On July 25, 2024, CMS held a 9-person advisory panel for pricing over 100 new lab tests. I posted cloud archive video and a (imperfect but readable) auto-transcript. Here.
SEE CMS VOTING RESULTS
CMS has now posted the voting results (at the meeting, voting is silent).
At this webpage:
https://www.cms.gov/medicare/payment/fee-schedules/clinical-laboratory-fee-schedule-clfs/clfs-advisory-panel
Scroll down for Panel Recommendations and download the file for 2024.
I'm also offering a Google Sheets cloud version:
Most Votes Unanimous
I haven't tallied the 100-plus voting results, but on inspection, it's clear that many (80%??) are unanimous. This means the panel as a whole agreed with the suggestion of its subcommittee for that code. Now and then there's the odd duck out of line (0426U, line 50, 8 votes to crosswalk 80307, 1 lonely vote to gapfill).
Cologuard: Split Vote 5/4
Some, like the vote for Cologuard new version (0464U, line 773) were split (5 for the current price 1x, 4 for the current price 1.25X which was Exact's request on June 25 at the public comment meeting.)
5/4 Split is Rare!
The Cologuard vote was one of only 4 votes where the voting included a "5," indicated a 5-4 or even more split vote.
AI CORNER
I gave the spreadsheet to Chat GPT 4 and asked it to count how often "gapfill" line had a nonzero number of votes. It said 41.
I asked for a formula I could apply myself in Excel (where I could visually inspect the accuracy.) The formula Chat GPT provided was,
=IF(AND(B1="Gapfill", C1<>0), TRUE, FALSE)
Before I tried that in Excel myself, I asked if Chat GPT itself could apply this formula (as opposed to the verbal instruction I gave it earlier.) It asserted it could, and the formula (in the hands of Chat GPT) returned 41 gapfill.
I asked how often the gapfill vote was "9" and it said 15 times - that is, 15 cases were unanimous for gapfill.
Confirming by Hand
In my own Excel spreadsheet, I applied the formula and then took all the rows where the result was "true" (choice = gapfill "and" N <> 0). I sorted those and made a graphic. This graphic shows how many times - amongst the 41 votes that included at least one "gapfill" - had a vote for gapfill from 1 (single odd duck) to 9 (unanimous).
 |
| Gapfill votes for nonzero |
I count 38 where the panel had a strong consensus (supermajority 9, 8, 7 or only little minority 3, 2, 1 vote for gapfill). Or, there were 24 codes where the panel voted 7 or more (of 9) for gapfill.
Putting together with our earlier results, the panel produced a 5/4 vote four times, only one of which was on a gapfill vote.
