Examining Medicare Data

Examining Medicare Data

By Charlie Carbery

As a company focused on providing cost transparency in health care, we’re always on the lookout for new streams of data. Recently, we’ve been looking closely at Medicare’s 2012 Provider Utilization and Payment Data. This dataset, released in April this year, offered an unprecedented look under the hood of how Medicare payments are actually distributed. At the time, the data made significant waves, including a New York Times piece that called out certain physicians receiving suspiciously high reimbursement amounts.

While the release received a lot of attention on its insight on total reimbursement amounts, the data also includes average reimbursement amounts for each physician. Since this data is broken out by HCPCS code, this data has the potential to be really useful in the cost transparency field. Specifically, by looking at the average reimbursed amounts for each procedure, we may be able to ascertain how individual doctors bill Medicare patients in relation to the fee schedule amounts.



Our team started to evaluate the dataset by looking at the allowed and reimbursed amount averages for specific zip codes and comparing these amounts to the Medicare Physician Fee Schedule (MPFS) allowed amounts. We focused on radiology codes, since Stroll’s initial release will only offer these procedures. Surprisingly, we found that the payment data returned results lower than the MPFS allowed amounts. For example, when looking at the HCPCS code 72195 (MRI Pelvis w/o Contrast) for the zip code 94109 (Stroll HQ!), we found an average allowed amount of $124.59, compared with an MPFS allowed amount of $505.61 for the San Francisco area. As we looked at more and more HCPCS codes across more and more zip codes, we consistently saw this discrepancy.


Professional vs. Technical vs. Total

Confused, I reached out to the Kaiser Family Foundation with this question and got put in touch with one of their associates on Medicare policy. We spent some time going over the data, and may concluded that the problem may be due to how the CMS data is rolling up procedures. In certain cases, doctors billed Medicare for only the Professional Component or only the Technical Component of the procedure. While each procedure requires a Professional and Technical Component, the Medicare data counts each instance as two separate billed procedures. When rolled up an averaged out, this will lead to estimates well below the true cost of a specific procedure.

Since this professional/technical divide is especially prevalent in radiology, this was a particularly important issue for us. In looking at the raw data, we found that each doctor listed either ‘Facility’ or ‘Office’ as their primary place of service.  Comparing these data to the MPFS accepted rates, we noticed had an ‘F’ (Facility) value for place_of_service seemed to exclusively bill for the Professional Component. Doctors with an “O” (Office) value were less predictable, either billing for technical, professional or total cost, often in combination. 


Digging Deeper

I wanted to test our hypotheses concerning the place of service field in the 2012 CMS data:

-Physicians who reported “Facility” as their place of service show results analogous to the Professional (26) component in the Medicare Fee Schedule.

-Physicians who reported “Office” as their place of service show results analogous to the total cost (i.e. Professional + Technical) in the Medicare Fee Schedule.

-The allowed amount represents 100% of the fee schedule amount, while the paid amount represents the amount the CMS reimburses the physician (typically ~80%).

To validate our thoughts, I downloaded the 2012 CMS data and compared it to the Medicare Fee Schedule amounts. The file was exceedingly huge, so I started by only looking at two CPT codes (74177 – CT Abd. And Pelvis w/Contrast, 78264 – Gastric Emptying Study) and only doctors with last names starting with ‘A’:


74177 – CT Abd. And Pelvis w/Contrast, Doctors A
ModifierMPFS Allowed Amount*2012 CMS Average Allowed
* MPFS Allowed Amount takes prices from National Locality (0000000)
78264 – Gastric Emptying Study, Doctors A
ModifierMPFS Allowed Amount*2012 CMS Average Allowed
* MPFS Allowed Amount takes prices from National Locality (0000000)


I wanted to make sure that ‘A’ doctors represented a good sample, so I performed the same exercise on doctors with last names ‘K’ or ‘L’. The results were almost identical:


74177 – CT Abd. And Pelvis w/Contrast, Doctors KJ
ModifierMPFS Allowed Amount*2012 CMS Average Allowed
* MPFS Allowed Amount takes prices from National Locality (0000000)


78264 – Gastric Emptying Study, Doctors KJ
ModifierMPFS Allowed Amount*2012 CMS Average Allowed
* MPFS Allowed Amount takes prices from National Locality (0000000)


Though the Office amounts were between the MPFS total and technical allowed amounts for the 74177 CPT code, it wasn’t clear what these amounts represented. Initially, I suspected that certain physicians were exclusively reporting the Technical cost, while others only reported the total cost. I then took a look at a frequency distribution of the average allowed amounts for 74177. If my theory was correct, the costs should have a bi-modal distribution, clustered around the Technical cost ($234.64) and the total cost ($327.42):

While the costs show a clear clustering in $60-$90, there is not a clear second clustering. As such, I couldn’t conclude that certain doctors consistently reported either Technical or total costs.



Based on this analysis, I have reached the following conclusions:

-The Facility amounts are almost identical to the CMS ’26’ component. The 2012 data results were consistently below the fee schedule amount, but the difference is small enough where we can attribute it to other factors. These factors may include variance in the allowed amount by carrier locality and the increase in allowed amount from 2012 (CMS data) to 2014 (fee schedule amounts).

-The Office amounts were significantly lower than the total allowed amount. Due to the magnitudes of these differences, I think it is reasonable to conclude that the Office amounts are analogous to the total allowed amount. Similarly, the Office amounts were consistently higher than the technical component (TC). My guess is that Office averages represent both physicians reporting total cost, physicians reporting solely the Professional component (26), physicians reporting solely the Facility Component (26) and many more physicians reporting some combination of the three. Unfortunately, these results do not give us the ability to clearly determine Medicare participation status for these physicians.



Leave a Reply

Your email address will not be published. Required fields are marked *

Making healthcare as easy as a Stroll