected for severity of illness nor do they provide insights into quality or safety. No statistical outliers were removed. For demonstration projects or other purposes, services provided by more than one provider may be billed under a single provider number. The list goes on and will get longer.

It is still a good thing.
Nevertheless, examining the variety and extremes apparent in the data will be extraordinarily educational for both the public and policymakers. We will be treated to examples of exemplary and atrocious medical care; honest providers and fraudulent ones; and efficient providers or wasteful ones. Make no mistake, policy and practices will change because of the release of this information– mostly for the better.

To allow my Kentucky readers to help in this venture, I attach links to Excel files containing all the data for Kentucky, and a separate (and smaller) file for Louisville. These contain physician and non-physician providers. Let me know how you find them and share your insights with me. It is easy for me to extract data from the main file for individual states, cities, provider types, and the like. If you are interested in expanding this first look yourself, contact me and, if feasible, I will provide the files. The methods and definitions used for the main Public Use Files are presented by CMS here. The methods and definitions for the aggregate files are contained within the Excel files themselves. Contact me for clarification, or for that matter help me clarify for myself!

Peter Hasselbacher, MD
President, KHPI
Emeritus Professor of Medicine, UofL
April 13, 2014

Downloadable Excel Files:

Main Public Use File for Kentucky Providers.  (29 MB Excel file)
Aggregate Provider Payment File for Kentucky. (2.5 MB Excel file)

Main Public Use File for Louisville Providers.  (9.4 MB Excel file)
Aggregate Provider Payment File for Louisville.  (0.6 MB Excel file)

Methodology & Definitions for Detailed PUF Files. (0.3 MB .pdf)

Rheumatology Providers Only:
Just for fun and future analysis, here is the data for all 4053 Rheumatology Providers in the aggregate database. (0.8 MB, Excel .xlsx file)

Want your specialty or city?  Just ask.