• A
  • |
  • A
  • Text size

Release of physicians' Medicare claims data

On April 9, 2014, the Centers for Medicare & Medicaid Services (CMS) released physicians' Medicare claims data—including billed charges and total payments—to the public. This unprecedented release of data garnered considerable attention, in part because of the sheer volume of information it contains and the belief that such data will empower patients with a new way to make decisions about their health care. While the AMA is committed to transparency and supports the release of data that can help improve quality of care, it is also crucial that certain safeguards are put in place to ensure that accurate information is presented to the public. The manner in which CMS is broadly releasing physician claims data, without context, can lead to inaccuracies, misinterpretations and false conclusions.

This set of data does not provide a complete picture—there are a number of limitations that need to be considered. CMS released only raw data about physicians' Medicare claims, making it easy for patients, reporters and others to draw inaccurate conclusions about individual physicians.

Below are the key things physicians need to know about their Medicare claims data.

What is included in the data?

CMS released 2012 Medicare Part B claims data for individual physicians, organized by National Provider Identifier (NPI) and Healthcare Common Procedure Coding System (HCPCS) code. The data include the provider's name, address, gender, specialty, the number and type of services provided, the number of unique Medicare patients treated, the average and standard deviation in charges submitted, Medicare allowed amount and the total amount Medicare paid for those services.

How can I look up my data?

The data are posted to the CMS website in 12 separate spreadsheets, which are listed in alphabetical order by last name. The individual Microsoft Excel (.xlsx) files are located at the bottom of the page under the header Microsoft Excel Format. These Microsoft Excel files can be downloaded as compressed folders (.zip), which also include documents that define the data content of each spreadsheet. You can find your specific data by searching for your National Provider Identifier (NPI) or sorting by your last name. CMS has also provided the Medicare claims and payment data in an interactive search tool geared toward patients and in a searchable database. A few media outlets have used information from the CMS data release to create their own databases.

What can I do if I have concerns about the accuracy of my data?

At this point, CMS has denied requests to establish a process for physicians to report inaccuracies in their data. Although the AMA continues to press the agency to provide a feedback mechanism for correcting errors, CMS is also being urged to post an expanded list of problems with the data that users would be forced to click through prior to accessing the data. CMS will not correct the data, but physicians can e-mail their complaints or questions about their data to MedicareProviderData@cms.hhs.gov.

What are the limitations of the data?

1.  The data could contain errors. Physicians don't have a way to correct the information reported.

2.  Care quality can't be assessed from the information reported. The set of data focuses solely on payment and utilization of services and doesn't include explicit information about the quality of care provided.

3.  The reported number of services could be misleading. For instance, residents and other health care professionals under a physician's supervision can file claims under his or her National Provider Identifier, and the data may not properly detail who performed the services. Similarly, many surgeons will appear to have highly inflated utilization rates due to the failure to account for modifiers such as those associated with assistants at surgery and co-management. Services which can be split into technical and professional components may also be over-counted. Conversely, data for some physicians are missing or incomplete because some or all of their care was billed under their group practice's NPI or because CMS excluded services performed for 10 or fewer Medicare beneficiaries to protect patient privacy.

4.  Billed charges and payments aren't the same. CMS will report both the physician's billed charge and the actual amount paid, which is set by the Medicare Physician Fee Schedule. Payments generally are much less than the billed amount.

5.  The data set doesn't represent the physician's whole patient population. The database only includes fee-for-service Medicare patients. Without data for privately-insured patients, the total number of times a physician does a particular procedure cannot be determined. Failure to include Medicare Advantage patients will also skew regional comparisons of physician practices.

6.  Payment amounts vary based on where the service was provided. Medicare pays physicians less for services provided in a hospital outpatient department than for services provided in the physician's office to reflect a difference in the practice costs. But Medicare makes another payment to the facility to cover its practice costs when services are provided in the outpatient department. That means that in reality, the total costs to Medicare and the patient are often higher when a service is provided in a facility setting.

7.  The data set doesn't enable clear comparisons of physicians. Specialty descriptions and practice types aren't very specific, so physicians who appear to have the same specialty could serve very different types of patients and provide a dissimilar mix of services, making some subspecialists appear to be "outliers." 

8.  Important information is missing. The data set doesn't account for patient mix or demographics, and it doesn't point out that a significant share of Medicare payments is used to cover such costs as office overhead, employee salaries, supplies and equipment.

9.  Reimbursement for drugs purchased and administered by physicians is co-mingled with other physician payments. Such reimbursement is not separately identified and there is no indication that these payments are compensation for the price of the drugs themselves, many of which are very expensive and are required to treat such serious conditions as cancer and macular degeneration. The lack of any information on dose size and alternative drugs that would have been billed under Part D rather than Part B also prohibits accurate analysis.

10.  Coding and billing rules differ over time and across regions. Changes to Medicare's coding and billing rules need to be taken into account in any analysis because these rules frequently change over time and across different parts of the country. In addition, some physicians, such as pathologists, may bill as either physicians or suppliers, but only claims billed by physicians are included in the data.

What should I tell my patients or the media?

If your practice receives inquiries from patients or reporters about charges or payments, you may want to give a response that provides context for the data. While circumstances will vary on a case-by-case basis, here are some of the most important points to clarify. You can provide this fact sheet to your patients, or download a customizable version. It includes information on some of the primary limitations (which you can modify) that patients should keep in mind when reviewing the data.

Medicare payments aren't the physician's personal income. These payments are practice revenues that must cover business expenses, including pay and benefits for practice staff, billing and other professional services, office rent, utilities, professional liability insurance, medical equipment and supplies.

The average breakdown, according to CMS' Medicare Economic Index (MEI), places about half of payments into one of these other expense categories.

Depending on the physician's specialty, location, practice type, patient mix and other factors, the percentage physicians actually take home as their personal income can be much less than the MEI estimate.

In addition, the Medicare claims data released include reimbursement for drugs the physician administered, which is not figured into the MEI payment breakdown. CMS fails to explain in the data release that doctors purchase drugs and that Medicare payments simply cover the price of these services—many of which are very expensive and are required to treat such serious conditions as cancer and macular degeneration.

The majority of physicians don't receive noteworthy Medicare payments. Despite some attention-seeking news headlines that focus on sizeable pay-outs to physicians, the average physician doesn't generate that much revenue from Medicare payments. In fact, 75 percent of physicians and other health care professionals receive less than $85,000 per year in Medicare payments as reported in the CMS data file. The median payment amount is scarcely more than $30,000.

Attribution issues could distort the data. Because the data are tied to the National Provider Identifier (NPI) under which the services were billed, some physicians who provide Medicare services may not be included in the data at all because their claims were filed using a group NPI. Other physicians may be included in the data release, but the services attributed to them may not reflect the care they actually gave because some of those services were reported using their group NPI.

Similarly, some physicians may appear to have provided fewer services than they did because CMS excluded services that were performed for 10 or fewer Medicare beneficiaries to protect patient privacy. For instance, the sets of data for some surgeons report only their evaluation and management codes, not the procedures.

On the other hand, some physicians may appear to have a higher number of services because their data are attributed with services and payments provided by other clinicians in their practices, including residents and other health care professionals. Even where the data are complete, the way the data are displayed may lead to misinterpretations, such as the appearance that surgeons working with an assistant at surgery have done an inflated number of procedures.

The data set doesn't include quality. The claims data focus on payment amounts and number of services provided but don't provide explicit information about quality, such as outcomes, necessity and comparison to services provided to non-Medicare patients.

The data set shows only a portion of the full picture. Patients and others shouldn't make an assessment of their physicians based on the Medicare data alone because it only provides a subset of information. For instance, claims data aren't included for patients who are covered by Medicare Advantage plans, private insurance or Medicaid.

What is the AMA's position?

In a May 15, 2014, letter to CMS, the AMA stated its agreement with the Administration that Medicare data will be a key ingredient in any successful effort to move toward a value-based Medicare payment and delivery system. We recognize that access to Medicare data, as well as other payer information, has the potential to improve the quality of care while reducing costs. However, we also believe that as demonstrated in the recent data release by CMS, raw Medicare claims data can be potentially harmful.