Wednesday, September 30, 2020

Mysteries of MolDx: Advance Notice of CareDx LCD Changes

 A few days ago, I published a long blog article that inferred and pieced together some apparent rules about how MolDx operates, and why and how some Noridian coverage is presented in different formats than can be found at any other MolDx MAC (here).

Today, I was checking progress of the CareDx Allosure LCD which was proposed in draft from in Fall 2019.   DL38255 is the proposed policy, and it shows on CMS as being still in the (E)Edit (review) phase, not released to final.  CMS link here, today's screen shot next:


However, if you click on the DL38255 link at CMS, you get the proposed LCD, with a recently appended comment that DL38255 was "Released to Final" on 8/12/2020.  Link here, screen shot next.  I'm told that in this use case, "released to final" is a CMS LCD term of art, and only means the LCD has entered a final edit process.

OK, no major surprises so far.

Surprises Are Tucked Into 150-Page Palmetto MAC PDF

Later, I was reviewing the new October 2020 150-page Provider Bulletin PDF on the Palmetto website, and it DID hold a few surprises.  Find the October Bulletin here.   Cloud copy here.  (See upper right hand down-arrow to download).

In the October Bulletin, you can find information concerning the AlloSure policy that is not yet released at CMS.gov.   See page 122 of the PDF.

Specifically, we can read that the new title will be, "MolDx:  AlloSure OR EQUIVALENT Cell Free DNA Testing for Kidney and Heart Allografts," the number remaining L38255.  L38255 will be accompanied by an article (shown in the PDF, but not yet on the CMS website), A58387, with billing instructions to submit code 81479 and the correct Z code.  (As of today, A58387 isn't on the CMS website yet.)  

DL 38255 (quick covers AlloSure Heart & Kidney) is stated to replace the much older, single purpose AlloSure Kidney LCD, L37266.

Finally, Palmetto via this PDF publishes the public comments received during the open comment period for DL38255.  There were two, and we can read them on pages 124-128 of the Palmetto PDF.   

  • The first comment asks that the LCD be finalized as covering both AlloSure "OR EQUIVALENTS" - as Palmetto has done.  
  • The second comment is a lengthy comment from the CareDx company, running 3 pages.  Palmetto provides a brief response noting that some of the CareDx remarks cite unpublished evidence, which Palmetto will evaluate in the future as it is published.

Analysis

The proposed LCD was for AlloSure test alone.  The final policy is edited to become an "umbrella" or "foundational" policy for the class of DNA tests used in transplant rejection.   Thus, this is another data point that validates theories in my "MolDx Mysteries" article a few days ago here.

For example, if you are entering the market with a DNA graft rejection test, you might be able to get coverage in a few weeks based on proving "equivalent" to AlloSure, whereas a few years ago, you might have had to wait a year or more for your own LCD, or a revision in the LCD itself that discusses your test.  In "umbrella" or "foundational" Palmetto LCDs, the new tests are added by a coding/billing article, and no change is required in the text of the LCD itself.

It's not immediately obvious how this policy L38255 for "Allosure or equivalent tests" will interact with a separately- and more recently-proposed policy (April 2020) for "Liquid biopsies [no brand stated] for solid organ transplantation" which is Palmetto LCD DL38779.   



Tuesday, September 29, 2020

Very Brief Blog: The James Buchanan Genomics and Health Economics Blog

 Interesting reading, interesting website.  James Buchanan, who's with the Health Economics Research Centre of Oxford, runs a weekly blog that updates on publications for genomics & health economics.

Find the blog here:

https://healtheconomicsandgenomics.com/



Below the break I've done a simple cut/paste of some recent entries.

Monday, September 28, 2020

Will CMS Release Provider/Lab Data for CY2018? It's Almost October 2020

Since around 2014, CMS has annually released data on physician and laboratory utilization, as well as separately collated utilization data on DME suppliers.  The release involved a FOIA lawsuit and I believe Wall Street Journal and ProPublica were either both involved, or, both were early users of the data.

Because the Trump Administration places great emphasis on transparency at CMS, even against lawsuits, I would expect the 2018 version of the database to appear soon now in 2020.

It seems later this year, though, than in recent years.  The database home page (part of data.cms.gov) is here.  The full name of the data set with lab data is Medicare Provider Utilization and Payment Data: Physician and Other Supplier.

For CY2015, the dataset was created in February 2017, and updated in September 2017 (88K views).

For CY2016, the dataset was created in May 2018, and updated later in May 2018 (75K views).

For CY2017, the dataset was created in May 2019, updated in June 2019 (64K views).

When will day for CY2018 be released?  October 2020?  November 2020?  Ever?  Never?  See CMS comment here.







Saturday, September 26, 2020

2019 Medicare Molecular Microbiology Spending: $365M and Overview

In a recent blog, I reviewed CY2019 Medicare pathology & laboratory spending, which is about $7B, focusing on human molecular pathology (e.g. 812xx 813xx 814xx 815xx) which was about $1.7B.  Here The Part B data doesn't include hospital outpatient or hospital outreach claims.   I added a deep dive on 81479 (unlisted code) here.

The Molecular Pathogen Code Series

In this article, I'll briefly describe an area that some readers are less familiar with, molecular microbiology payments.  These are CPT codes 87471-87801, then skip to 87900-904.

In my hands, for CY2019 Part B, these tally $365,846,197 allowed charges, or $365M.

Using the same metrics, CY2018 was $294,470,931 (growth +24%).

Charges Are Highly Concentrated

As occurs in other areas of lab medicine, charges are enormously concentrated.  

Of 87 codes, 27% of payments go to ONE code, and over half of payments go to only 5 codes.  

The highest paying code is 87798, pathogen, other, amplified probe, with 2.6M services and $100M of allowed charges (27% of total).  It's about $39 a pop.  

The second-highest paid code is 87633, respiratory panel, 12-25 probes, with $34M for only 74,000 services, because it's $459 a pop.

Data Display; Cloud Raw Data

I've put a screen shot below (click to enlarge) and I've put my working Excel file in the cloud here.


ANALYSIS

A lot of microbiology spending goes through pretty non-specific codes.  This was a problem in the human molecular pathology realm, due to hundreds of millions of dollars of fraud being driven through 81408, a non-specific complex gene code (see my 2019 analysis here ending with a focus on 81408 and is problems.)

87798, the highest-dollar microbiology code at $100M, has no LCD or article at all at most MACs, suggesting it will auto-pay.  (And the only articles that even mention 87798, do so solely in the context of Zika testing (!) or GU testing; see A55326, A55327, A56791 (retired).)   

87633, the second highest-code, has Medicare Coverage Database edits only at the MolDx MACs (WPS, CGS, Noridian, Palmetto).  Screenshot below taken 9/27/2020.


Payments, at least through 2017, were incredibly concentrated.  For the 12-25 pathogen panels for GI and for RESP, 87633 and 87507, about half of all the nationwide Part B payments went to 4-5 providers, and not particularly large places (not Quest, etc).





2019 Medicare MoPath Spending: Rank and Distribution of "81479" Spending (Unlisted Code); 95% MolDx

Top Line

There was $1.6B in Part B Medicare spending for Molecular Pathology in CY2019.  Hundreds of millions of dollars were fraudulent.

Among the normal spending, the #3 molecular pathology code with some $200M of claims, was Unlisted Code (Miscellaneous Code) 81479.  

In some states, this code was over 90% of spending, and 95% of all use of code 81479 was in MolDx states.  Data below.

Background

In a previous blog, I discussed 2019 Medicare Part B molecular pathology spending, highlighting several trends including explosive growth in fraudulent spending, where and under what codes.   Here.

In this blog, we take a deep dive on spending under the MoPath Unlisted Code, 81479.  I collated data from the 50 state spreadsheets and found it summed correctly up to data in the single US national Part B spreadsheet.  All data shown use "dollars allowed."  (Since MoPath doesn't have copays, this is similar to actual Medicare program dollars spent.)

RESULTS

US Part B MoPath spending was $1,690,000,000 in CY2019.   Of this, $202,933,000 was coded as 81479, Unlisted MoPath code.   12% of all MoPath spending flowed through 81479, and 81479 was the #3 MoPath code in terms of dollar volume.

Nearly all spending under 81479 was in the MolDx states.  

MolDx states spending $192,480,664 on 81479, and non-MolDx states (about half the US) spent $9,452,177 on 81479.   (Services were 103,263 and 5,164 respectively.  Spending per single service was about the same in MolDx and non-MolDx states ($1,864; $1,684).

Per state, the highest percentage spend on 81479 was 96% for Ohio (15% of US total 81479).   

Arizona was next, with 52% of molecular pathology spending there, under 81479 (16% of US total).  California was next, with 33% of state spending on MoPath being 81479, and 64% of all US 81479 spending flowed into California.

The highest non-MolDx state for 81479 spending was Pennsylvania, with 4% of the US 81479 total, and 12% of Pennsylvania MoPath dollars flowed through code 81479.  No other non-MolDx state contributed 81479 spending above data dust, since 4% of 81479 went through Pennsylvania, then 95% through MolDx states.

All the data I've quoted is in the table below (click to enlarge).   


It's not an Excel artwork but I've put the underlying data worksheet in the cloud here.

_____

2019 MoPath spending was $1.7B, all Lab/Path spending was $7.3B, so MoPath was 24% of all lab spending, a far higher percentage than ever before.

_____

I don't know who Novitas/Pennsylvania spent the 4% or $8M on.  However, in CY2017, Novitas/Pennsylvania 81479 spending went to only 3 labs, with 95% or $3.2M going to Interpace for 1,839 "81479" services at $2304 allowed each.

Friday, September 25, 2020

CMS Releases National Data File for CY2019: 81408 Is The Fraudomatic Code in Some States - Submit Your Claims Tonight!

On September 16, 2020, I noted that CMS had released Excel spreadsheets of Part B spending by CPT code for every state - here.   On September 22, I released some deep dive analysis of MoPath spending by state - showing massive increases in fraudulent spending, in some cases, the states and CPT codes where the spending was concentrated.  Here.   I noted that the most egregious cases - the ones that popped out high to the naked eye, like Florida, D.C., and Oklahoma - were either under the Novitas or FCSO MACs, both subsidiaries of Florida BCBS.

  • New!  CMS has now released a consolidated file of national Part B spending, by CPT code, for CY2019.  Find the original CMS website data here.
    • I make no promises for transparency or clarity but I've put all my worksheets for this blog in the cloud too, here.

The files can maddeningly tedious because they are broken into dozens and dozens of individually tiny files (e.g. 13 kb, 21 kb) with short runs of CPT/HCPCS codes. 

Luckily PATH series codes, including "U" or PLA codes, are in one file named Y2019_0001U.xlsx on the CMS website linked above.

All Pathology, Lab Medicine

I sorted all the Path/Lab codes by TOTAL (this sums technical component, professional component, other, where they exist).  Total Path/Lab dollars allowed were $7,133,811.

Sorting by "Allowed charges," the top 15 codes were paid 56% of the total or $4.0M.

The highest code was 88305 (surgical biopsy) with 20M services and $1B in dollars allowed.

The next highest code was 80053 (routine chemistry panel), with 29M services and $342M dollars allowed.  As you can see by eyeball, with around 30M services and around 300M dollars that's in the ballpark of $10 per service.

Three MoPath codes were in the top 15:  81408 (Tier 2 Level 9), 81528 (Exact Cologard, circa $500 per case, 482K cases and $245M, sounds right).  Followed by 81479 (mopath unlisted code, used almost exclusively in MolDx states in other recent years), at 109,555 services and $202M cryptically spent dollars - cryptic, at least, to the public with this CMS file. Click to enlarge:


MoPath Codes

I merged all the PLA ("U") codes with MoPath (81162-81599). 

Allowed charges was a collosal $1.7B - last year in 2018 it was about $1B.   

The top 15 codes concentrated 78% of payments, or $1.3B.

The largest code was 81408 - Tier 2, Level 9, $2000 each, 146K services for $290M allowed charges.  That's likely due to stupendous growth in payments for this code in Novitas/First Coast states like FL, DC, OK (here).  

Next was Cologard at $245M.  

Next was Unlisted Code, 81479, traditionally used only in MolDx states, at $202M.  Then BRCA testing (BRCA1-2 & Dup Del), 81162, for 60,000 services for about $120M allowed charges.  Finally, in the top 5, was 81519, Oncotype Dx, with 22,000 services at $85M. Click to enlarge.



For a breakdown of 2019 81479 spending (95% in MolDx states), here.

Above, I've described Human MoPath codes (e.g. 812xx, 813xx, 814xx, 815xx).  There are also microbiology MoPath codes (I believe 87471-801 and 87900-904).  Those tally $365,864,197.  I briefly describe in a separate blog here.

Most PLA Codes Unused

Total PLA code spending was $116M, but if you leave off the top 2 (FMI and Oncotype Prostate) the total PLA spending plummets to $16M out of the $1.6B for MoPath - a rounding error.

The highest "administrative MAAA" code was 0003M, at #18, with only $18,781 allowed for 42 services.  Click to enlarge, but most of the numbers will still be pretty small.



The Reward For Most Unbelievable Code Goes to - 81408

Here's the basics:

2017 -  5,817 cases, $9,550,000 dollars

2018 - 62,000 cases, $123,000,000 dollars

2019 - 146,000 cases, $290,000,000 dollars

What code are we talking about?  81408!  This is AMA CPT Tier 2, Level 9, CMS price $2000.

That's 14X growth 2017-2018, and 32X growth 2017-2019.  If spending on this unbelievably rare code set had been kept at 2017 levels, national MoPath spending in 2019 would have been nearly $300M dollars lower.

It's pretty tedious to do 81408 spending in CY2019 by MAC - you'd use the 50 state spreadsheets one at a time - but it would probably show that some MACs were completely resistant to this fraud and others gushed out money like a firehose.  For example, I quickly found that despite TWO YEARS of explosive fraudulent use of this code, according to the CMS Coverage Database, the Novitas MAC still has a billing article A52986 that says "81408 has no edit codes at this time."  Ya think? I think they already noticed that.  This clipping next is from TODAY.  Wake up and smell the coffers being drained.

Click to enlarge, but it may give you nightmares:


81408 seems to be the most popular code with fraudsters, based on other analyses I've been doing.  There are only 24 allowable rare genes listed under 81408 - little gems like "LAMA2 congential muscular dystrophy, full sequence" and CEP290, Joubert syndrome, full sequence (a cerebellar malformation that cause gross maldevelopment in infancy with mean age at death age 7.

Fraudulent payments in CY2018 should have been released by now in an annual CMS database containing physician/lab payments.  However, CY2018 is not being released - perhaps because of the coming election and the fact the egregious Operation Double Helix overpayments were occurring on this administration's watch.  Here.
____
____

In 1HCY2016 PAMA data, reported to CMS in 1Q2017, nationwide commercial payer payments for 81408 were $28,800,000 for 16,159 cases, with a price range from $2.16 to $22,547 and a median of $2000, which set the CLFS for CY2018-2020.  Double the cases and dollar volume to approximate commercial payments for 81408 in CY2016.  It's unknown how many of these are legit uses of the orphan genes stored under 81408 or how many might have been concerning but simply autopaid at a percent charges.  Excel.

 









You Can Understand It: MolDx and MolDx MACs Journey Toward Liquid Biopsy Coverage (For Both Solid & Heme Cancers)

Regular readers of Genomeweb may have been puzzled by two recent short-format news articles.  One announced that "Palmetto" had finalized coverage of the Natera Signatera test for minimal residual disease detection, in colorectal cancer.   That was September 5.   

But Medicare-oriented readers will be aware that Natera bills under LCDs in California (the Noridian MAC), which simultaneously nixed (aka retired) the same LCD.  

So here's a new item at Genomeweb on September 25, that MolDx has released a master LCD for MRD in solid cancers (including Signatera in colorectal cancer).   And that the new proposed LCD would consolidate MRD coverage in solid cancers and hematopoietic cancers in one document.

What's going on?

I'll explain it.  Let me just flag the bottom line: Signatera (the test most often mentioned) is paid the same in any MolDx state, but through different routes, which are today's "Discovery in Health Policy."


Zip File All-in-One

Also, I've put all the relevant Medicare LCDs, draft LCDs, and articles in one convenient ZIP file in the cloud, here.  [Note, this zip file was updated and corrected at 830 PT 1130 ET].


What are the MOLDX MACs?

28 states operate under MOLDX, a consolidated set of LCDs, articles, and coding edits.  MOLDX is a special subcontract or statement of work for Palmetto GBA LLC, a MAC in South Carolina which is a subsidiary of SC BCBS.   The other MACs - Noridian JE, Noridian JF, WPS J5, WPS J8, CGS J15 - operate MolDx based on joint agreements.  (Here).   

They share a common edit file of some 20,000 lines.  Most of the information in the common edit file is public (with email registration) at a website called DEX (TM) Diagnostics Exchange (here).   For example, I tested that the Master Edit File was 5.9MB and matching data files on DEX (TM) were 4.9 MB by my count.   Data fields in the lightly larger M.E.F alone include (a) Z-code identifiers and (b) carrier-set prices, but note that all the carrier-set prices are explicitly releasable under FOIA per Noridian and its public FOIA website instructions and rules here.




MOLDX MAC MAP

The MolDx MACs are shown here by the red "M" symbol:





The Saga and Journey of LBX Coverage: ROUND 1

Regarding links, some links at CMS can change, but the ZIP file cited earlier in this blog has a copy of every document that I mention or link.

Around August 2019, all the MolDx MACs released a draft coverage of DL38290 (or equivalent local MAC L-number) which was entitled the "Signatera" LCD for MRD in colorectal cancer.   

In Summer 2020, Palmetto and other MACs except Noridian finalized this LCD.  \

Noridian explicitly announced in a web article Noridan was rejecting and NOT finalizing the 2019 proposed Signatera coverage - as an LCD.   

The title of the LCD (for MolDx MACs that did finalize the LCD) was changed to delete the word Signatera, but the topic was still MRD in CRC and Signatera was covered by name within the article.   

This final LCD was also released with a coding and billing article (A58283) and a Q&A document (A58280) responding to public comments.

So at the end of Round 1, an MRD LCD had been proposed in all MolDx MACs, this LCD had been deleted in Noridian MAC, and it was finalized elsewhere, with a more general title, and with some associated implementation articles.  Whew.


The Umbrella LCD for MRD Liquid Biopsy: September 2020: Round 2

On September 24, 2020, MolDx (meaning the Palmetto MAC) released a new LCD which will likely replace L38290 when the new LCD is finalized.  The new LCD is numbered L38779 (DL = Draft LCD DL38779) and is titled simply, "Minimal Residual Disease Testing for Cancer."

Get the DL38779 text here (or in Zip file).   Public comments run from yesterday, 9/24, to 11/07.  There will also be an opportunity for comment on November 2, 2020.  I assume that will be virtual (telecon or webinar) due to COVID.

I've included the coverage text in the LCD link just above, the Zip file, and in a separate HTML file here.


BOTH NORIDIAN AND OTHER MOLDX MACS 

ISSUE COVERAGE ARTICLES; DIFFERENCES


Now we get to the really fun part.  If you go to the newly redesigned and improved CMS Coverage Database (database here, my article here), today, you'll get identically titled Billing Articles for SIGNATERA at every one of the four MOLDX MACs - WPS, CGS, Palmetto, Noridian.  

So you'd probably think the identically titled articles are identical, no?  NO.  It's not that simple!


The harmonized MACs (WPS, Palmetto, CGS) have a short shell of a Coding Article that primarily points its reader to the LCD (the colorectal LCD posted last year and finalized in August).   

In contrast, the Noridian article with the same title is longer, because it quotes a full multi-bullet point paragraph of coverage LCD text that Noridian deleted/didn't finalize.   Compare A58448 at Noridian here, with A58330 at CGS here.


ANALYSIS

After the public comment closes on 11/07, and closes in OTHER MolDx MACs as well, a final LCD could be released in as few as several weeks (requiring among other things a couple-week CMS review cycle for new final LCDs).  But don't count the days.  Final LCDs typically appear after 6 months or more, based on real-world data I've run in the past. 

I think of these LCDs as "umbrella" LCDs but note that MolDx refers to them as "Foundational" LCDs.  MolDx strongly wants to shift from its history of "branded" single test LCDs to issuing only "foundational" LCDs.   An approximation - I'm not sure if it's perfectly true but it seems to work as a generalization - is that the Noridian MAC since May 2020 has generally been rejected (ex post facto) any MolDx LCDs with a branded name in the title.   Since Noridian processes 90% of the MolDx MoPath payments as recently as CY2019 (here), something that Noridian won't allow isn't a good thing for MolDx to propose.  

One issue raised is that under these non-branded umbrella (foundational) LCDs, future tests will be included (or excluded) on a rolling basis based on internal MolDx reviews - which are potentially faster than public comment period activities but provide less visibility.   

For example, if your test is covered in 2 weeks after an internal tech assessment you'll be very pleased, but if your test is non-covered and your competitor's is covered, you'll be less pleased, especially if you think the two tests are equally good.  
The full set of methods and rules for implementation of "foundational LCDs" with internal tech assessments is an issue that will be hammered out in the coming months.

 

IMPACT OF PATHWAY ON NATERA

Overall, Natera has performed extremely well over the past year, with share price doubling from about $35 to about $70 in twelve months (Yahoo Finance), and market cap reaching $5.9B on September 25.  Natera was up from $65 to $71 on the morning of September 25. 

As noted above, Natera in the Noridian MAC has the same coverage under an article (A58448) as it "would have had" in the several other MolDx MACs, under an LCD paired with a shell billing article that pointed to the LCD.





















   

Thursday, September 24, 2020

Is Noridian's Coverage Article for COVID Testing Not Right?

 I was recently on the CMS website for new articles and by accident I noticed a Noridian article that seems "not right" regarding COVID testing.

The article is A57338, and it's secondary to their master LCD for molecular pathogen testing L37301.  Famously, this legislates against coverage of viral (or pathogen) panels when the test has more than 1-5 pathogens.

Find the A57338 article at CMS here and a cloud copy here.

Very Odd Things Happen

In Article Group 1 codes, the article provides coverage ONLY in limited circumstances, being hospital or emergency rooms, or else, if ordered by an infectious disease specialist.   (If not an infectious disease specialist, you are expected to be in a remote rural location or island).  

But the article then lists codes 87631 (respiratory pathogens any type, 3-5) and CMS COVID codes U0003, U0004.  

That can't be right.  Use of CMS COVID codes like U0003 can't be restricted to hospital and ER only and when ordered by an ID specialist.  (For just one example, CMS explicitly pays for COVID testing collected in nursing homes (here).)   But we all know CMS is supposed to pay for COVID tests ordered in a doctor's office or collected by a Quest.  These CMS COVID codes can't be covered "only when ordered by an ID specialist," or only for inpatients, etc, but the article says so in black and white.

Article Non-Covers Codes Already Explicitly Listed by CMS as Covered

And CMS has an explicit listing of codes COVERED FOR COVID TESTING, and it includes codes NEVER COVERED by Noridian like 87632.   Crazy!  Look here (click to enlarge):


Go figure.  And by the way, I pulled the CMS coverage codes from the CMS website today, so one would assume it's correct. 

__

Nerd Notes.  When it came out, the article and LCD provide coverage for inpatients, or urgent care/emergency room.  In general, CMS bundles ALL tests for inpatients, and bundles tests OTHER THAN HUMAN GENETICS in the emergency room (hospital outpatient).  So the LCD actually covered molecular pathogen tests in settings where they weren't payable.   (A little different today since CMS has set certain COVID codes to be payable in ER setting).

Nerd Notes.  The Article has been update multiple times, most recently 7/17, 8/13, and 9/16.  However, the revision history section explaining what's new at each step, stops at 7/30.   

For an April blog (not updated) on CMS coverage for COVID here.

Very Brief Blog: CMS Updates Coverage Site User Interface, More Features

For years and years, CMS has had a website that provides search services for national coverage decisions, local coverage decisions, and articles. It's here.  It's quirky and clumsy but it's what we had.

Earlier this month, they released some new search features which are discussed in a two page PDF - here.

But wait, there's more.

CMS has now rolled out a new, more modern, search page.   Enter it here.



Note that this web page is like Google, it will start auto-filling what it thinks you might want:

Once you have gotten to a document, something new appears, a number of pull-down menus left-to-right that let you narrow your search (by MAC, by State, etc).


Don't Ignore Old-Fashioned On-the-Menu Reports

There are a lot of times you still want to use the old search interfaces and reports - for example, if you want all the active LCDs for Florida.   However, the new tools and new interface option make it clear they are trying hard to keep up with modern web design and they deserve credit for that.

Another feature to be aware of, the What's New, Local Coverage "REPORT" (click REPORT tab), which gives you all new articles in that week's update either nationwide or in a geography of interest.



___


Depending on how you enter the "new" search website, it may give offer you a pop-up tutorial (in 5 or 10 pop up screens in sequence).  I don't see it again, so it may be smart enough to offer that only once.





Very Brief Blog: MOLDX DEX (TM) Website Down Again This Weekend; Transition to Palmetto GBA LLC Management

Last weekend, the DEX (TM) Diagnostics Exchange was down for transition to Palmetto GBA LLC management (the contractor that manages MolDx).

MolDx is posting that DEX (TM) will be down again this weekend (September 25-28) for further transition to Palmetto GBA LLC.  See DEX (TM) here.


In August 2020, MolDx updated rules for using DEX (TM).  It's M00148, V14 here, here.



Exact Sciences Stock Skyrockets on Liquid Biopsy News

This week, Genomeweb reported that UBS (which in part issues investment advisories) had downgraded Illumina's stock rating (here).  The major recent activity at ILMN was buying Grail, a well-funded company advancing tests for cancer detection.  

Today, at the Cowen Liquid Biopsy investment summit, Exact Sciences includes a slide that while 166 people need to be screened to detect just 1 colon cancer, a Multi Cancer liquid biopsy test could potentially be able to detect a cancer while screened only 30 people (19 page deck here).  I haven't heard the whole talk yet but it has a link on the Exact website here.  Cowen's next slide releases product-specific data.




On the last slide above, see the specific data from its clinical trials on a future product.  Note that data is broken out on the right by cancer - liver, lung, ovarian, pancreatic, and so on.  Note, of course, that you can't possibly screen 30 people to get 1 case of pancreatic cancer, the cancer itself isn't common enough.

Stock Shot Upward - Billions in Market Cap

Exact stock shot up by what CNBC is currently rating at 24%.  Since the market cap has been about $10B , that's about $2B-3B in stock market value. 


The five-day performance from Yahoo Finance is here:

___

Note that while today's activity is very exciting, if you take a one year view, today's peak is still below past highs over $94 that occurred at several timepoints in the past twelve months:

Coverage of the day at Genomeweb, here.


Wednesday, September 23, 2020

Nerd Blog: If you ever see, "Access to CMS.GOV denied" on your browser

Update

PS None of my frenetic internet-recommended fixes and efforts and reboots on 9/22, 9/23 helped, but the CMS website worked fine for me on 9/24.

PPS  Problem came back Sept 30/Oct 1, then gone on October 2.  Go figure.

_________


Quick note since some others may see this sometime.  


A few years ago, and again today, I'm having difficulty accessing CMS.GOV on either of my laptops, although I can access it on the same laptop if I unplug my computer's ethernet and use my phone as a hot spot.   This is clearly some kind of DNS or internet protocol error.

Specific to CMS

I'm not the only person who's ever seen this.  Here are two websites that specifically discuss difficulties that can come and go in reaching CMS.GOV

Could Occur with Other Websites

There's also a more general condition; you can just google ( You don't have permission to access _ on this server ) where _ is nothing - don't put anything - and you'll see many Q&A's of people tearing their hair out.  (For example, one Windows trick is going to the CMD file and typing ipconfig /flushdns , which, for example, did not work for me, but gives the idea of the nerdiness of the problem.)  

Pesky

I did things like reset my DNS and cleared browser caches and nothing has worked yet.  The same problem is on both Macs and PCs in my house, but there's no problem if using an iPhone hotspot.  Rebooting did not help.  Would rather not use my iPhone as a hotspot every time I need a CMS document on my laptop.   

The main message is, Google shows that others have had the same problem sporadically and you're not crazy.





Very Brief Blog: CMS Releases Updated Mid 2020 Medical Director Directory

 For years, CMS has published an online directory of Medicare Contractor Medical Directors (MAC CMD's).   It's been updated to June 2020.  Find the PDF online at CMS here.

You should be able to convert the resulting PDF to an Excel file via this website, but I don't know anything more about that website and don't otherwise vouch for it.  I've used it.



DOJ Settles $11M Case Against Lab; Focus on Software Expenses between Lab and Providers

On September 22, 2020, DOJ announced a settlement of $11.5M against Bio-Reference laboratories which "admits to improperly billing government."  Part of the billing was for inpatient tests not segregated from outpatient tests during the billing process.   Both Medicare and Tricare and discussed.  note that since about 2014, Medicare also bundles many outpatient lab tests (but not genetics) under a complex and regularly changing set of rules.

The much larger part of the case relates to payments and values exchanged related to software used in the provider's systems.   I don't know the details of the law, but this must be a complex area because of the increasing reliance of the modern health system on electronic transactions, electronic pre-authorization through shared records and software, collating and integrating EHR and lab data, imaging, appropriate use criteria, electronic orders, etc.    

I note CMS issued a proposed rule to modernization anti kickback and value sharing regulations, in February 2020 (here).  As I understood it (and I'm not an expert here), the rule excluded labs from the modernized value-based or shared savings rules which are required to integrate and cooperate in modern health systems and reduce healthcare fragmentation.  In August 2020, CMS announced it was delaying that final rule due to complexities in its structure (here). 




_______



FOR IMMEDIATE RELEASE
Tuesday, September 22, 2020

Acting Manhattan U.S. Attorney Announces $11.5 Million Settlement With Biotech Testing Company For Fraudulent Billing And Kickback Practices

Bio-Reference Laboratories, Inc. Admits to Improperly Billing Government for Hospital Inpatient Testing and Donating Cost of Medical Software to Physicians Based on Volume of Business

Audrey Strauss, the Acting United States Attorney for the Southern District of New York, Scott J. Lampert, Special Agent in Charge of the New York Regional Office of the U.S. Department of Health and Human Services, Office of Inspector General (“HHS OIG”), and Leigh-Alistair Barzey, Special Agent in Charge of the Northeast Field Office of the U.S. Department of Defense - Office of Inspector General’s Defense Criminal Investigative Service (“DCIS”), announced today an $11.5 million settlement of a False Claims Act case against BIO-REFERENCE LABORATORIES, INC. (“BRL”), a New Jersey-based biotechnology company that provides molecular and diagnostic tests.  The settlement resolves claims that from 2009 to 2012, BRL fraudulently billed federal healthcare programs for testing conducted on hospital inpatients that should have been billed to the hospitals instead, and that BRL knowingly donated the cost of electronic medical records software to physicians’ offices throughout the country based solely on the volume of business generated by those practices, in violation of the False Claims Act and the federal Anti-Kickback Statute.  
Under the settlement approved by U.S. District Judge George B. Daniels, BRL will pay $11,500,960.00 to the United States to resolve the fraudulent billing and kickback claims.  BRL also made extensive admissions regarding the company’s conduct.  

Acting U.S. Attorney Audrey Strauss said:  “Bio-Reference Labs received millions of dollars from federal healthcare programs through its fraudulent billing and kickback schemes.  The company knowingly and recklessly billed the government for tests it should have billed to the hospitals instead, and provided kickbacks to doctors in order to induce them to order more tests.  Our Office will continue to hold healthcare providers accountable when they engage in fraud and other illegal conduct.” 

HHS Special Agent in Charge Scott Lampert said:  “The irresponsible behavior by Bio-Reference Labs compromised the integrity of the Medicare program, and wasted millions of taxpayer dollars.  Working with our law enforcement partners, HHS-OIG will continue to ensure that healthcare providers that do business with federally funded health care programs do so in an honest fashion.”

DCIS Special Agent in Charge Leigh-Alistair Barzey said:  “Fraudulent billing and kickback schemes threaten the integrity of TRICARE, the Defense Department's healthcare system for military members and their families.  Today’s settlement is the result of a joint effort and it demonstrates the DCIS’s ongoing commitment to work with the USAO-SDNY and HHS-OIG to investigate and prosecute companies that seek to fraudulently profit at the expense of federal health care plans.” 

As alleged in the Complaint filed in Manhattan federal court:

Fraudulent Billing Practices & Kickback Scheme

From 2009 through 2012, BRL knowingly and willfully billed Medicare and Tricare for certain testing performed for hospital inpatients that should have been paid by the hospitals themselves.  As a result, BRL received reimbursement from Medicare and Tricare for tests that the federally funded programs had already paid for, because hospitals receive payments for all items and services provided to the patient under the inpatient prospective payment system (“IPPS”), unless an exemption applies, which is inapplicable here.

In addition, in violation of the Anti-Kickback Statute, BRL knowingly and willfully offered and paid remuneration, in the form of a percentage of the cost of electronic medical records software, to physicians based on the volume of business generated by those physicians in order to induce them to use BRL’s services.  The Anti-Kickback Statute prohibits medical service providers, such as testing facilities, from paying any remuneration to providers in order to induce them to refer medical services.

As part of the settlement approved today, BRL admitted, acknowledged, and accepted responsibility for the following conduct:

            Inpatient Testing Claims

  • From 2009 through 2012, BRL billed Medicare and Tricare for certain testing (i) listed on the Clinical Lab Fee Schedule (“CLFS”) and (ii) performed on beneficiaries who were hospital inpatients at the time of service. 
  • Specifically, from 2009-2012, approximately 2.51% of all of BRL’s Medicare and Tricare billing originating from hospitals consisted of testing performed on hospital inpatients and listed on the CLFS.
  • For example, from 2009-2012, BRL did not bill Triad of Alabama/Flowers Hospital in Dothan, Alabama (“Triad”), for any inpatient testing.  As a result, from 2009-2012, BRL improperly billed Medicare and Tricare for approximately 2.51% of all testing BRL performed for Triad and its associated pathology practices on behalf of Medicare or Tricare beneficiaries.
  • In 2009, BRL’s requisition form – the form BRL provided to hospitals to order tests for their patients – did not contain any place for a hospital to indicate whether the patient was an inpatient or an outpatient.  But as of at least January 2010, BRL management had a clear understanding of the necessity to bill hospitals – and not Medicare or Tricare – for testing performed on hospital inpatients and listed on the CLFS.  Indeed, on January 27, 2010, the Director of Genpath Accounts Receivable wrote to management, “I’m afraid that we can end up billing Medicare for hospital patients.”  Nevertheless, the requisition forms remained the same, and through at least 2012, BRL billed Medicare and Tricare for hospital inpatient testing listed on the CLFS.


            Software Cost Donations

  • In addition, from 2009 through 2012, BRL provided a percentage of the cost of electronic medical records transition software (“EMR Software”) to physicians’ offices based on the volume of business generated by those offices. 
  • Specifically, from 2009 through 2012, BRL engaged in a practice – at the direction of its management – entitled the “3 to 1 calculation,” meaning that BRL conditioned the provision of payment for EMR Software to physicians’ offices on whether a physician’s office would generate revenue equal to three times the value of the EMR Software BRL provided. 
  • For example, on January 24, 2009, a BRL employee, in an email to BRL management, applied the 3 to 1 calculation to a particular physician’s office and suggested that BRL provide the payment for EMR Software, but noted, “You find the legal way to say that.  I don’t feel they will make us put it in writing.” 
  • Similarly, on January 7, 2011, BRL management evaluated a BRL salesperson’s request for payment for EMR Software to a particular physician’s office, and directed that salesperson to “[b]uild volume to meet 3x rule.”
  • During this timeframe, BRL provided payment for EMR Software based on this formula to 69 separate physicians’ offices. 

 

*                *                *

BRL agreed to pay a total of $11,500,960.00 to resolve these claims: $1,396,386 to resolve the Inpatient Testing Claims and $10,104,574 to resolve the Software Cost Donation claims.  OPKO Health Inc. (“OPKO”), which merged with BRL in 2015, will serve as guarantor of BRL’s obligation to pay the settlement amount.

In connection with the filing of the lawsuit and settlement, the Government joined two private whistleblower lawsuits that had previously been filed under seal pursuant to the False Claims Act. 

Ms. Strauss thanked HHS-OIG and DCIS for their assistance with the case. 

The case is being handled by the Office’s Civil Frauds Unit.  Assistant U.S. Attorneys Michael Byars and Ellen Blain are in charge of the case.

Topic(s): 
Health Care Fraud
Contact: 
James Margolin, Nicholas Biase (212) 637-2600
Press Release Number: 
20-201

Tuesday, September 22, 2020

CMS 2019 MoPath Spending Analyzed: Massive Growth, Fraud Made Visible

In the past several weeks, both the OIG and MEDPAC have reported on Medicare molecular pathology spending in CY2018.  See OIG here, MEDPAC here.   In a nutshell, investigators for both organizations found that between CY2017 and CY2018, MoPath spending shot up from about $500M to $1B.  

Before this 2017/2018 doubling, MoPath spending had been roughly level for several years, with that $500M dominated by a few large tests like BRCA, Exact Cologuard, Oncotype Dx.

In several recent past years, I've found that the MolDx program channels a large part of CMS Part B mopath spending, such as 80%.  As we'll see in a moment, that's no longer the case at all.   

Now that CY2019 state level data has been released (here), there's a shocking discovery that MoPath spending growth has skyrocketed in unusual states, in several cases by 500-700% in one year.    This may be related to massive genetic test fraud reported by the Department Justice through their sting operation "Operation Double Helix" - here and here.  CMS funds seem to be protected in states that use NGS MAC edits (CMS JK, J6) or alternately, in states that use MolDx edits.  

Data in a Nutshell

I've put my Excel research file in the cloud here.   Methods - Basically, I went through 50-plus Excel files state by state, pulled a sum of spending for 811xx 812xx 813xx 814xx 815xx, plus adding spending for 0037U (FMI F1) in Massachusetts.   (Except for 0037U, I didn't assess PLA codes.)  Please note I'm assuming CMS assignments between state jurisdiction codes and actual states were correct during several hours of manual work, and assuming that other manual data collation was correct.  The data doesn't include hospital outreach labs.

Total MoPath 2019 Spending:  $1,642,287,344

     Up from circa $500M just 2 years ago. (FN1)

     I'm just showing CY2019 Part B; CY 2019 Part A + outpatient Part B will be closer to $2B. (FN1)

Spending in MOLDX states:  $569,720.113 (35% of total)

     Spending, within MOLDX, in Noridian states: 90%

     In recent prior years, MolDx was closer to 80% of total.

Spending in Non-MOLDX states:  $1,072,567,231 (65% of total)


State-Level Data Shows Fantastic Growth Rates

Here's where it gets really interesting.   

  • Spending in Florida was $123M in 2019, but only $17.5M in 2018, a 7X growth rate in one year.
  • Spending in DC was $43M in 2019, but only $7.5M in 2018, a 5.7X growth rate in one year.
  • Spending in Oklahoma was $123M in 2019, but only $33M in 2018, a 3.7X growth rate in one year.  

Some of this 500%, 700% growth in a matter of months almost certainly represent fraudulent payments under Operation Double Helix.  CMS will release payments by individual providers and labs for CY2019 in about mid 2021 (here) and we'll know more about where the $500M landed by NPI number and street address.   

Wisconsin spending seems high at first at $243M or 15% of USA total, but this includes about $120M for Exact Sciences Cologard.  In Massachusetts, FMI garnered $79M under code 0037U.

Louisiana seemed a little high at $70M, but that's actually down 20% between 2018 and 2019.

How Florida Got 7X Payments and $123M in One Year

Not encouraging.  25% or $32M came from Tier 2 code 81408.  14% or $17M came from BRCA.  Another 9% or $11M came from Tier 2 code 81407.  

That's half the total or about $60M of $123M in just three CPT codes, two of them vague ones.



For more on 81408 fraud, see my September 25 blog here.

_____

MolDx is Protective, Though It's Not Required

Note that none of the MolDx states had the super high growth rates seen in other MACs, in Florida, Oklahoma, DC.

However, you didn't have to be in a MolDx MAC to successfully avoid the eye-popping super-fraud.  None of the many states under the NGS MAC (J6, JK) had any eye popping fraudulent growth, and toward that end they didn't require in NGS MAC the level of rules and suprastructure built up for MolDx.  

In addition, note that while neither NGS MAC states nor MOLDX states have such large Double Helix Fraud as to be visible to the naked eye, like a 10X increase in 81408, newspapers last fall did report that both NGS MAC states (Illinois) and MolDx states (Georgia) did have some focal cases that DOJ connected to Double Helix (here).  E.g. per that link, an Illinois case involved just $4.6M (NGS MAC) while a Georgia case involved $154M but may have been charged and billed in Florida.

Pay and Chase - Alive and Well

I worked in the CMS system 2004-2008 and MUCH was spoken that "as of now and forward, we're catching payments up front, we're avoiding pay and chase."  

If you look at 2018, 2019, 2020 announcements, someone will regularly say, "From now forward, we're avoiding pay and chase."  2019 statement here.  I can guarantee you the press release verbiage was the same in 2004.  

Comparing New Data to DOJ Press Release in Fall 2019

DOJ reported $2.1B in fraudulent "charges."   First, we wouldn't expect to see the dollars in "payments" that appear in "charges."  Medicare doesn't pay charges, they pay fee schedule rates.  A charge might be $2000, a fee schedule payment $1000.  I believe that CMS payment files I am data-mining for CY2019 reflect paid claims, adjucated and not (at the time of compiling) recouped.  

I believe (but I'm not certain) a claim rapidly recouped or reversed wouldn't have shown as a payment 18 months later.  In any case, it's clear from the 2019 state payment data that at Novitas and FCSO MACs, it's still pay and chase on a colossal scale, even on errors the naked eye can see - no machine learning or AI required.

___

FN1

As I noted in the introduction, OIG reports $1B in Medicare genetic spending in CY2018, and I find $1.6B in CY2019.   

BUT: My data doesn't include hospital testing, while OIG's does.  

The comparable number in OIG terms for CY2019 may eventually be published as $1.8B or $1.9B, adding in (legitimate) hospital genetic payments.

Oveall, hospital lab spending at CMS is 29% of the annual CLFS spending.  If that held true for MoPath spending, one would scale up 1.6B x 1.40 (=100/71) to get $2.2B in CY2019 mopath spending in both compartments, independent lab spending and hospital lab spending.  However, I believe that the $1.6B in independent lab spending is inflating by hundreds of millions of dollars of fraudulent spending at fly by night labs (as DOJ admits), which isn't likely to be reflected in the hospital ledgers.

Source for 29%: I've extracted this in prior years by comparing CMS carrier Part B spending with spending summed up by OIG annually, the different being hospital lab spending.   In this case, I've used the 29% from an October 9 report by the investment firm Nephron.

CMS Releases Proposed Prices for Clin Lab Fee Schedule CY2021 - Public Comment Open

CMS has posted Preliminary Payment Determinations for the CY2021 Clinical Lab Fee Schedule on its website.  I've also placed a cloud copy here.   

Notation in the Excel spreadsheet informs us that comments are accepted until October 21, 2020 at: 

CLFS_Annual_Public_Meeting@cms.hhs.gov

There are 132 agenda items, the most ever.


GAPFILL BONANZA

By my count (and using the Excel filter/display function) about 53 of 132 codes are gapfilled.

THESE PRICES CAN CHANGE

Based on past years, a significant proportion (a third?) of these values can change between the proposed and final pricing determinations.

COVID

CMS covers the first COVID testing codes from last spring at about $50 (carrier rates, standard COVID) or $100 (CMS administrator's rate, high throughput COVID) - U0001-U0004.   CMS recommended all of these codes be "gapfilled" during 2020.

CMS also recommended additional COVID codes be gapfilled: 87635, 86328, 86769, 

PSYCH PGX

For observers of Myriad's Genesight test and pricing (currently locally priced at $1569 by MolDx; see recent news reports quoted here) - CMS sends the 14 gene psych PGx panel 0173U and the 15 gene psych PGx panel 0175U both to gapfill.    0173U and 0175U belong to RPRD Dx and Genomind, respectively.  If these codes enter the Gapfill process, which is dominated by the 28 MolDx states controlling the gapfill median, the simplest prediction is that MolDx prices these PGx tests at the parity rate to Myriad Genesight and AltheaDx test both at $1569 and due to the 28 MolDx prices, the 2021 gapfill median will be $1569.*

RECONSIDER?

Six codes were under the reconsideration process, meaning stakeholders were unhappy with their pricing by crosswalk last fall.  4 of the 6 are to be gapfilled in CY2021.  In these cases, panel experts had unanimously or almost unanimously recommended gapfill.  In the other two cases, a large majority of panelists had recommended the same crosswalk.

MAAA MICROBIOL

In a new trend, two multi-pathogen tests were approved by AMA as MAAA CPT codes (81XX4, 815X3).  They were crosswalked to 87631 (resp virus 3-5 $143) and 87506 (GI virus 6-11, $263), respectively.    The text for the tests tests doesn't sound so dissimilar - 


Genome-web cytogenomics

Three codes for genome-wide cytogenomics were gapfilled.  0209U: No public comment; panel recommended gapfill.   0156U: Applicant asked for current microarray cytogenomics code 81229 x 1.5 (about $1500).   021U: (chromosomal analysis + exome sequence): applicant asked for very elaborate crosswalk stack (I believe about $12,000); gapfill.  

_____


That's a long sentence but shows my three years of studying German grammar in college weren't wasted (long sentences).









Monday, September 21, 2020

Very Brief Blog: Next PLA Deadline October

 

PLA codes continue to grow - numbers up to have been assigned, and another couple dozen codes will be released at the end of September.   If you want to get on the PLA code bandwagon, the next AMA CPT deadline is October 1, 2020.   The AMA PLA committee will vote on the codes within a few weeks although the codes won't be released until around December 30.  See information on the application process from links in the main PLA web page here.  It's a short online application just a few paragraphs long.

In addition, on the main CPT PLA Codes Homepage here, CPT should release the results of a couple dozen new PLA codes in the July 1 quarterly cycle, around September 30.  Currently they're up to PLA code 226 (0226U), so the next batch will probably approach 0250U at the top end.



___

AMA CPT makes PLA codes on a quarterly calendar but they've been releasing COVID codes on a rapid basis, in a couple weeks.