The Strange World of PAMA Neuro Test Pricing

Header.   For 24 neuro codes, most were PLA codes with no PAMA data.   For codes with PAMA data (e.g.  5,000 payments for generic ABeta 1-40 at various prices), median prices rose significantly (from $118 to $175).   Interestingly, raw PAMA data shows that dozens of labs often got over $200, even over $300, for these codes despite later CLFS prices of $118.

##

I asked Chat GPT to look at CMS's current clinical lab fee schedule and pull out full text codes for dementia-related codes (A-Beta, Tau 217, etc.)   It found 24 rows of data.

I then asked Chat GPT to search the CMS PAMA Preliminary Median file and wherever it found a median (some are blank), to insert that in the table.  I then had it make ratios of "PAMA Median over Current CLFS" and plot the ratios (where possible) on a vetical bar chart.   Note the bar chart has bins 1 to 24, and the table in Excel also has ordinal rows 1 to 24 (right column G).  

Find the Excel data in Google Sheets here. Note there are several tabs.

This shows the full table but won't be legible (see Google Sheets)  "Down" price in pink box; "up" prices in green box. Again, quite a few are PLA codes with no PAMA data so those rows are ineligible for an up/down color.

not legible, see google sheets

Whereas many prices fall under PAMA, here in Neuro World, only one of 24 neuro codes fell and not by too  much (0035U prion, from $540 to $432, row 1.)  

Six of the codes rose between 1.2X and 1.8X, including 0443U, 82233, 82234, 83884,  84393, 84394.  Most of these rising prices were generic CPT codes like Beta amyloid 1-40 and Beta amyloid 1-42.  These were reported by dozens of TINS (101 TINS in the case of 84393 Tau phospho eg 217. 

Often, PLA codes had no reported PAMA data (blanks in the bar chart).     

I had Chat GPT go to the "raw PAMA data" (each reported code, price to the penny, and number of services) and make a bar chart of the price distribution for 82233 and 82234, which are Abeta 1-40, and 1-42.   These had CLFS prices of $118 (hard-won, since CMS attempted to price first at $18).   Prices go UP from $118 to $175 for each code (which had about 20 reporting labs). 

  

The CLFS price $118 is not a peak, and a lot of prices came in over $200 dollars.  But this makes sense because these codes were in gapfill (no national price) during the PAMA survey period of 1H2025.

Note that CMS calculates medians, so it doesn't matter how far to the left or right the data fall, only what is the 50th percentile price.   You could push the 82233 spike at $260 to $360 or $460 and it wouldn't affect the median of $175.   

It's interesting the distribution below is so diffuse and non-Gaussian. 

Another tidbit - the codes, generally for Alzheimer testing, WERE getting paid.  But the data represent for each, about 5,000 services, not many considering there may be over 10M dementia patients.   The 5000 services are a circa $1m market.

##

Nerd note.  How-to.  For 82233, I went to raw PAMA data and pulled the rows for 82233, about 500 rows, into its own Excel.  I uploaded that simple excel to Chat GPT with this instruction:

• "Please bin these PAMA prices and then make a bar chart where the number of ser vices reported at a price is the bar height.    We have done this before.   Bin the prices in $10 increments from $0 to $300.   The height of each bar will be the aggregated number of services for that bin.'