Wednesday, June 10, 2020

Very Brief Blog: CMS Releases MAC Gapfill Proposed Pricing for CY2020

Last year, CMS had to price some 100 new CPT codes (mostly PLA codes) and sent about 37 to the "Gapfill" pricing method.  Under "Gapfill" pricing, CMS doesn't set a price, but allows its MACs to set proposed prices for the new lab tests in the first half of the next year.

Gapfill Pricing Released - Comment to August 10

On June 9, 2020, CMS released the MAC gapfill prices for CY2020 for public comment.    Comment is open until August 10, 2020.  CMS will circulate public comments to its MACs in August, and final pricing will be released in the fall.

The public meeting home page is here.   Near the bottom look for the link, "2020 Gapfill Preliminary Recommendations."  This connects to a zip file here.   

I've also put my own cloud copy of the Excel file in the cloud here.  My copy has a few extra pages such as a comparison of Noridian and Palmetto MolDx pricing, and an analysis of Minimum, Maximum, Std Dev of proposed prices.

Brief Analysis

Almost All PLA:  There are 37 gapfill codes on the spreadsheet (lines 8 to 44 inclusive).  Only one is a Category 1 CPT code (81542, prostate MAAA test).   The rest are PLA codes in the range 0046U to 0138U.  

MolDx Underprices Novitas. The biggest surprise is that the MACs differed from each other substantially on about 30 of 37 codes, with MolDx pricing lower than Novitas/FCSO on about 20 of those codes.  

Biofire Results:  In a rare move in 2019, Biofire created several new PLA codes for its FilmArray panel tests (e.g. 20, 21, 22 targets, codes 0097U-0100U) and asked they be priced less than the reference code (87633, 12-25 targets, $416).   They got their wish, with 0097U priced at $363, 0098U-100U priced between $247 and $298.  

The highest priced test is rapid sequencing of whole genome (e.g. in a couple days), at $7582 (0094U).  All MACs had around the same price.

MAC Discrepancies and Ranges

Note that each code has a Rationale for Pricing on a separate tab from the pricing information itself.

In the cloud spreadsheet, I ran extra columns of data for the MAC's minimum, maximum prices and standard deviation, as well as standard deviation normalized against the median price.  

MAC prices varied quite a bit, with an average normalized standard deviation of 32%.  

Only a small handful of codes had only a 0-5% standard deviation in proposed prices.  Several codes had >100% standard deviation in proposed prices (e.g. 0067U, 0083U, 0095U).  For example, for code 0083U, the median price is $167 but the max price was $672.   For 0067U, the median was $507 and the max $1897. 

I believe as in recent years the MolDx MACs price uniformly and drive the median, because they have more than 50% of the localities and the final price is driven off the median, not the average.   

This year, there are 21 codes where Novitas priced significantly higher than MolDx (at left in chart below), and 10 codes where MolDx priced significantly higher than Novitas (at right in chart below). Excel in cloud here.

Novitas/FCSO Match;  So Do Noridian & Palmetto

Novitas always matches FCSO.  The Novitas MAC and FCSO (Florida) MAC are ultimately under joint ownership of Florida BCBS, and always priced the same (Pennsylvania always matches Florida).  

Noridian always matches MolDx (Palmetto).  Recently, it has been news that the Noridian MAC has deleted several MolDx LCDs that have been finalized in other MACs of the MolDx system.  That raised the question of whether Noridian would agree with other MolDx MACs on gapfill pricing.   I compared California and South Carolina proposed prices and they all matched.  

NGS MAC also always matched MolDx (Palmetto).


Footnote - Cryptic Explanations

MACs are required to provide "rationales" for pricing decisions, and I'm sure they are trying.  But the rationales are still pretty cryptic.  On the Rationales tab, Column C is invariably this text or a minor variation of it:

Pricing based on comparable relevant service or services (D) and resources required (B)

Initially I had no idea what the parentheticals in Column C (A,B,C,D) meant.  I now believe they trace to 42 CFR 414.508(a)(2), where A=Charges, B=Resources, C=Other payors, D=Charges/resources of comparable tests.  Based on this assumption A=3 uses, B=12 uses, C=0 uses, D=24 uses.  One case 0094U used both B and D.  

Column D information is more variable and occasionally explicit (e.g. "Contractors used the rate for procedure code 87800 and six times the rate for procedure code 88271 as otherwise relevant tests to get the recommended rate for code 0086U. ")   However, often this column D states, "Contractors applied [or were unable to apply] the 67/37% split" without telling the reader what the 67 and 33 refer to.  (I think it's direct then indirect costs, or vice versa, but it's not so stated).  Typically in the R&D-driven lab industry direct costs are circa 20% of total costs, similar to pharma and medical devices (public financials for COGS and other costs) whereas in the national large-lab industry (e.g. Quest public financials), direct costs are about 50% of total costs. 

In 8 cases, Column D says pricing information was requested but none submitted.  In some additional cases, Column D says "only a recommended price" was submitted (not viewed as actionable information) and goes on to explain how a price was rationalized by the MACs.   Overall, Column D contains a lot of information collectively about the different types of reasoning used in price-setting.

Nerd Note 

While regulations state that prices will be calculated as "median of MACs," CMS continues to collect prices for about 57 different zones (all states, some parts of states, Puerto Rico) and calculate the median of that.   MACs submit the same price for all zones under their control (e.g. Noridian submits the same price for CA1 Zone 01112 and CA2 Zone 01182).  I believe MolDx pricing would dominate the median whether calculated on the basis of 57 zones, states, or MACs.