CMS has been releasing physician and laboratory payments by year and by CPT code dating back to 2014. The CY2019 data has just been released.
The 3 GB 10M line database can be searched online by using filters and queries in a cloud database.
For the lab industry:
I've put a condensed version of 47MB and 888,000 lines in the cloud, including all 80,000 series (lab & pathology) CPT codes. Excel can usually open 1M lines, so this should open in Excel. Across the 888,000 lines of charges for lab & pathology codes, total allowed charges were $7,374,664,794, or $7.4B.
How to Find the CMS Data for CY2019
- I've made a simple a five-minute video on how to do this - YouTube here.
- Finding the data:
- The home page for all data resources at CMS is here.
- Interesting data pages include "Medicare Fee for Service Parts A&B" and "Part B Carrier Summary Data File" and "Part B National Summary Data File."
- For the new 3GB of data, the data page for Physicians and Other Practitioners is here.
- Data is by geography (e.g. all uses of a CPT code in a state), by provider (with detailed summary data by provider such as ages and races of patients seen), and the most granular, "by provider and service." That's where the new 3GB data of physician & CPT usage is found.
- Cut to the chase: You'll get to this webpage:
- Where you want to click on the white box, left side, for VIEW DATA.
- How you see the data in the cloud. Between last spring and now, CMS has revised the data layout and I think it is worse, but the data is still there. Note there is a small drop down box in the upper left where you can select year (2019, 2018, 2017, etc).
- Link for 2019 data.
- Note that the data view, before any filtering, has 10M rows.
- You will want to filter it by "managing columns" and "filtering" (e.g. for only code 81445), and then use the EXPORT button to export data to your hard drive in Excel. My simple video guide here.
I say the data set is worse because, for example, you used to be able to filter CODE by "Starts with" or "range" and now, you can only filter it by standard number equations like "equal to" or "less than." The visual user interface also used to be better, and the column names used to be reader-friendly.
- I believe that the genetic test fraud situation called "Operation Double Helix" was at its peak in 2018 and 2019 - press here.
- For example, a person with an interest in doing so, could look up every lab that was named publicly by DOJ in "Operation Double Helix" and see how much each lab was paid, and by what MACs, and for exactly which CPT codes. (See more at the end of this blog.)
- Here I show a table (click to enlarge) of payments for code 81445, tumor, 5-50 genes, with 7,325 services going to 24 labs for a total of $4.2M in 2019. The point is: You can do this kind of data for any CPT code, or range of codes. You could look up all the urologists who used a particular prostate surgery code. And so on.
- DOWNLOAD ALL LAB PAYMENTS.
- If you'd like to ask any question about lab codes, labs, and payments, but you don't want to mess with the CMS cloud data, I've tried to help.
- I've put in the cloud a shortened, downloadable 47MB, 888,000 line file of all physicians and labs who used any 80,000 series CPT codes in CY2019. >> Here. <<
- For a couple years, I've tracked CPT code 81408 (rare long genes, $2000). This code was virtually unknown before 2018, and use exploded in 2018 and 2019.
- Intense users of this code were disproportionately mentioned in Operation Double Helix, and almost all use in 2019 was in two MACs, Novitas and First Coast. These MACs had "no edits" on this $2000 code, meaning it would autopay if billed and in units up to 2 per day. Blog here.
- 81408 came out of nowhere with lightning speed to be the top billing genetic code in Medicare in 2019 ($283,634,684). I mean, wow - an unbelievable story.
- There's essentially no billing of code 81408 by familiar national genetic labs like Invitae, Ambry, Labcorp, Quest, Bioreference, and etc. Click to enlarge.
- Medicare Part B pending by state for 81408 in CY2019 was as follows:
- That's 81408. For data on CY2019 users of code 81479, see the next blog.