Associated with a great deal of fanfare, particularly in the popular press, on April 9 Medicare released data about how much it paid physicians and other suppliers in 2012. Popular reporting quickly focused on the “top billers” to Medicare for health services, including setting up “look up your own doctor” type of search websites. Who doesn’t want to estimate how much their doctor is making, and whether they are gouging or pricing services reasonably? Consumer groups tended to welcome the availability of the data. The AMA seemed to respond both ways at the same time, praising “transparency” while cautioning against misuse and misinterpretation.
As usual, the truth is in somewhere between. My initial review of data directly related to mental health services reveals data that, with certainty, are missing (more on this later) and that must be inaccurate (either due to provider inaccuracy of claims or CMS inaccuracy in the database itself). The old computer science term GIGO (garbage in garbage out) comes to mind. Even a cursory review finds that the data are no better in quality than the very preliminary stages of a graduate student project, much less adequate to serve as an “empirically supported” basis for understanding health care financing and determining public policy. Nevertheless, there is some potential for useful information that may add to our understanding of the price and delivery of health services. But caution in use and interpretation of the database appears wholly warranted.
With that in mind, here are some preliminary views specifically of mental health services as revealed by the data.
How much and what type of mental health services are being provided to Medicare beneficiaries?
The following table shows an initial view of the Top 10 services provided in 2012.
It appears that, contrary to popular belief and data that show greater growth in medication treatment relative to psychotherapy, the most frequent mental health service provided to Medicare beneficiaries is 45-50 minutes of psychotherapy, with about 4.16 million sessions provided in 2012. However, a quick further look at the information shows that medication management visits in both Office (3.47 million visits) and Facility (e.g., hospital and similar) settings (1.77 million visits) combined (5.24 million visits) outstripped 45-50 minute psychotherapy sessions provided in Office and [based on the description of the procedure code] those those occurring in Facility settings combined (4.9 million sessions).
The numbers in the table above raise the first evidence that the raw data provided by CMS are flawed in some respects. The reader will note that I included the “Location” where services were provided. For those who may be unfamiliar with this, claims submitted to Medicare must show the location where services are provided, as Medicare reimburses different amounts if the patient is seen, for example, in the doctor’s office or in the hospital.
Note line 7 shows the service described as “45-50 minutes therapy in facility” and the “Location” nevertheless shows “Office.” Similarly, line 7 describes the service as “20-30 minutes therapy, facility”, and the “Location” is shown to be “Office.” These data were obtained directly from the CMS database and have been checked and rechecked, and may be checked by others, as the database certainly is “open source.” What this means, however, is that either somewhat more than 1.4 million service items were miscoded on claims, or CMS has incorrect data in its database, or some combination of both. Of course if claims are miscoded, this potentially puts those doctors and other providers at risk of “fraud” charges by CMS, with potential sanctions such as taking the payments back or worse. If not miscoded, it puts those of us who would like to rely on accurate data with huge uncertainties.
With the above problem in the data noted, here is a modified version of the above table showing the 10 most common mental health procedures for Medicare beneficiaries in 2012, corrected for services with consistency between the type of service and location where the service should have been delivered.
Ten Most Common Mental Health Services for Medicare Beneficiaries in 2012
In 2012, 5.52 million 45-50 minute psychotherapy visits were provided (including .72 million with additional evaluation and management [E&M]) in offices and facilities around the country, and 5.24 million medication management visits were provided. Psychotherapy services do appear to have been more common, albeit with only a slight edge, than medication management-only visits. Considering 45-50 minute therapy visits only with no E&M gives the edge to medication visits being more common than psychotherapy, at 5.24 million to 4.8 million visits.
How much are doctors and other mental health clinicians charging?
Here are the minimum, average and maximum charges for the ten most common services provided to Medicare beneficiaries in 2012.
On average, in 2012 the charge was $227 for an initial evaluation interview in the office and $131 for an individual therapy session of 45-50 minutes, and $97 for a medication-related visit in the office. These are national averages across all practitioners providing these services, including MD, PhD, MSW and potentially others such as advanced nurse practitioners. Analysis of differing charges among mental health service providers is possible given the available data.
Given the figures for average charges submitted, some if not all of the maximum charges reported seem improbable. This adds to the uncertainty of the database. Are these charges miscoded by the health professional and/or biller? Are these misrepresented in the database? Are these true actual charges submitted? If the latter, it becomes apparent how readily CMS is able to spot outliers. If errors by the practitioner or some method of data entry into the CMS database, it becomes apparent how readily a charge and practitioner might be inaccurately flagged as “fraudulent.”
How much is Medicare reimbursing relative to amounts charged?
Of course popular reports tended to leave the impression that if a doctor receives $X in reimbursement from Medicare, that reflects net income. Anyone who runs a business immediately sees the error there. Beyond the ability to look at charges health care professionals are submitting to Medicare, the database allows the opportunity to understand amounts Medicare actually reimburses. That should be an interesting lesson. Here is a look at actual reimbursement relative to amounts charged for the ten most common mental health services.
Actual reimbursement from Medicare for the ten most common services provided ranges from 52% below the billed amount at best (for an initial diagnostic interview in the office), to 73% below the average amount charged at the worst, for medication management in the hospital or other facility. Either thousands of health care providers are overcharging, or Medicare is substantially lowballing reimbursement for common mental health services.
An important analysis that awaits is whether the relative gap in reimbursement for mental health services relative to usual charges submitted by mental health care professionals is similar to that seen in other health care specialties and services. If relative reimbursement is substantially below that seen in other health care services, the question follows why the relative bias in reimbursement for mental health care in Medicare? The database does set the stage to raise and potentially answer these questions.
And the final evidence that the database is missing data? Despite a thorough search in a number of ways, data from my practice in 2012 are nowhere to be found in the database. How such an omission could occur, and the extent to which other professionals may have been omitted from — or incorrectly included in — the database are exactly the questions I posed to CMS. We will see what they say.