Discoveries in Health Policy: The Distribution of US Payer Price PAMA Data Is Bizarre

Saturday, October 24, 2026

The Distribution of US Payer Price PAMA Data Is Bizarre

Here's an analysis run mostly by AI, based on my ability to "vibe code" instructions of what AI had to do with raw PAMA Excel data.

Data shows the bar charts of the private payer price distribution for about 15 codes. The distributions are strange and not many are "Guassian" or "normal." While some are tightly distributed and others spread evenly around the price range.

See a prior study in 81455, 81456 here. See a prior study of CPT neuro code data here.

I went to the giant raw data files for PAMA price reporting and created smaller spreadsheets for PLA code data and for 815nn (MAAA code) data. I asked Chat GPT to consider only codes with more than 200 rows of data.

This gave 4 just charts for PLA data (of 500 PLA codes in PAMA) and 11 charts for 815nn data.

Note that CMS does not usually provide sole-source lab data, so I'm not sure why there seem to be data for 81528 Cologuard but not others (e.g. Oncotype Dx Breast 81519, Veracyte AFIRMA 81546, etc). With distributed kits like Biofire you'd expect to see data from many TINS, though, and CMS posts the data.

Data jumps from 81539 to 81596:

Data jumps from 81517 to 81528:

Anyway, so far, we've identified codes in the PLA series or 815/MAAA series with > 200 rows of PAMA data.

Then I instructed Chat GPT to take the price data for each code that qualified, and bin the price data in 8 bins (equal size, but I let AI figure it out) as vertical bar charts per price bin.

See all the results in ZIP file.

0359U IsoPSA