Tuesday, March 30, 2021

Very Brief Blog: NYT Exposes Elite Hospitals Charging $3000 to Payors for Single COVID Tests

 In brief, a New York Times story today describes elite Manhattan hospitals - like Lenox Hill - charging $3000 for single COVID tests.   They note that Congress requires COVID tests to be covered without a fee to the patient, but assert there is price gouging by some suppliers.

https://www.nytimes.com/2021/03/30/upshot/covid-test-fees-lenox-hill.html?action=click&module=Top%20Stories&pgtype=Homepage






Very Brief Blog: JAMA: After Vaccination, Strong Real-Life Neutralizing Antibodies for COVID Variants

An interesting short open-access article in JAMA this week by Edara et al:

https://jamanetwork.com/journals/jama/fullarticle/2777898


There are two types of COVID antibody tests, simpler tests for the presence of anti-COVID antibodies, and more complex tests that actually use cell culture to detect whether there are toxic "neutralizing antibodies."   In this report, the authors set up not one, but four types of cell culture systems for neutralizing antibodies, against four major variants.   

Two findings.  One, vaccinated persons (Moderna version) produced stronger neutralizing antibodies than either new or recent COVID patients.   Second, the antibodies performed well regardless which of the four viruses was used in the cell culture system.  Click to enlarge.


Note that in the left panel, a few hospital patients 10 days into their illness still had minimal detectable neutralizing antibodies (their dots sit at bottom in each column.)   Note also that the convalescing patients (middle panel) were several months post infection, and expected to have gradually falling antibodies, whereas the Moderna patients (right panel) were exactly 2 weeks after their second injection, which might be the natural "peak antibody" timepoint.


AMA Releases Agenda, Proposals for May CPT Meeting; Permission to Comment Available to April 15

As AMA prepares for its May 2021 CPT meeting, it has released the agenda of code proposals.  Interested stakeholders can obtain a copy of the application packets by requested by April 15, and then, responding by April 22.  AMA screens the requests to comment for being a validly involved stakeholder (e.g. not a journalist, not just curious, etc).   For example, a competitor company is a valid stakeholder.

See the code proposal list here:

https://www.ama-assn.org/system/files/2021-03/cpt-panel-may-2021-agenda.pdf


There are circa 54 total agenda topics.  It looks like 17 are for Category III codes.  The codes range from Tab 6, Revise Non Face to Face Inter-Professional Consultation (99451), to Tab 60, a proposal to create a code for cytogenomic (genomic wide) analysis by the low-pass sequencing method (styled as an addition to the 81228/81229 cytogenomics code family).   This technique has been reported to have less adverse racial bias that some other methods (here).

My vote for most interesting/unusual/unexpected application is Tab 8, "Climate related health counseling."  On the digital front, see Tab 15 (CT post processing with artificial intelligence for low dose CT), and the pair Tab 46 (validated digital online therapy program) and Tab 47 (remote cognitive digital behavioral therapy.)

It might be a bit scary for owners of Cat III codes that there are ~17 new applications for these codes, but I only counted 1 application to elevate a Cat III to a Cat I code and 17 applications for simple deletion of Cat III codes.  

###

Lab test items had an earlier comment cycle in the beginning of March (entry point here) to allow the comments be available for pre-May subgroups for lab tests (Molecular Pathology Advisory Group MPAG, Pathology Coding Caucus PCC.)





MGH Study on Genomics, Low-Pass Sequencing, Racial Bias

We periodically hear that too many genetic studies are overpopulated with people of white European backgrounds (e.g. here).   This week, a paper by MGH's Alicia Martin et al. is the basis of a lead story by Christie Rizk at Genomeweb.

  • Genomeweb: Low-Coverage Sequencing Effectively IDs Novel Variants in Underrepresented Populations. (here
  • Amer J Hum Genet: Low-coverage Sequencing Cost-effectively Detects Known and Novel Variation in Underrepresented Populations. (here)
Martin et al. write, "[M]ost genetic studies use genotyping arrays and sequenced reference panels that best capture variation most common in European ancestry populations....Low-coverage sequencing approaches surmount the problems induced by the ascertainment of common genotyping arrays, [and] effectively identify novel variation particularly in underrepresented populations..."

To my knowledge, the largest study directly comparing low pass genome sequencing to microarrays appeard in 2020 in J Molec Diagnostics, by Chaubey et al., a team based at Perkin Elmer (here), using 409 cases, each detected by multiple lab methods.  (See earlier coverage of Chaubey et al., here).   

There is no Category I CPT code yet for low pass cytogenomics, although there are several PLA codes, including one at Mayo Clinic, one at Perkin Elmer, one at New York Genome Center (0012U, 0209U, 0156U.)

More From the Martin et al. Study

Genomeweb goes on to describe the following:
They [Martin et al.] further found that 1x sequencing was among the more affordable options, costing less and performing similarly to or better than commonly used lower-density arrays such as the Illumina GSA. 

They also noted that the GSA is composed of variants that are most common in European populations and so it's therefore not the most appropriate technology for studies of participants with primarily non-European ancestry. 

Aside from cost, low-coverage sequencing had several distinct advantages compared to GWAS arrays, particularly more accurate identification of genetic variation across the allele frequency spectrum in underrepresented populations. 

In the NeuroGAP-Psychosis data, the researchers found that 38 percent of common variants could not be imputed from the 1000 Genomes Phase III data, most likely because of a lack of eastern and southern African diversity in that panel.  

_____ 

For two Economist articles on racial bias in medical technologies, see here and here.

Sunday, March 28, 2021

Very Brief Blog: Three Interesting Articles: (1) DNA Horizons & COVID (NYT); What's Next Moderna (Forbes); Payer Medical Directors as Leaders (McKinsey)

Three interesting readings came across my desk this weekend.

  1. Genomics Boom.  New York Times Sunday Magazine runs a feature article on the boom in genomics in the last several years, the explosion in sequencing technologies triggered by COVID, and what's next for healthcare and society.  By Jon Gertner.
    1. Here.
  2. Moderna's Horizon.  Perhaps even more interesting, Forbes runs a long interview with the CEO of Moderna, and talks about what's next - cancer vaccines, a far more effective flu vaccine, and much more, based on the horizons made possible by injectable RNA technologies.  By Leah Rosenbaum.
    1. Here.
  3. Chief Medical Officers. The global consultancy McKinsey and Company releases a 5 page white paper on the increasingly important strategic role of payer medical directors and chief medical officers.
    1. Here.


___

Unrelated to the McKinsey article, an article appeared by subscription in Financial Times on "the value of corporate chief medical officers."  A copy of the FT article appeared to be reflected here.

Saturday, March 27, 2021

Mark Your Calendars: MedPAC PAMA Report on Lab Policies, Friday April 2

In December 2019, Congress delayed PAMA implementation by a year (and in Spring 2020, by an additional year).   However, in return, Congress required the federal advisory body MedPAC to report on PAMA policy and issues by April 2021.

That day has come.  

  • MedPAC has released its agenda for Thursday/Friday, April 1/2, 2021, and it includes a briefing on its Report to Congress regarding PAMA.   
  • Find the home page for the meeting here.  
    • MedPAC hasn't released the slides or text for the report yet.  
    • Transcripts are released a few days after the meeting.
    • At the website, look for a video registration links for GoToWebinar.  There's separate links for Thurs AM, Thurs PM, Fri AM.  They ask you to please, only sign up for sessions you will attend due to limited broadcast bandwidth.
    • There is a Public Comment box, but it's on the home page for Public Meetings, not on the home page for the April 1-2 meeting.  Here.  It just leads you to a comment mail box at meetingcomments@medpac.gov .

PAMA TOPIC is scheduled from 11:00am to 12:30pm on April 2 (ET).  

MedPAC discussed lab pricing under PAMA in September 2020 (entry point here).  Their discussion of lab pricing falls in the context of increased molecular pathology spending, in round numbers from $500M in 2017 to $1B in 2018 to $1.5B in 2020.   Both OIG and MedPAC have pointed responsibilty to what they call "expensive tests."  OIG here.  In  is annual multi-topic report to Congress earlier in March, MedPAC suggested Congress or CMS restrict telemedicine orders of expensive DME or lab tests, due to  program vulnerabilities revealed by "Operation Double Helix."   Operation Double Helix asserted there had recently been $2B of genetic test fraud, much of it by fly by night telemedicine orders.  ("...fraudulent genetic cancer testing has resulted  in charges in five federal districts against 35 defendants associated with dozens of telemedicine companies and cancer genetic testing laboratories.")  For an example of a conviction of this type of telemarketing/telemedicine scheme after a 4-day trial, here.

The assertions about telemedicine misuse conflict with the complexity of genetic testing and the need for expertise (and, for commercial insurers, often preauthorization), which can often be provided by telemedicine by expert genetic counselors and physicians.   See for example the highly legitimate acquisition of Trapelo's precision oncology services and platform by NeoGenomics last week - here.  see also the alliance between Foundation Medicine and Informed DNA announced earlier this month - here.



I''ve argued in a series of 5 detailed reports that the Double Helix expenditures were almost entirely driven by a stupid lack of edits of very costly genetic test codes at just three MACs (Cahaba (now defunct), Novitas, FCSO).  Here.


The explosive growth in MoPath tests was only partly accounted for by reasonable growth (e.g. more tumor tests to guide drug choices).  Much of the growth was accounted for by genetics labs named in Operation Double Helix.  See for example the explosive growth in the code 81408, unedited-for in southern states.  But look equally at where 81408 was never paid!  Never billed by labs like Quest, Invitae, Ambry, Bioreference, etc.  And 81408 was essentially never paid in the MolDx 28 states (which have elaborate test controls, funding, and infrastructure) nor in the NGS MAC states (which have no novel controls or contracts for genetic testing.)

Practically100% of 81408 Spending, 2019, in Novitas & FCSO MACs





___

In December 2019 legislation, Congress delayed PAMA but required MedPAC to report on its implementation, options to streamline data collection.  However, Congress  added a phrase allowing MedPAC to also report on any issues it deemed relevant.

The PAMA delay is confusing.  Originally, national lab prices from 1H2019 would be submitted to CMS in 1H2020 to set prices in 2021, 2022, 2023.  Now, lab price data from 1H2019 will be submitted to CMS in 1H2022 to set prices in 2023, 2024, 2025.


Friday, March 26, 2021

Nerd Note: Finding Out if an IDTF is Enrolled in Medicare

Medicare has a unique type of provider (created by regulation not statute) called IDTF - Independent Diagnostic Testing Facility.

For example, this is how iRhythm is enrolled as a Medicare medical provider, as an "IDTF."  

HeartFlow 

One of the most innovative providers in cardiology imaging is HeartFlow, which provides sophisticated computed fractional flow reserve based on prior high resolution cardiac CT imaging.  Website here.

As a cutting-edge innovator, Heartflow faced a series of problems with CMS that are a matter of public  record.  

  • For example, they experienced denials when attempting to enroll (here 2016, here 2017).*  
  • Later, when Heartflow got a CPT code (Category III), I recall that CMS initially wanted to categorize it as "bundled" (ouch!!!).  
    • But with some good arguments, CMS later recognized Heartflow analysis as separately payable, and assigned it a substantial hospital outpatient APC payment level (I think around $1500).

I noticed that Noridian has a brand new pending LCD for HeartFlow which will be effective in April (L36615, A58097).   Seeing those LCDs, it led me to as, Is HearFlow enrolled as an IDTF now?


Is Heartflow Enrolled as an IDTF now?

Start by finding the NPI Number

First, it's really easy to lookup an NPI number - you can google Heartflow and NPI, or go to a database

The NPI for Heartflow is 1982019980.

What Doesn't Work

There's a CMS Physician Compare database that lets you look up whether a physician NPI is enrolled in Medicare, or a hospice,  but this dataset doesn't include IDTF as a category to look up within.

https://www.medicare.gov/care-compare/?providerType=Physician&redirect=true


What Did Work

Support staff at Noridian pointed me to a public database called MEDPARD that gives LOTS of types of providers, many more than the prior link.  Use this Noridian link:

http://norweb.noridian.com/medpard/main.aspx

Here, the drop down menu gives you all the physician specialties (hematologist, internal medicine, endocrinology), plus a plethora of institutional categories (hospital, hospice), plus... IDTF.



Select IDTF and select Northern California, and you get a list of 200 IDTFs, one of which is HeartFlow.  


Bingo!


Footnotes

Misnamed?  MEDPARD is "Medicare Participating Physician Directory," but as just indicated, it seems to be misnamed, in as much as it displays a large variety of provider types besides physicians.  

Other MACs, Different Windows to MEDPARD.  I may have been lucky with the Noridian webpage into MEDPARD; I tried to search the MEDPARD database at the NGS MAC for any IDTFs in Illinois and got "0" which seems unlikely.   

I tried Novitas for Texas, but it required a county to be named in Texas, before initiating its Medpard search, and when I picked one by chance, it gave an error message.

___

* Based on information I got via FOIA, these enrollment challenges between HeartFlow and Noridian continued past 2017 into 2018 or 2019.




Wednesday, March 17, 2021

SoCalBio: Fourth Annual Digital Health Conference

VIDEOS POSTED:

http://socalbiodh.com/

 


#####


Did you know that Los Angeles and Orange Counties are one of the top several investment zones for digital health?   See the 4th Annual Digital Health Conference, sponsored by SoCalBio, the Southern California Biomedical Council.

The SoCalBio website is here.

The page for the digital health conference is here.  As you scroll down, you'll see the full agenda.

The conference runs March 23, 24, 25 (Tuesday, Wednesday, Thursday).  The three-day list of topics and speakers is pretty impressive.

I'm speaking on a panel on digital health policy and payers - 3 pm PT on Wednesday, March 24.  Our topic is, "Digital Health - Regulatory and Reimbursement Issues to Watch."  We'll have myself, Vivek Thakkar, Digital Health Program Director, Genentech; Rick Gannotta, Chief Healthcare Administrative Officer, Masimo; Evan Seigel, CEO, Ground Zero Pharmaceuticals; and the panel is moderated by Peter Blaisdell, Chair, SoCalBio Innovation Catalyst Program.




Webinar at Dark Report: Genetic Testing Marketplace, What's Changing (March 25, 1 ET)

See a webinar convened by Dark Report on Thursday, March 25, 1 ET (10 PT):

"State of the Genetic Testing Marketplace--Getting Paid for All Your Lab's Genetic Test Claims: What's Changing, What's Not, What's Working Best."

Home page is here.

Registration page is here.

Discount code, GENTEST.


Speakers will include Robert Metcalf, CEO, Concert Genetics; Healther Agonstinelli, AVP of Strategic Revenue Operations, Xifin; and Bruce Quinn.


Tuesday, March 16, 2021

Trapelo Hosts Podcasts with Medicare and Commercial Payer Viewpoints

Trapelo Health is an interesting organization that aims to improve the patient experience for cancer patients, reducing complexity and making decisions more clear

  • Find their website here.
  • The week after this blog was written, Neogenoimcs acquired Trapelo for $65M - here.

Trapelo offers some interesting media collateral.   (Email sign-up required).

  • See an interesting 60-minute webinar, an interview between Lee Newcomer MD, former chief medical officer of United Healthcare, and Clynt Taylor, CEO of Trapelo Health.  Here.
  • See their podcast series with leaders in precision medicine.  Sign up for access here.  Of particular interest to reimbursement issues, see:
    • Episode 44, February 22, 2020, 25 min, with Dr. Gabriel Bien-Willner of MolDx.
      • Bien-Willner is also heard in Episode 40, October 29, 2020.
    • Episode 20, July 25, 2019, Chet Burrell, former CEO of CareFirst BCBS.
    • Helpfully, Trapelo provides {DFs as well as audio.
    • Podcasts are hosted by Jerome Madison, Trapelo's VP for Provider Relations.

_____________

Tangents from the topic of Trapelo Health.  See the new alliance between Foundation Medicine and Informed DNA (here), and read about an alliance between PWN Health and Sema4, the latter recently valued at $2B.  See also the approach of Cancer Commons to clarifying the confusing journey of cancer patients (here).  Grand Rounds, which has one line of business in oncology second opinions, merges this week with Doctor on Demand in a "multi billion" dollar deal.  Doctor on Demand recently raised $75M, while Amazon announces virtual-first healthcare services here.   CMT Solutions has prior authorization as a core competence, but advertises a range of services including genetic counseling; raises $10M (here).  XIFIN, which began as a lab billing firm, doesn't forward-integrate into patient care services but does provide advanced financial data like copay estimators (here).  Yet another firm aiming to sherpa over the patient-payer divide is Transcarent, recently raising $44M.


In a similar vein, see news at Genomeweb (here) that ACMG has published a new guidance on bringing together genetic screening tests with effective clinical counseling, follow-up, and care (here).   Challenges include lack of funding for genetic counseling in most health systems.






Informal Blog: Diagnostic Tests and Saving Money

I was on a call recently with a startup lab company that has a new technology it asserts will save money.  The startup doesn't have any modeling or data yet, but was making the general argument that better diagnostics save money.

Do diagnostics save money?  Well, sometimes yes, for sure, but not necessarily.   I put together some citations and I'm logging them here.

Do Preventive Services Save Money?

On the topic of preventive services saving money, much has been written.  In general, preventive services that are widely adoptive are at least cost effective (the cost in dollars per each life year saved), but not literally cost-saving.  See a New England Journal article from 2008 here.  See a 2012 Congressional Budget Office report (discussed in NEJM) here.  See a 2018 Aaron Carroll article here.  

Looking at imaging, mammography is beneficial, but likely not literally cost-saving, with costs per life year in the $10,000-$70,000 range (herehere).

For one example, when smoking prevention avoids a lung cancer death, that person lives on to die of something else - a stroke, colon cancer, Alzheimer's disease, and so on.   

A CMS-funded study in 2019 seemed to show that 7 different colorectal cancer prevention strategies all saved money (at least a bit, a bit better than cost-neutral) against no-screening, but some of the more expensive methods were less effective (per this study) and ;ess cost-effective than less-expensive methods. So just because a method is more accurate and more expensive, it is not necessarily more cost-saving or more cost-effective.  (I think this publication has had rebuttals; take it as a for-example).

Do Diagnostic Tests Save Money?

Sure, some do.  See Pimentale 2016 for cost savings with better diagnosis of irritable bowel syndrome (here).   See Pliakos 2018 on saving money with more rapid diagnosis of blood stream infections (here.)  

But is that the big picture?  If we want a general answer, we should probably turn to a comprehensive review.  Fang, Neumann et al at Tufts did a 2011 comprehensive systematic review of diagnostic test cost effectiveness (here).  Reviewing 141 studies, about 15% reported results that were literally cost-saving; others were "cost-effective."   See figure, in which the left-most column is the number of the tests reported as "cost-saving:"

Fang et al. 2011 Fig 3 
Industry Funding?

In J Clin Oncol in 2018, Wang et al. looked at numerous cost effectiveness studies of a 21-gene breast cancer recurrence test (here).   They reported that they found a range of cost effectiveness values, critiqued the methodology of some reports, and reported that the two studies with the "worst" cost effectiveness (>$50,000 per life year), were "not" industry-funded.  A few studies, but only a few out of 20 or so, reported "cost savings" per se.

Most recently, writing in Value in Health in 2021, Koldehoff et al. looking at the cost-effectiveness of genetic testing for breast and ovarian cancer risk (e.g. BRCA testing) - here.  BRCA testing could run from being cost-saving to have a cost of $22,000 per life-year (QALY).  In other settings, multi-gene testing ran from $15,000/QALY to $70,000/QALY, depending on scenarios and health care costs in selected countries.   

Are diagnostic tests cost-saving?  "It depends," and if you are depending on an insurer like BCBS to pay you a lot more, "because you are saving the payer billions in costs," they'll probably require some convincing.





Landmark Study of Personalized Medicine Integration in Health Systems; Health Advances & Personalized Medicine Coalition

The Personalized Medicine Coalition (PMC) and the consultancy Health Advances have published a landmark study on how U.S. medical systems are integrating (or failing to integrate) personalized medicine into their healthcare delivery and workflow.

The work was previewed at the November 2019 PMC annual meeting in Boston.

See the open access article at J Pers Med online here (PDF, 13 pp).

See the March 16, 2021 press release at PMC here.

click to enlarge

___

For another perspective on adoption of personalized medicine at health systems, see a November 2020 online article by consultancy Decibio, here.

Monday, March 15, 2021

MEDPAC Releases Annual Spring Report to Congress (531pp); Targets Genetics by Telemedicine

Each spring, MEDPAC releases a lengthy annual Report to Congress.   This year it weighs in at 531pp.

  • Full report here.
  • 24 page chapter on telehealth, telemedicine, here.
  • 3-page summary press release, here.

Nix Telemed Orders for DME and Genetics?

MedPAC formally recommends (page 470-471) that higher cost clinical lab tests (e.g. genetic tests) could be ordered only after clinicians provide at least one face-to-face visit with the patient within six prior months.  They note this is in response to cases where DOJ has charged 300 persons with submitting $6B in "false and fraudulent claims" (quote) of which $4.5B was related to telemedicine "without interacting with patients" or "only a brief telephone conversation" with patients.  

Note that Medicare already requires that a test-ordering physician must be treating the patient and managing his/her disease for which the test will be used (42 CFR 410.32).  A telemed doc who has a 1 minute call with the patient and never talks to her again, would not meet the "physician managing the disease" requirement.

Telemedicine

There has been a radical expansion in telemedicine in Medicare under the COVID pandemic, and providers have been paid at office-based rates which are higher than facility-based rates.  MedPAC recommends continuing this for 1-2 years after the pandemic ends to "gather data," although they recommend that during this extension period, CMS cut payment rates to the facility-based rates regardless of the physician's location (e.g. office). 

Incident-To Under a Microscope

They recommend CMS collect data about incident-to services, focus on potential problems with incident-to telemedicine services, and prohibit incident-to billing if a provider can provide the service directly (e.g. a physical therapist or psychologist can bill Medicare directly; 471-472.)


Friday, March 12, 2021

CMS Delays Start Day for "Breakthrough Device" Coverage (New 30 Day Comment Period)

Comments ran to mid April and CMS posted 97 comments on April 22 - here.   Advamed comment here.

CMS will post final decision on MCIT by about May 15.

####


On Friday, March 12, after market close, CMS announced it was delaying implementation of the Medicare Coverage for Innovative Technology (MCIT) program.  The program, which was released as final by the Trump administration in mid-January, would provide 4 years of coverage for any devices cleared by the FDA as "breakthrough pathway" devices, from a start date of 3/15/2019.   The regulation also "codified" longstanding CMS rules defining "reasonable and necessary" as a coverage criterion.

The delay was pre-posted as a PDF on March 12 and published in the Federal Register on March 17  published in the Federal Register on Wednesday, March 17.     Home page for the new rule here.   See the green box, "Submit A Formal Comment."  PDF of rule here.  Citation, 86 Fed Reg 14542-45.  Comments due April 16.  

 The change is styled as a "60 day delay with comment period."   The new implementation date, if nothing else changes, will be May 15 instead of March 15.  Note that the regulations for MCIT have already, technically, been finalized for Code of Federal Regulations, all ready to launch (42 CFR 405.601ff).  However, perhaps because of this new rule, the January "final regulations" for MCIT don't appear in the eCFR (here).



Update, March 13-15

  • See my four-minute video explanation at YouTube, here.
  • On March 10, NEJM published a four-page critique of MCIT, by Rathi, Johnston, et al.  My original blog had cited a Nature Biotech paper by Johnston et al., critical of the Breakthrough program.   
    • The March 11 NEJM critique is found here.
    • RAPS summarizes NEJM here.  MedTechDive summarizes here.
    • See 11/2020 MEDPAC comment on MCIT here.
  • Fierce Healthcare here.
  • RenalytxAI (Publicly held small lab company) here.
    • As of Monday 10 ET, stock had only dipped a few percent.
  • See Blog on LinkedIn by AdvaMed, March 23, here.



__ Original Blog Continues___

CMS Adds Some Comment

The publication today was not strictly "pro forma."  The administration gives several types of concerns.  

Operationalizing the Rule.  For example, CMS notes the rule might be hard to operationalize, since it promises coverage "from the first day of FDA approval."   I would have assumed, or the rule stated, that existing coding and payment rules apply, and if a new code was needed, it would take a few months and providers would have to hold claims.  For example, NCDs are effective "the day they appear" but CMS often takes months to issue full-fledged codes or coding rules. On the other hand, if a device was bundled, it would be "covered" but without any net impact.  CMS adds that as the implementation drew closer, it realized it had "underestimated" operational challenges.  

Bruce's takeaway:  I assumed "coverage" would be first-day but "claims" would take several months to implement, sometimes.  But that has occurred in other situations.  A lot of things might be "covered" but no added payment to a outpatient or inpatient bundle.

DME Rule.  CMS also cites some overlapped with a proposed, but not finalized DME rule that appeared last fall.

Bruce's takeaway:  I've read the DME rule and didn't see much conflict.

Volume.  CMS notes that breakthrough device volume was estimated, in the rule, by CMS, at 2-5 devices per year.  This made no sense to me, since a July 2020 publication had reported there were 200+ devices in the MCIT pathway already at that time.  CMS cites public remarks by FDA in February 2021 that 400+ devices were in the breakthrough pathway.  FDA here.

Bruce's takeaway:  Yup, the CMS estimate of 2 devices a year seemed oddly low.

Bruce's takeaway:  The FDA devices annual report page is worth seeing on its own.

Validation in the Elderly; first do no harm.  CMS notes that some devices might not be well-validated in the elderly and there could be "patient protection issues" which would point toward non-coverage of the device.

Did your complaint get ignored? CMS notes that in the original public comments, some parties had asked for more details on implementation.   CMS gives   any negative stakeholders a chance to assert that their negative views were not properly discussed, aired, and dealt with in the CMS final rule.

The rule specifically cites criticisms that appeared in New York Times and Health Affairs.

___


Footnotes.

Diagnostics Are In (for now).  Whereas the August rule proposal had been ambiguous about Diagnostics coverage, the final rule in January made clear it applied to all FDA breakthrough devices, including diagnostic ones.

NEJM Article.  After my original March 12 blog, I became aware of a March 10 paper in NEJM critical of MCIT, Rathi, Johnston et al. here.


Two recent Breakthrough devices.  As I was researching this for new headlines, up popped two press releases in recent days.   Anuncia received FDA breakthrough status for review of its Anuncia Reflow Mini, CSF shunt; and Inovise received FDA breakthrough status for review of its Inovise "Audicor" remote heart failure management system.






Very Brief Blog: Foundation Medicine CDx LBx Test Gets "ADLT" Status

This quarter,  Medicare added a 9th test to its list of "ADLT" tests - advanced diagnostic laboratory tests. ADLTs are a specially defined type of lab test whose code is priced initially at its market list price (assuming it hadn't been previously priced on the CMS fee schedule) then annually repriced to its market median price.

The ninth test is Code 0239U, Foundation Medicine Liquid CDx test.  It was approved 1/25/2021, with its "New ADLT Initial Period" to run 4/1/2021 to 12/31/2021.   Prior to 4/1, the test is "contractor priced."  The ADLT price will be $3,500.  

Historical note - The initial paraffin-based FMI CDx test was the "first ADLT," with code 0037U, and granted ADLT initial pricing status from 7/1/2018 to 3/31/2019. 

ADLT 0239U was applied-for July 2020, released October 2020, effective January 2021.  The FDA approval was October 26, 2020 (here).  It was an FDA "breakthrough" device.  The  64 page safety summary P200006B is here.  An additional November 6, 40 page safety summary P200016B is here, adding genes.  (I believe that FDA has to issue new "P" PMA numbers rather than updating the original - ?).  The ADLT application at CMS would have waited til the actual FDA approval dates, I think, at which point it was too late to make the deadline for a January 1 kickoff ADLT pricing date. 

See the updated ADLT price list here:

https://www.cms.gov/files/document/advanced-diagnostic-laboratory-tests-under-medicare-clfs.pdf  

Quirks of ADLT

  • Don't Mis-Time your ADLT Application versus the CMS Annual Normal Pricing Process

The ADLT pricing process is driven by the creation of a CPT code.   If a lab test has a CPT code (including a PLA code) that completes the annual new code pricing process before it completes the ADLT process, then the CLFS fee schedule price will apply rather than the rule regarding the new ADLT's list price.   So usually, you want to time your code creation to ensure it will finish the ADLT pricing process before CMS handles the code through the traditional gapfill/crosswalk annual process.

  • Tricky Way Initial List Price Is Defined

Due to tricky ways that the ADLT regulation is written, the initial 9 month ADLT price will be the lowest price publicly available on the first day the test is available for ordering.   For example, as I read it, if your price on Day One is $3000 list, $500 for uninsured patients, then your ADLT initial price will be $500.   

On the other hand - as I read it - if your Day One price is $3000 list, and on Day Three you add a $500 uninsured price, only your Day One price is used to set the ADLT initial price at $3000.  

See the regulation at 42 CFR 414.522, but paying extra-special attention to the definitions of the wording found there.

  • Some ADLT Initial Prices Have Been ReSet by the Annual Process

ADLT's are priced annually, based on reported market prices, but it's a slow process.  As I understand it, for example, 1H2018 might be a claims collection period for the ADLT, 1Q2019 a reporting period, and January 1, 2020, the new ADLT price date.   So it's annual, but slow.

Two ADLT prices have been reset downward by the PAMA annual process, with Myriad test 0172U Myriad My Choice CDx [FDA] dropping from $4040 initial to $3030, and Myriad test 0090 MyPath Melanoma dropping from $1950 to $1755.

click to enlarge






Very Brief Blog; MOLDX Publishes Article Defining "Algorithm" Valuation in Lab Tests

CMS releases new local contractor policies and articles on Thursday, and yesterday, MolDx released an article with the concise title, "Algorithm Definition."  It's logged as CMS article A58650.

Find the link here:

https://www.cms.gov/medicare-coverage-database/details/article-details.aspx?articleid=58650&ver=2&bc=CAAAAAAAAAAA

I've cut-pasted the article in full within this blog.  It has effective date, 3/11/2021.

Other Definitions of Algorithm

CMS also defines algorithm for a lab test by law at 42 CFR 414.502, but it's beyond my scope to assess if there's a meaningful difference between 414.502 and A58650.  I've clipped the 414.502 definition at bottom.

AMA defines algorithm in the CPT code booking regarding MAAA tests - I've also clipped this at bottom.

International Group - Polygenic Risk Scores

In related news, The Clinical Genomics Resource Complex Disease Working Group has published a paper in Nature on defining polygenic risk scores.  Nature here, Genomeweb here.


Palmetto MolDx Article Text:


This contractor defines an algorithm, as a distinct component of clinical laboratory processes, as follows:



An algorithm may be considered a clinically valuable and independent component of a laboratory process when ALL the following conditions are met:

  • It is an unambiguous problem-solving operation that includes deploying a set of rules or calculations requiring computer processing;
  • The test result (or a component of the result) is the calculated output of this process, and not an intermediary process;
  • The same or similar test result could not be obtained without the use of this process;
  • The input for the computation is derived from biological samples using analytical processes, and must include data from the sample submitted for the test;
  • The process must:

Either be required for the analytical result, OR

If adjunct to the analytical result as a post-analytical process, the calculation itself must be independently found to be reasonable and necessary apart from the other components of the test.

Examples. [Note that examples 2,3,4 are of "non-algorithms".]

  1. A gene expression profile test wherein sequencing data must be compared in a calculation to an existing and validated set of profiles to bin it in one of several possible risk stratification groups would require the use of an algorithm as defined above.
  2. A next generation sequencing (NGS) test that uses computation to identify variants in a sample is not considered as using an algorithm in this context. The calculation in this scenario is seen as an intermediary process.
  3. Calculations using only clinical information not derived from analytical services on biological samples are not considered algorithms in this context. Examples would include using the clinical information from the patient in a calculation to assess their risk stratification or using a similar process to identify relevant clinical annotations derived from literature as associations with sequencing variants.
  4. A test that inputs resultant analytical processes that are reasonable and necessary (such as gene variants or protein markers) that are post processed by computation, but wherein that subsequent computation is not independently established as reasonable and necessary above and beyond the other lab components, shall not be considered an algorithm as a valid component of a laboratory test.

###

Other sources with definitions of algorithims include (A) Medicare statute, (B) Medicare regulations, and (C) AMA CPT.

(A) Statute Definition

SSA 1834A

  • The statute states only that an algorithm is used and yields a single and patient-specific result.

"The test is an analysis of multiple biomarkers of DNA, RNA, or proteins combined with a unique algorithm to yield a single patient-specific result."

(B) Regulation Definition


42 CFR 414.502 (ADLT law)

  • The regulation adds a few terms such as "predicts the probability" and "will develop" and "will respond to."

Advanced diagnostic laboratory test (ADLT) means a clinical diagnostic laboratory test (CDLT) covered under Medicare Part B that is offered and furnished only by a single laboratory and not sold for use by a laboratory other than the single laboratory that designed the test or a successor owner of that laboratory, and meets one of the following criteria:
(1) The test -
   (i) Is an analysis of multiple biomarkers of deoxyribonucleic acid (DNA), ribonucleic acid (RNA), or proteins;
   (ii) When combined with an empirically derived algorithm, yields a result that predicts the probability a specific individual patient will develop a certain condition(s) or respond to a particular therapy(ies);
   (iii) Provides new clinical diagnostic information that cannot be obtained from any other test or combination of tests; and
   (iv) May include other assays.
(2) The test is cleared or approved by the Food and Drug Administration.


CMS further discusses the meaning of algorithm at 81 FR 41057ff (June 23, 2016).

That the analysis of the biomarkers 
must be ‘‘combined with a unique 
algorithm to yield a single patient- 
specific result’’ indicated to us that the 
algorithm must be empirically derived, 
and that the ultimate test result must be 
diagnostic of a certain condition, a 
prediction of the probability of an 
individual developing a certain 
condition, or the probability of an 
individual’s response to a particular 
therapy. Furthermore, the statute 
requires the result to be a single patient- 
specific one, so we proposed that the 
test must diagnose a certain condition 
for an individual, or predict the 
probability that a specific individual 
patient will develop a certain 
condition(s) or respond to a particular 
therapy. 

...Laboratories would have to 
submit to CMS evidence of their 
empirically derived algorithms and 
show how their test provides new 
clinical diagnostic information that 
cannot be obtained from any other test 
or combination of tests. 

...We considered the 
commenters’ suggestion to use only the 
exact statutory language and not define 
unique algorithm as we proposed to do. 
However, we do not agree with this 
approach for the following reasons. 
First, using only the exact language of 
the statute would leave the public 
without any specific guidance on how 
to interpret ‘‘unique algorithm to yield 
a single, patient-specific result,’’ and 
would leave us with no criteria by 
which to evaluate whether a test meets 
that requirement. Second, without such 
criteria, the requirement that a test have 
a ‘‘unique algorithm to yield a single, 
patient-specific result’’ would be, to 
some extent, self-determined by each 
laboratory requesting ADLT status. 
Without specific guidance, the 
laboratory seeking ADLT status would 
interpret the requirements under 
criterion A in whatever manner it chose, 
which could potentially vary depending 
on the test, and which could also vary 
from other laboratory interpretations. 
Third, if not further defined, the 
criterion could apply very broadly to 
nearly any test on the CLFS that is only 
done by one laboratory, which would be 
inconsistent with our view that ADLTs 
are innovative tests that are new and 
different from any test already on the 
market. Therefore, we believe it is 
necessary for us to interpret what it 
means for a unique algorithm to yield a 
single, patient-specific result, and to use 
that interpretation in establishing the 
requirements a test must meet to qualify 
as an ADLT. Additionally, as noted 
previously in this section, we are 
revising criterion A of the definition of 
an ADLT to include protein-only tests. 
However, we continue to have concerns 
about granting ADLT status for protein- 
only tests that are not advanced tests. To 
that end, we believe our proposed 
application of the unique algorithm 
requirement ensures that simple protein 
analyses would not be considered 
advanced tests as they are not likely to 
produce a patient-specific result that 
cannot be provided by any other test. 
For the reasons discussed previously 
in this section, we are finalizing our 
proposal for the unique algorithm, and 
will reflect it in the definition of ADLT 
under criterion A as proposed. 

PAMA final rule fact sheet

Medicare also has an ADLT application form that involves reporting and describing the algorithm, but doesn't add anything to the definition of algorithm, it just repeats the language seen above.



(C) AMA CPT Definition of Algorithm (MAAA)

"In contrast to genomic sequencing procedures and other multi-analyte assays, the assays in the [AMA CPT MAAA section] represent algorithmically combined results of analyses of multiple analytes to obtain a risk score or other value which in itself represents a new and distinct medical property of independent medical significance relative to the individual component test results in the clinical context in which the assay is performed.  ...Typically reported as a numeric score as or as probability.   MAAAs are typically unique to a single laboratory or manufacturer.

Wednesday, March 10, 2021

Very Brief Blog: New Splash Screen at MolDx "DEX" Website: Prices are Displayed

 Log on to the Palmetto MolDx Diagnostics Exchange this week, and you'll see this splash screen:

click to enlarge

(To get into the DEX website, you need to log on with an email and create a password).

The Price Field

If you look up a lab test, there will be a new field, in "beta," which is the price. 
  • For some tests, the field says "See Fee Schedule."  
  • In some cases, like PGx panels, you have no idea what the coding is, so an instruction "see fee schedule" is not informative.
  • Generally, when a test is not covered by MolDx, it does not show a price.  
The Coverage Field (Bottom)

For some tests but not others, MolDx also lists a data section called "Coverage."   This will state a test is covered by Medicare, or not, but sometimes it is blank and sometimes the "Coverage" field is simply not there.   MolDx informed me that this depends if the test originally got a Z code with the intent to have coverage, or registered solely for a Z code.  In the latter case, the Coverage tab at the bottom is missing and gives no information either way.

Example: Allosure Coverage and Pricing Look-Up

Let's take a simple example of the database in use.  The CAREDX ALLOSURE test is code by generic code 81479.   It's covered by a MolDx LCD.   The price field gives a price for ALLOSURE.  ($2,840).   
Down at the bottom, there is a tab for COVERAGE and it is stated to be covered.  

click to enlarge

In an early version of this blog, I had looked up Myriad Genesight and I was very surprised there was no price.  On further review, I discovered there are multiple entries for Genesight with slightly different names, in both Ohio and Utah, some priced, and some not.


Part of a larger pattern?

I've heard that several different MACs have been posting their Category III code prices, which in the past they generally didn't do.  (This recently resulted in a famous headache for iRhythm - see article.)   Possibly CMS has encouraged MACs to post internal non-fee-schedule prices, when available.   I know that Noridian, among others, has always listed internally-set prices as being available to the public, by FOIA upon request.

Quirks

Based on work I've done in the past, and based on Noridian FOIA requests, there were about 1000 MolDx locally-priced codes.   All other tests were either (A) at fee schedule codes and prices, or (B) not covered and thus not priced.

Regarding price, the results of the DEX search can still be cryptic.  For example, the Advanced Genetics lab in Connecticut has a "AGL Cardiology" pharmacogenetics panel test that is listed as "Covered" and "See Fee Schedule," but one has no idea how it is coded, so, you can't actually look to a fee schedule anywhere to find the payment.

click to enlarge

Some other little mysteries, puzzles, or items of interest.

  • The Foundation Medicine liquid biopsy test is listed in italics, and it's not clear what this means.
  • Some tests like FMI HEME test, are carefully coded by MolDx as 81479, unlisted code, but carefully priced at exactly the fee schedule price to the penny (for 81455 (50+ tumor genes)) typically used by other insurers and MACs. 
  • The CARIS BRCA test is listed as covered at $2380, although the more common coding would be 81162 for about $1800.
  • I've seen tests listed as NOT COVERED in MolDx DEX, while the same lab directly was told their diagnostic test was out of scope of review by MolDx.  This seems like a Catch-22?
Some screen shots used to develop this blog in the cloud here.


 

Brief Blog: Perkin-Elmer Publishes Pivotal Clinical Paper on Low-Pass NGS-Based Cytogenomics; Chromothripsis; Update OGM

A few weeks ago, I had the chance to write and post a 25-page white paper on advanced genomics coding and reimbursement, focusing on the evolving position of PLA codes - here.

Looking at genome-wide services, like exome, genome, and cytogenomics, I noticed something unusual.  There were at least three codes for germline (constitutional) cytogenomics using NGS, but there was no corresponding Category 1 CPT code.  This is in contrast to various codes for pharmacogenetics, hereditary panels, exome or genome, where the PLA codes are at least related to an existing CPT code.  IO was pulled back to this topic last week, when AMA posted proposed CPT codes for May 2021, and there is now a proposal for a low-pass NGS Category I code (here).

Interested parties can see the AMA CPT proposal here and follow instructions to request permission to comment by March 12.


Coding Today

There are several cytogenomics codes today, but all specify "microarray" cytogenomics as the technological method.  These are 81228, 81229 (microarray with SNPs), and 81277 (tumor microarray), all designed to report chromosomal abnormalities rather than point mutations.   They're priced by CMS at $900, $1160,. and $1160.

In the PLA world, there are codes for sequencing-based cytogenomics, of which I count at least three for germline tests, being 0012U (Mayo), 0156U (NY Genome Center), and 0209U (Perkin-Elmer).  The Mayo code is priced by CMS at $2515, while the NYGC and Perkin codes are newer and still in the 2021 gapfill process at the MACs.

In 2020, ACMG published technical standards for interpretation and reporting of constitutional (germline) copy-number variants (Riggs et al.), and remark, 

"Tremendous strides have been made in understanding the effects of copy-number variants (CNVs) in both affected individuals and the general population...continued broad implementation of array and next-generation sequencing–based technologies will expand the types of CNVs encountered in the clinical setting."   

(I recall the first time I read a headline about NGS-based chromosome analysis 6 or 8 years ago, and in the interim it sounds like it's clearly come of age.)


New Publications

New to me is Chaubey et al., J Molec Diagn 22:823-40 (2020).  This team at Perkin-Elmer assessed 409 cases for microarray and low pass genome sequencing for CNV, and suggest that NGS methods are economical at scale and potentially more accurate (depending on copy depth; they use 5X) than microarray.  Find it open access here.

Chaubey et al. conclude,

Several cases with pathogenic CNVs were detected that were missed by CMA. This study demonstrates that LP-GS (5X GS) was able to reliably detect absence of heterozygosity, microdeletion/microduplication syndromes, and intragenic CNVs with higher coverage and resolution over the genome. Because of lower cost, higher resolution, and greater sensitivity of this test, our study in combination with other reports could be used in an evidence-based review by professional societies to recommend replacing CMAs.

A lab that is new to me is Gencove, whose website headline is "Industrial-scale genome sequencing."  They cite collaborations with Broad, Coriell, and others.  The website states, "Gencove’s low-pass sequencing platform is setting the new standard for high-throughput genomics research and diagnostics applications."  

See a new open-access publication Li et al. using 0.5X and 1X LP-NGS and published in Genome Research, February 2021 - here.   These authors summarize, "We conclude that low-pass sequencing plus imputation, in addition to providing a substantial increase in statistical power for genome wide association studies, provides increased accuracy for polygenic risk prediction at effective coverages of ~0.5× and higher compared to the Illumina Global Screening Array."




####

Traditionally in tumors, we've used NGS to detect mutations or indels in 50 to 500 genes.  Another aspect of tumor biology is "chromothripsis," the widespread disruption of chromosomal structure in some cancers.  See a March 2021 paper in Nature by Shoshani et al. here.   See a review on cell biology of chromothripsis by Kolb, 2021, here.  See a new March 2021 paper in NEJM on the need for whole-genome sequencing in myeloid cancers, Duncavage et al., here.   Put low-pass NGS (0.25X) together with chromothripsis in a myeloma paper from Belgium by Rengifo, 2021, here.  Separately, see Zhao et al. 2021 on chrotin modification in cancer here.

####

Update May 18, 2021.   We've talked here about microarrays and low-pass NGS.   Another entry in the field is "optical genome mapping" - OGM - not to be confused with OMG - see a review on Bionano's work in this space in Genomeweb, May 14, here.

###

For a retrospective on how much we already knew on tumor chromosomes in 2003, see Albertson, 2003, Nature Genetics, here.


Sunday, March 7, 2021

Very Brief Blog: NYT Summarizes Fast Spread of COVID Variants

Very brief blog.  On March 6, 2021, NYT published an excellent article with first-class graphics on the spread of COVID variants in the US.  Multiple references to work performed under a federal contract by HELIX.   In addition, the article references an online 24 page PDF (Feb 2021) from the Johns Hopkins Center for Health Security.

NYT here -

https://www.nytimes.com/interactive/2021/03/06/us/coronavirus-variant-sequencing.html

And Johns Hopkins 24 page PDF here -

https://www.centerforhealthsecurity.org/our-work/pubs_archive/pubs-pdfs/2021/20210216-covid19-variants.pdf

JH PDF (Feb 2021)

NYT Graphic - March 6
Take Home: B1117 Mostly 20% by March 3 In All These States

___

For those with a Genomeweb subscription, see an article based on the current AGBT conference with COV-SEQ discussions by the Wellcome Institute in the UK and the British Columbia Center for Disease Control - here.

Nature has an open-access article on COVID variants in the US - March 5, 2021 - here.