Friday, November 30, 2018

Very Brief Blog: GAO Releases Report on CMS Lab Overpayments and Risks

GAO has issued an unusual report that implementation of PAMA lab pricing "May Lead to Billions in Excess Payments."   This will probably be a head-spinning conclusion, to anyone watching lab payment rates fall annually under PAMA.

UPDATE Feb 2019:
ACLA and Advamed complained about the report, here.  See 360DX article in Feb 2019,  here.
  • The GAO report home page is here.
  • The one page summary is here.
  • The full 47 page report is here.
    • See comments in response, from HHS to GAO, page 38-40.
    • Footnote 67 on page 30 references a House report language request to CMS that I cited in a blog on August 31.
Update:  Grassley asks CMS to follow up on the GAO report, January 2019, here, here.

I think I was one of the first to report in 11/2017 on the potential implications of CMS dropping its longstand practice not to pay more for clinical chemistry panel components than it would pay for the clinical chemistry panel.  GAO picks up this same issue in its summary now in 11/2018. In addition, their Figure 6 page 29 almost exactly replicates the figure I published in the bottom of my blog in 11/2017.
CMS stopped paying a bundled payment rate for certain panel tests (groups of laboratory tests generally performed together), as was its practice prior to 2018, because CMS had not yet clarified its authority to do so under PAMA, according to officials. CMS is currently reviewing whether it has the authority to bundle payment rates for panel tests to reflect the efficiency of conducting a group of tests. GAO estimated that if the payment rate for each panel test were unbundled, Medicare expenditures could increase by as much as $10.3 billion from 2018 through 2020 compared to estimated Medicare expenditures using lower bundled payment rates for panel tests.

In addition, I have previously published (e.g. in August 2018) panel pricing graphics that almost exactly match some later used by GAO.

click to enlarge.  Quinn graphic left, later GAO graphic right.

Very Brief Blog: CMS Releases First Transmittal on NCD for Cancer Tumor Testing

In March 2018, CMS released its ground-breaking National Coverage Determination for next generation sequencing tests used in oncology. 

CMS has released its first official "transmittal to contractors" (MACs) on how they should implement the NCD.   This transmittal primarily replicates the effective coverage text of the NCD, and includes a long list of oncology codes that support payment under the NCD.   (The ICD-10 codes must be associated with patient qualifications, such as having advanced/metastatic cancer.)
  • Download the November 30 transmittal here
    • The PDF was very slightly updated after its initial release to clarify that one long set of ICD10 codes in the back (all solid tumor oncology codes) are for FMI CDx, and a one page set of lung codes is for Thermo Oncomine 0022U.
  • Since the NCD does almost nothing except cut/paste/replicate the NCD text, along with the full cut/paste of the ICD-10 oncology code list, it's unclear why it took eight months to be issued.

Workload and task instructions in the transmittal at 10878.4, 10878.5, request MACs to work collaboratively to ensure consistent editing and attend "up to 4 one hour calls" to discuss implementation of the NCD and to provide consensus recommendations on implementation in a "final report to CMS" (10878.5).

What This Is Not

This is not detailed claims processing instructions.  CMS states those will be released in the first as amendments to the CMS "Claims Processing Manual."   CMS says point blank in the introductory paragraph that "A subsequent CR will be released at a later date" with more claims processing instructions.

The NCD has a number of implications that are subject to interpretation, or on which experts disagree.  My best guess is that the Claims Processing Manual rules, when they come out, will generally quote from the existing NCD and may not resolve alot of the remaining fuzzy issues.

What It Says Re LCDs

Regarding coverage by LCDs, the transmittal says that:
Medicare Administrative Contractors may determine coverage of other diagnostic laboratory tests using NGS for patients with cancer only - 
when the test is performed in a CLIA-certified laboratory, ordered by the treating physician and the patient has: 
either recurrent, relapsed, refractory, metastatic, or advanced stages III or IV cancer; and,• either not been previously tested using the same NGS test for the same primary diagnosis of cancer or repeat testing using the same NGS test only when a new primary cancer diagnosis is made by the treating physician; and,• decided to seek further cancer treatment (e.g. therapeutic chemotherapy).
A diagnostic laboratory test using NGS is non-covered when cancer patients do not have the above-noted indications for cancer. 

MACs have covered NGS tests in women with breast cancer & and high family history, and Lynch testing in colorectal cancer, including when the cancer is not metastatic, so MACs will have to decide how to apply the above text in those situations.  (To not cover BRCA panel or Lynch panel in non metastatic patients can't possibly be an intention of the NCD authors, though, since they reviewed only CDx issues in cancer patients and didn't even touch the literature on BRCA panels or Lynch testing, so the effect of the NCD shouldn't pertain to those areas.) 

There are also puzzles such as a new first presentation of advanced leukemia, which isn't literally "recurrent" or "metastatic" or not known to be "refractory" since it hasn't had first-therapy yet, the words found in the bullet-points of the NCD.  But is surely "advanced cancer," the heading of the NCD as a whole.

The NCD covers gene panel testing in some situations that might seem surprising, such as if you have a very small, but recurrent, squamous skin cancer of the trunk (code C44.509).

CMS classifies some high-cost NCDs as payable by Part A/B rather than Medicare Advantage.  I haven't seen a statement that CMS has classified this NCD that way.

Tuesday, November 27, 2018

Very Brief Blog: HHS Floats New Part D Pricing & Policy

Briefly, on November 26, 2018, CMS released proposed new rules that would somewhat raise competitive pricing pressure among Part D plans, and shift some existing policies such as those surrounding protected drug classes like HIV drugs.
  • Trade press on new Part D policy here.
    • Pharma stakeholders argue:
    • Part D already handles protected drug classes economically, here.  (Study by Avalere).
    • However, DOJ continues to act against Part D suppliers for illegal patient supports and rebates, here, here.  $360M settlement regarding Tracleer and other oral drugs.
  • CMS announcement on new Part D policy here.
  • Proposed rule in full, here.
    • Comment open to January 25, 2019.
    • Final rulemaking in spring 2019 for CY2020.
    • Rules apply to direct Part D plans as well as the many beneficiaries who get "Part D" benefits through Part C - Medicare Advantage.
    • For separate Part C rules for physician-administered drugs (aka Part B drugs) under step therapy see HHS proposals in August 2018, here.
Recall that on October 25, President Trump gave a press conference in person at HHS discussing potential, and more speculative, possible future changes to Medicare Part B drug pricing such as indexing US Medicare prices to European prices; entry point here.  

The proposed rule, like the October announcement, are keyed back to the May 2018 "Trump Administration Blueprint to Lower Drug Prices."  Here, here, here.

Monday, November 26, 2018

Very Brief Blog: National Academy of Medicine Report on Genomic Disparities

The National Academy of Medicine runs an ongoing roundtable of events on "Genomics and Precision Health" - home page here.

On November 14, 2018, they released the proceedings of a June conference on "Understanding Disparities in Access to Genomic Medicine."   (I had the chance to chair one of the panels).

  • The conference webpage is here, including video archive and presentations.
  • The conference report is here.
    • Download PDF ebook for free.


In other news, in recent weeks, NAM released a new ebook on "Harnessing Mobile Devices for Nervous System Disorders" - here - and one on "Economics of Microbial Threats," here

FDA Will Soon Float New 510(k) Guidance; Positions on Drug/Software Pairs and Bayesian Designs

Over the Thanksgiving weekend, FDA Commissioner Dr. Scott Gottlieb tweeted that FDA was about to launch new initiatives to "modernize" the 510(k) process.  

This process is a pillar of the 1976 medical devices law, which generally grandfathered preexisting devices and required review or approval for new devices unless they had pre 1976 "predicates."    An important modernization was the introduction of "de novo 510(k)," a regulatory process that is much faster than a PMA device approval.

The Monday, November 26, 2018, press release from FDA (both Dr. Gottleib and Dr. Shuren of CDRH) is here.   The modernization will focus on shifting manufacturers away from use of "old predicates" and toward use of predicates less than ten years old.  They indicate they will issue a plan for public feedback, meaning either a draft guidance or draft regulation.  No specific date is offered.  FDA notes that some of its "goals" may require "additional guidance from Congress."
  • Although not directly related to 510(k) changes, this article also includes sections further below on:
  • FDA's recent announcements about software directly tied to pharma, and 
  • FDA's recent announcements about Bayesian and other complex trial designs (CITD).  


The FDA seems to be responding to years of criticism of the FDA 510(k) system as too lax, including a burst of new documentaries and books in 2017/2018, e.g. here.   One smoking gun, hitting the public media several years ago, was fatal cancers caused by use of tissue morcellators in minimally invasive uterine fibroid treatments (here; press here on GAO report here).  Another class of issue was manufacturers who viewed existing regulations as endorsing an abilityto get by with 1976-equivalent technology.  (Let's say, as a made-up example, if 1976 glucose meters were plus/minus 20% accurate, and modern ones 5%, you could merely meet the 1976 predicate today).

Links re 510(k) 
  • New Detailed November 26 FDA press release here.  
    • 3400 words.  At 200 wpm, about a 20 minute read.  
      • The current FDA regime uses these "press releases" as substantive and detailed reviews of the agency's plans and next direction, and how it views its own process and goals.  Ignore reading the whole thing, at your peril.
    • Early open access coverage at MedTech Dive, here.  
    • WSJ here.  
    • Genomeweb here.
  • The FDA also issued an 8-page "performance report" on recent achievements and improvements in the 510(k) policy system, here.
  • Last week's November 20 FDA press release on updated device post market surveillance efforts, here.
    • Much of this document describes NEST (see next).
  • November 26, FDA seeks $46M for national health surveillance system NEST, here.
    • FDA home page for NEST, here.
    • NEST involves funding from FDA to the MDIC (Medical Device Innovation Consortium) to set up another body, NEST Coordinating Center or NESTcc.
    • See 8 pilot test cases of NEST projects, here.  
      • The scope would be greatly expanded with $46M instead of several $M.
  • FDA current medical device predicate home page here.
    • Current 2014 PDF guidance on finding and using predicates (42 pp) here.
    • FDA sets the new 510(k) ideas firmly in context of its Medical Device Safety Action Plan, floated in Spring 2018, here.
  • Gottlieb's November 25 "FDA Sunday Tweetorial" on device improvements
  • In separate messaging, such as in December 2017, FDA discussed wanted to ensure that devices could enter the market with "acceptable uncertainty" and not be overly delayed by regulatory review, here.

FDA Sports Metaphor

The November 26 press release gives Gottleib and Shuren a chance to use the (overused?) hockey metaphor "skate to where the puck will be."   "We not only have to skate to where the puck will be, we also need to drive the puck to where it should be."  Both of these risk being obvious; of course you are supposed to skate toward where the puck is going, in order to hit it; and when you do hit it, you are supposed to hit it to where you need it to be next. 


Footnote 01: FDA and Software Directly Related to New Drugs

A year back, in November 2017, I was at a digital health conference where Amgen was discussing its plan to develop drugs closely linked to software packages.   FDA has been pushing forward on that front, too.
  • See a November 19, 2018, press release on FDA proposals to link digital tools with drug approvals, here.
    • See the FDA's corresponding official Request for Public Comment on the proposal (comment open until January 22), here.
    • Trade press on the above, Medtech Dive, here. At RAPS, here.
  • Earlier, May 2018 trade press on FDA and digital devices (Covington), here.
    • Gottleib comments on digital health at Datapalooza April 2018, here.
  • Trade press on digital therapeutics, June 2018, here.  Trade press (RAPS) on FDA and Dhealth and "paradigm shift," April 2018, here.
  • FDA report on how 21st Century Cures impacts FDA medical software policy, here (December 2017, 14 pp).
  • FDA draft guidance on clinical decision support (CDS) software, here (December 2017, 13 pp).
Pre-Certification.  Separately, FDA has been crafting a "pre-certification" system for some types of health related software, with pilot announcements in 2018 and more detail to come by early 2019.   In October 2018, Sen. Warren issued a 12-page letter to FDA requesting more detail about risks of this adventure (PDF from here, trade press here, here.)

Footnote 02:  FDA and Complex Innovative Trial Designs

The 21st Century Cures Act asked FDA to provide guidance and look closely at innovative trial designs (e.g. Bayesian designs).   Some argued that Congress was trying to lower the rigor of FDA design standards, but to my eye, Congress was only asking FDA to "provide guidance" - for example, it could provide guidance it hated Bayesian designs and it would meet the Congressional direction to provide guidance.

In any case, the FDA has latched onto "CITD" - Complex Innovative Trial Designs.  FDA held a full day workshop in March 2018; webpage here.   (See decks, transcripts, video; public comment docket with 12 comments).  See post meeting comments by BIO, PHRMA, Friends of Cancer Research (totalling 44pp); here.

See the FDA homepage for CITD Pilot Program here.

Press release here (August 29, 2018).   Corresponding FDA Fed Reg announcement here.  (83 FR 44274).   Trade press here, here.  RAPS here.

RAPS on September 2018 draft guidances, a pair of them, here.  See #1 Adaptive Designs draft guidance here.  See #2 "Efficient Master Protocol Design to Expedite Development..." draft guidance here.

Wednesday, November 21, 2018






original blog follows (October 2018)

Original Title from 2018:
CMS Posts CY2019 New Molecular Coding Rules!  Envisions More, Not Less, Use of 81479

Every November CMS updates a national coding rulebook called "National Correct Coding Initiative Edits."   This include massive excel spreadsheets, but also a zip file of PDF coding instructions.

The new files for CY2019 are now posted here and have some big changes for genetic coding.   NCCI edits and rules may be used by private payers as well.


At the CMS website here, click on the download for "NCCI Policy Manual...January 1, 2019...Zip".

Inside that Zip, the file for laboratory (codes 80000-89999) is Chapter X (10).  Inside that, Section F is Molecular Pathology (p9 ff).  I've also put up a cloud copy of Chapter X here, as it was on November 21.

Revisions in Chapter X, Labs, Section F, Paragraphs 7, 8, 9

The three revisions are new section 7, section 8, section 9. 

Section 7 states that individual Tier 1 or Tier 2 CPT codes (e.g. "EGFR") shall not be reported with a genomic sequencing procedure, when the CPT descriptor for GSP procedure includes a descriptor for the Tier 1, Tier 2 analyte.   E.g. don't report BRCA testing (81162) along with a Hereditary Breast Cancer gene panel (81432).   Don't report individual Lynch genes along with a Lynch panel (81435).

Section 8 states that if a lab analyzes multiple genes by "a next generation sequencing procedure," report only one unit of service of GSP, MAA, or PLA code.  "If no CPT code accurately describes the procedure performed, report CPT 81479 x 1."   And:  "The laboratory shall not report multiple CPT codes describing the component results."   

  • My joint reading of the two rules above, if you bill the 10 or more genes for an AMA CPT breast cancer or colon cancer panel, bill the panel codes (81432 or 81435) not individual CPT codes.   If you bill some subset, like 8 of the 10 breast panel codes, do not code stack them, but bill 81479.  Presumably, a MAC would not pay more for the subset panel as 81479 than it would pay for the whole panel (81432).  This fits recent Congressional instructions for CMS to handle panel unbundling (here).

Section 9 states that the NCCI Excel tables, or procedure-to-procedure edits, describes CPT codes that "should not be routinely reported together."   They give the example of 91292 MLH1 gene and 81292 MLH1 dup/del analysis, adding, "it may be appropriate to perform dup/del testing if the disease variant is not identified by full gene sequencing."

These are summaries and for this article I'll ignore some probably very esoteric or obscure or likely unintentional readings of the text.   Readers should refer to the full CMS text.

Molecular section F, new points 7,8,9.

They also include a similar paragraph, not in the molecular section, but in the introductory section "A" on page 4:


Section 7:  Don't report CPT codes that are found inside GSP (gene panel) codes  This seems like commonsensical coding advice.   I've been puzzled in the past by CMS data that MAC payments didn't seem to follow this rule.   There is a general principle to use the best-fit CPT code rather than components that add up to the same CPT code.    However, particularly MolDx may have given lab-specific instructions in the past in unpredictable combinations. 

Section 8:  Use of 81479 rather than code stacking Here, I think the key sentence is: "The laboratory shall not report multiple individual CPT codes describing the component-test results."   The most obvious fact here is that CMS seems to be nationalizing a MolDx rule called "Genetic Test Panel Alert" that multiple CPT codes (e.g. EGFR gene + KRAS gene + BRAF gene) should not be added up one by one, but coded as 81479 (unlisted procedure, unpriced procedure).   The MolDx rule is here (and cloud archive 11/21/18, here.)

  • Nationalizing a Quirky MolDx Coding Rule.  In the past, I felt the MolDx rule (that flipped every group of several genes into 81479 coding) violated the normal national correct coding rules and concepts, but now, the NCCI CMS manual appears to be updated to more or less match the MolDx multi-gene rule.
  • Labs May Complain There Isn't Time to Do This.   Labs may complain that this rule comes out near December 1, for implementation January 1, and there isn't enough time to recode LIS and billing systems to conform.  
  • I Predict Payment Chaos Without Further Instructions.  Without a payment rule, this creates huge opportunities for unfairness.  
    • What if one lab bills EGFR+KRAS+BRAF, let's pretend they historically add up to $200+$200+$200 =$600, and gets paid $100 for this use of 81479 and the lab across the street under the same MAC gets paid $500 for the same use of 81479?  
    • Remember that 81479 is unpriced and there is no pricing rule for it.  
    • Or different MACs across state lines might bundle and reprice genes under 81479 in very different ways.  This is inconsistent with the new nationwide fee schedules, which were deliberately meant to reduce state-to-state variation in lab test pricing.
    • Plus, there is a huge manual activity burden when 81479 is used, a burden on the MACs that process the claim.
    • Plus, if all payers follow this rule, PAMA genetic pricing surveys(which is based on individual genes for lab tests and ignores 81479) go haywire.
    • Auditors can't audit.  Let's say an OIG auditor finds a lab was billing for and getting paid for three CPT gene codes in 2019, and this instruction says to use 81479.  What overpayment would OIG assign?  It's undefined.  Would they just guess?  Spin a wheel?
    • In October 2017, CMS issued an instructions to MACs to "report to CMS" those labs that were using 81479.  Here.  At the time, my best guess is that CMS wanted to centrally track those labs doing a lot of 81479 billing, perhaps to encourage them to use specific codes or get a PLA code.   (Of course, CMS could always track labs billing 81479 anyway, since CMS has access to its own claims data.)  This November 2018 instruction would boost, not reduce, claims with 81479.   
    • Data dive:  CMS CY2016 data for 81479 billing by lab can be obtained here.   Total 81479 Part B payments were about $108M.  All but about 7% went through MolDx states.  The largest 81479 biller in CY2016 was (Assurex + Myriad) at $39M (36%; Myriad acquired Assurex in mid-year CY2016).  See screenshot of the top 18 81479 billers in CY2016, here.  8 labs provided 90% of 81479 billing.  Excel in cloud here.
    • Claims for some CPT codes that might have autopaid without further editing might be slowed down (requests for records, orders, physician notes) through the 81479 process.
  • Perverse Incentive to Tilt Away from NGS Sequencing.  These rules could be an incentive to do Sanger or PCR hotspot testing, as the bundling and down-payment applies only if an NGS method is used.  
    • The same incentive to Sanger or PCR  methods  is found in the CMS national NCD for NGS testing in cancer, which makes some types of genetic testing literally unpayable unless if performed by methods other than NGS, since the NCD applies only to NGS.
Section 9:  Dup Del analysis can be a reflex test, but don't report routinely with sequencing.  This is the only new rule that doesn't cite NGS as a method.   This is actually an improvement in some ways, because until recently MolDx had a published rule against paying for dup del testing such as 81433 (breast cancer panel dup del) at all; it was on a list of MolDx unpayable codes.   Here, NCCN says that Dup Del should be a reflex test, however, in most cases -- such as BRCA syndrome testing or Lynch/Colon syndrome testing -- the sequencing code will be negative >95% of the time so you will indeed reach the Dup Del code almost all the time. 

What I'm saying is that a new rule paying Dup Del 95% of the time on reflex is better than an old rule paying it never.   I've been told that some private payers paid Dup Del coding irregularly, but this confirms a payable role for Dup Del code use and payment. 

Since the sequencing code is typically negative 95% of the time, there will be a lot of endorsed use of modifiers to pay the second code, the dup del code, in those 95% of instances, at least when no other rule like use of a comprehensive code (e.g. 81162 or 81455) is available.

While Dup Del is sometimes reflected by a separate code (whether a separate single gene Dup Del code like 81294 or a separate gene panel Dup Del code like 81436) there are other times when Dup Del is already part of the CPT code (e.g. Dup Del is already inside the sequencing codes 81162 or 81455).     

Tuesday, November 20, 2018

Medicare's Public Measure for Hospital Cost of Pneumonia Care & Mortality

Medicare has a large number of publicly available hospital measures, some of which are more intuitive than others.   (For Medicare's sepsis management measure, here.) 

An unusual one I ran across is Payment and Value of Care

Online at CMS, you can download the "CSV" file at this website (most recent, 10/31/2018).   The file is called "Payment_and_value_of_care_-_Hospital.csv."  I recommend immediately opening and re-saving the file as "Excel Workbook." 

Generally, there are four metrics per hospital:  See Column J Payment in dollars over 30 days for AMI, or for heart failure, or for pneumonia, and payment 90 days for hip/knee replacement. 

In the snapshot below, column J is the measure, column K the summary result in words (e.g. above, below, or near the national average); column L is number patients and column M is average payment.  (I've hidden some additional columns for clarity).   Then it is interesting to see Column R is a summary metric for how mortality and payments compare.   This is rated categorically in Column S (e.g. average mortality + average payment, or worse mortality and higher payment, etc).  (Unlike payment, there is no absolute number shown for mortality, only a categorical rating.)

click to enlarge
Zooming in on Pneumonia With Worse Mortality

Once you have the Excel, you can do some basic sort-and-play maneuvers. 

For example, I made a copy of the worksheet page, sorted alphabetically by measure, and deleted all the measures except the 30-day-pneumonia measure.   Then I sorted it for performance on Value of Care

This gives you information like: CMS classified 240 hospitals out of 4794 with data as "Worse Mortality" for pneumonia. 

Of these 240 all labeled with "worse mortality," CMS subcategorized by payment costs.  54 of the 240 had worse mortality and lower payment, 58 had worse mortality and higher payment (whoops), and 127 had worse mortality and average payment.   (We can infer that CMS assesses worse and better as quartiles, with 50% in the middle).    In other sources, CMS states that mortality is propensity-corrected (e.g. by advanced age). 

Payment Level Range

Sorting pneumonia 30-day measures solely by average payment, the highest was $27,126 for a hospital in Oklahoma with 35 cases, the lowest was $10,816 for a hospital in Texas with 29 cases. 

Number of Case per Hospital

Sorting solely by number of cases per hospital, the highest was 3,328 for a hospital in Orlando, FL, which was one of only 8 hospitals with > 2000 cases for this pneumonia measure.  About 1850 hospitals had >400 cases.  (Hospitals with <11 are not reported). 

Medical Imaging, Spending Per Bene

CMS also publishes data on "use of medical imaging" by hospitals (outpatients) and "Medicare Spending by Beneficiary" or MSPB for all spending by any provider in Part B or Part B that is within 3 days prior or 30 days after a hospital discharge.   Cost standardization removes geographic wage index differences, etc, and there is "risk adjustment due to patient health status."  Explanation of these, here.  Data sources online here.   (The main zip file downloads a large bundle of Excel files, 277 MB).

Friday, November 16, 2018

Very Brief Blog: Forbes Publishes Deep Dive Article on Flatiron

At Forbes, in the November 30 issue, journalists Matt Herper and Ellie Kincaid publish a deep-dive, 4000-word article about Flatiron Health.   Its all here; the intrepid young founders; the tentative startup within Google; the funding round over $100M and what they did first with it; the alignment with Foundation Medicine and soon Roche.

Online here.


See also a MedCity article on a new joint ventury called Syntropy on cancer analytics, here.

CMS Releases CY2017 SEP-1 Sepsis Management Data; Academics Perform Badly

On October 31, 2018, CMS released hospital performance data on quality measures, for CY2017.   This gave the first full year of data for SEP-1, a measure of how well hospitals live up to CMS management standards after patients have been diagnosed with sepsis.   (The measure requires several actions within six hours, such as managing the septic patient by drawing blood for culture, starting fluids, and starting antibiotics.)

I reviewed earlier partial year SEP-1 data on August 4, and the full original article is here.  The median hospital performance is 49% - meaning 49 patients out of every 100 diagnostically septic patients get the CMS protocol of interventions.   CMS adopted the measure after the agency established that protocol performance resulted in a 30% relative reduction in death rate in patients diagnosed with sepsis (79 Fed Reg 50236).

About 500,000 patients were under the measure, and it's a very complicated chart-reviewed metric that may require up to an hour per patient to classify.*  (That means at 2000 hours of work per year, 250 FTE's were squirreling away their statistics on this CMS measure.  So much for "patients over paperwork.") 

Rather than repeat all the findings in the August blog, let me highlight a few here.  For one, academic hospitals (per the US News & World Report ranking) did terribly.  Exactly half of them scored below 40% (remember that a perfect score is 100%), and three actually scored below 25%, the lowest being Vanderbilt at 15%.   (That's very close to the performance of the lowest-ranked of states or territories, Puerto Rico at 11%.)  The highest academic center score was 78%, at Baylor, followed by 72%, at NYU Langone.

For academic articles on recent SEP-1 data, see

  • Barbash et al., 2018, national performance on SEP-1.  Here.
  • Truong et al., 2019, should we drop one-size-fits-all SEP-1?  Here.
  • Rhee et al, 2018, does SEP-1 really correlate with outcomes?  Here.  Similarly Sanghvi et al., 2018, here.

How Did Henry Ford Hospital Do?

Remarkably, Henry Ford Hospital in Detroit, which is the measure steward of record for the National Qualify Forum for this metric, performed below the national median, at 48%, meaning less than half of its own patients got care that passed the CMS quality metric.   For two press releases by Henry Ford touting its role in creating the metric, now required by all hospitals seeing Medicare's 40M patients, here (2013) and here (2018).  For Forbes article on Henry Ford and sepsis measures, here.  As far as I can tell, there is no Henry Ford press release touting its actual, below-median performance.

NIH faculty criticized SEP-1 publicly earlier this year, bringing a rebuttal defending use of SEP-1 in patients from Henry Ford authors, although Henry Ford is at best below-median in actually using SEP-1 in patients.  Additionally, when CMS rolled out SEP-1, it asked the public to "recommend any changes in this measure" to "the measure steward" (HFF) rather than to CMS (79 Fed Reg 50240).

Unlike some hospitals in Michigan, like Covenant in Saginaw, which reported data on all its patients (with a higher score than Henry Ford), Henry Ford Hospital reported to CMS on only a sampling of patients that met sepsis criteria, so the exact performance in their Medicare patients can't be discerned.

The SEP-1 measure is bafflingly complex, with hundreds of pages of CMS guidance documents and - wait for it - a 254-question "Q&A" document.  See examples here.

How Did Mayo Clinic Do?

Mayo Clinic was also well below the national median.

How Did Vanderbilt Do?

I mention Vanderbilt because it has one of the highest-profile Centers for Precision Medicine in the U.S.   With a score of 15, Vanderbilt ranked all the way at the very bottom of US News' Top-20 medical schools for CMS management quality after patients are diagnosed with sepsis by CMS criteria, and also near the bottom of all US hospitals reporting to Medicare.   This ranked Vanderbilt at about #2900 of about 3000 reporting U.S. hospitals.

How did Emory do?

Emory University Hospital reported a low score of 25.  Emory has an internationally known critical care center and its director edits the journal Critical Care Medicine.  This ranked Emory at about #2740 of about 3000 reporting U.S. hospitals.

Recent Literature

Use of the measure has been criticized in some circles (e.g. here for summary and here for recent article in Annals of Internal Medicine by Pepper et al.)   For a current article in JAMA by Klompas et al. at Harvard, here.  Rhee at Harvard has published some of the first articles on SEP-1.  Rhee et al. reports that, in early data, meeting or not meeting the SEP-1 measure was not associated with any impact on outcomes in corrected data; here (trade press here). She discusses other limitations of SEP-1 relative to measuring quality, here.  (Rhee has also written about biomarkers to help manage antibiotics in sepsis; here; SEP-1 lacks any biomarkers except lactic acid for organ failure.)

Wide variation in SEP-1 performance was reported as early as January 2018 by Venkakesh, here, for which an op ed commented, "SEP-1: A measure in need of resuscitation?" (here).

The background of creating SEP-1 was reviewed by Faust & Weingart (here).  For critique in MedPageToday, here.  For a strong article on the challenges and dilemmas of implementing sepsis metrics across populations and subpopulations, here.

Does CMS Get What It Pays For?

CMS pays for diagnosing sepsis, as this can upgrade DRG payments markedly, driving by electronic medical records keyed to billing systems.  Other than posting SEP-1 performance, as discussed in this blog, CMS does not penalize hospitals if they don't do things like draw culture or start antibiotics.

The Rankings

Of the top 20 medical school medical centers in the US, only 4 even ranked in the top-fourth of the 3000 community and other hospitals reported by CMS.

For additional links and sources for the actual CMS data, entry point here.

Making It Personal

I was cheered that the closest hospital me, Olympia Medical Center in Los Angeles, and next to the preschool my kids went to, was tied for the top 6 nationally, at 99%.

Bringing back memories, I logged time at four of the top twenty.  Three of the places I trained or worked - Stanford, NYU, and Northwestern - were above the median; one (UCLA) was not.


*  CMS estimated the work burden of 3 chart based measures at 1.6M hours for US hospitals, or about 500,000 hours per chart measure.  For two chart measures, CMS estimated the work burden at 741,074 hours, or about 370,000 hours per measure.   For one other chart measure, 858,000 hours of work.  Exact numbers are unavailable, but I assumed SEP-1  (with extremely complex rules) so a 500,000 hour estimate of work is conservative.  See CMS, FY2019 final rule, 83 Fed Reg pages 41689-90, August 17, 2018.


It could be that patients at academic centers are on research protocols different than SEP-1; more recently, CMS made an allowance for this.  However, it seems like research protocols wouldn't randomize away from the basic aspects of urgent clinical care; or if they did, it means there were clinical community equipoise for the alternative treatment being used instead, in which case, hospitals should not be penalized for that alternative.

CMS Releases CY2019 Final Crosswalk Pricing

On October 19, 2018, CMS released final gapfill pricing for CY2019 lab codes that were under the gapfill process.

On November 16, 2018, CMS released final crosswalk (or gapfill) decisions for CY2019.
  • See the CMS final crosswalk PDF here (41pp). 
    • 99 codes went through this new code crosswalk/gapfill process.

A Few Results

Thermo Fisher Oncomine FDA Test
The code for the Thermo Fisher Oncomine test, an FDA-approved lung cancer gene panel test covered under the CMS NGS cancer NCD, was originally proposed at a $600 price - much lower than its circa $2000 MAC price.   The final CMS decision is to gapfill code 0022U in CY2019.   Presumably, this will memorialize the existing MAC prices for the 2020 fee schedule.   Because the NCD promises coverage to all FDA-approved tumor gene panel tests within their FDA labeling, there is likely to be an influx of tests from multiple venders (or sole source labs) popping up with FDA approval over the next 12-24 months.   This should promote competition (for example, for turnaround time or sample size or gene panel or helpful remotes) against the Foundation Medicine test, a sole-source gene panel currently priced by CMS at $3500. 

CMS also upgraded the valuation of 0040U (a high-sensitivity FDA-approved BRC/ABL1 test from MolecularMD) from 1x81206 (the legacy LDT price) to 2.5x81206.

Pan-Ethnic Carrier Screening
A new Category I CPT code for expanded carrier screening, 81443, is crosswalked to the existing code for Ashkenazi carrier screening, 81412 ($2448).

AMA CPT produced a new largely new set of BRCA codes for CY2019.   However, as public data has previously documented, nearly all current CMS BRCA testing is either under codes 81162 (for combined BRCA1-2 full sequence and full dup del, $2027 in CY2019 and $1825 in CY2020), or under a gene panel codeset 81432/81433.   New codes were created for BRCA1-BRCA2 sequencing alone, BRCA1-BRCA2 dup del analysis alone, BRCA-1 testing alone, BRCA-1 dup del analysis alone, and BRCA2 dup del analysis alone.  (A code for BRCA-2 sequencing alone previously existed and isn't changed.) 

Pricing of new BRCA codes was finalized, I believe, as proposed.   Like with other codes, CMS used codes in the Tier 2 code set as price benchmarks when possible.
  • 81163, B1B2 SEQ, is 81406+81216, or $282+$185, or $467.   
    • Code 81216, BRCA-2 sequencing alone, had a low PAMA price of only $185, which CMS is re-using here.    
  • 81164, B1B2 DUP DEL, is 81405+81406, or $302 + $282, or $584.*   
  • 81165 B1SEQ, 81406, $282.   
  • 81166 B1 DUP DEL, 81405, $302.
  • 81167 is B2 DUP DEL, 81406, $282.
New BRCA Codes Likely to Be Obscure and Rarely Used 
Prior codes, or similar codes, for these piecemeal BRCA services were virtually never used in either PAMA commercial payer data nor in CMS data. 

Since the new codes have prices that would "code stack" below 81432/81433 or below 81162, they are unlikely to be used more than very rarely.  (There is no incentive to use them, which there would have been if they code stacked to payments well above 81162). 

As far as I could tell from the relevant AMA CPT meeting in San Diego in February 2018, AMA was simply proposing to make equal and similar codes for any possible combination of BRCA1 or BRCA2 sequencing or dup del analysis, regardless of how often that service was ordered on its own or whether it had any literature on its own.**

Appeal Process

The 2019 prices can be appealed by submitting a formal appeal to CMS in the next 60 days.   The result would be that the code you appealed will simply re-enter the crosswalk/gapfill decision process again next summer.


Appeals Fizzle

Two codes were under appeal (81326 PMP22, and 81334 RUNX1) and CMS declined to change existing pricing.


Gapfill = 18 Codes, Or About 18% of New Codes

Using a wordsearch method across the 99 codes, I counted 18 under gapfill (0018U, 0019U, 0021U, 0022U, 0023U, 0029U, 0030U, 0035U, 0041U, 0042U, 0043U, 0044U, 0048U, 0050U, 0053U, 0055U, 0056U, 0057U.)   That's about the same number as were under gapfill this past season.  In some cases CMS used gapfill because there was truly no comparable test; in other cases, because CMS was concerned that code stacking of CPT codes might not reflect economies of scale.   

Your Little PLA Code Could Be $3873!

Several codes were finalized with a crosswalk to 81519, the Oncotype Dx test, including PLA codes that received this favorable pricing.


* Note that the Tier 2 system is quirky.  Each tier reflects more extensive sequencing than the tier below, but there is a rank order anomaly between 81405 at $302 and the higher more complex code 81406 at $282.

** BRCA coding is now like those Staples ads, "Yeah, we've got that."  AMA, you got a code for BRCA-1 sequencing alone?  Yeah, we got that.   OK, AMA, you got a code for BRCA-2 dup del analysis alone?   Yeah, we got that.

Monday, November 12, 2018

Next AMA PLA Conference Call on New Codes: November 19, 2018

Get the holiday Thanksgiving week off to an exciting start by listening in on the quarterly AMA PLA Applications public conference call, November 19, 2018.

With a warning that these links may be updated by AMA, the home page for PLA code policy is here  The listing of current long descriptors (circa 79) as finalized in September for October 1, 2017, is here.

The next public panel call is November 19, 2018, at 7 pm CT (8 pm ET).  Register by Friday November 16 via thilani (dot) attale (at)  .   This is also on the PLA home page cited above.

The Public Agenda for this November PLA meeting is here.  There are 7 new codes and one minor text revision on the docket.  If I read the calendar correctly (here). this round of applications was due October 9 and posted publicly by AMA on October 16, with a window until October 23 for public comment submissions.  The PLA subcommittee voted on November 6, which is not a final AMA vote but rather, sends recommendations forward to THIS public hearing on November 19.   You can listen in, just like you can attend and watch a CPT meeting, but often these final public calls are pretty brief ("Questions?  Hearing none, please vote.") 

These 4Q2018 PLA codes are finaled and public on January 1, but I believe CMS requires a quarter of cycle time before adopting them on April 1.


I noted in a previous blog that this PLA cycle is sort of like a solar eclipse - it is the shortest possible cycle time between getting a PLA code and getting a market based PAMA rate.   Your code will be active in Q2 2019 (for CMS purposes.)   Let's say you envision a market price of $1000, but in July 2019, CMS crosswalks your code to $200.    That $200 price will hold for CY2020.  However, in early 2020, CMS will collect your market price data - let's say it is $1000 per payment for your private payers during Q2-2019.   That private payer rate of $1000 will then become your CMS PAMA payment in CY2021, 2022, 2023.   You will only be exposed to the CMS low crosswalk price of $200 for one year (CY2020).    This is the shortest possible exposure to undervaluation, according to current PLA and PAMA cycles.

Tuesday, November 6, 2018

Very Brief Blog: Google AI System Speeds Surgical Pathology Work

In the field of radiology, Qure AI has raised over $30M to bring AI methods into radiologic diagnostics (here), and the impact of AI on radiology is widely discussed (here, here, here).

The Dark Daily news blog this week highlights a publication in Am J Surg Path finding that pathologists self-report faster and easier detection of lymph node micrometastases with AI-assisted vision.   Dark Daily here.  (Includes numerous links of further interest).   The new article is open-access at Am J Surg Path; see here by Steiner et al.

It's AI work coming of Google/Verily.   Verily has a college-campus-sized set of glass buildings in South San Francisco, just north of SFO and just west of the Genentech campus.


For a trade journal article on pharma investments in digital, here.

For November 2018 trade press on "Deep Lens" and affordable AI for pathology, here.

For a November 2018 article in APLM on new FDA approval for whole slide imaging, Evans et al., here.

Very Brief Blog: Scott Gottlieb Discussed FDA's New Vision at Milken Institute (October 2018)

On October 24, 2018, as part of the Milken Institute Future of Health Summit, Scott Gottlieb gave a detailed interview on his vision for the FDA, interviewed by Tanisha Carino, executive director of Faster Cures.

See the Milken agenda here.  See a detailed transcript and links to the Youtube video interview here.


Gottlieb, a physician with a background ranging from federal policy to investing, and a cancer survivor himself, dicussed keeping FDA focused on a "values oriented strategy" and one that incorporates patient perspectives.

He noted that FDA will soon release new guidance on incorporating digital apps and monitors into drug approvals.  (See also an open access article at ISPOR, here.)  He noted that digital tools could also have post-market or post-regulatory roles.   (For a November 2018 trade journal article on pharma investments in digital, here.)

He discussed the new for new regulatory tools in areas like neurologic diseases, and improving cognitive tools in Alzheimer's. 

He noted that FDA was restructuring the drug pipeline and wants to move things like filling spreadsheets away from expert medical reviewers and let them focus on high-impact medical decision making during the review process.  He also noted that the FDA needs much better knowledge management systems in house.   This is the classic dilemma in organizations, "How do we know what we know?"  The agency is also committed to more objective public disease-specific guidances (less need to rely on tacit experience or to reinvent the wheel.)   The agency will create "problem based review memos based around "critical questions."

He discussed the wave of pending retirements and that the FDA is a highly technical working agency, not just one that assigns and passes through money.

He referred to the agency's commitment to areas of self-regulation or self-certification (e.g. in some areas of digital health) and to the 70 pages of comment FDA recently submitted to Congress proposing new legislation for regulating LDTs.

There was a discussion, pivoting off of a Harvard economist Amitabh Chandra presentation, the R&D in pharma wasn't well aligned to major public health needs (e.g. too directed to orphan diseases rather than pneumonia and heart disease).   Some disease areas are piled high with "good-enough generics" that are a barrier to new research.

He noted, as he has in several recent speeches, that sometimes we don't get reimbursement right, and that is a barrier to progress as well.   He raised the concern that some areas may get stuck with "natural monopolies" of one or two drugs and it's too hard for more diverse new entrants to get a foothold.

Very Brief Blog: Resources on Pharma Digital Disruption (McKinsey); Opioid Addiction Digital Disruption (Rock Health)

Citations to two detailed white papers today.

Digital Disruption in Pharma 

McKinsey provides a whopping 204 page collation of its articles on digital disruption in the biopharma industry, online here.

Digital Disruption in Opioid Addiction

The Hill passed opioid addiction funding and legislation in October - 98 to 1 in the Senate, 393 to 8 in the House (here).  Funding breakout here

Rock Health provides a very up-to-date 12-page (or 50 iPhone screen) review of digital medicine and startups in the field of addiction treatment, online here.  

Friday, November 2, 2018

Very Brief Blog: CMS Publishes Notice of Future Rule on Innovative Medical Devices

Update - See also article in MedTechDive, here, October 2018.

Over the past several months, summer/fall 2018, there have been several "tips" dropped in the press that CMS is planning to produce a rulemaking change to encourage faster coverage of innovative devices.
Let me fill in some back-story.  A full year ago, back about 12-18 months, there were rumors that Advamed and other stakeholders and CMS might shortly come up with a new special program for breakthrough devices and rapid coverage.  But the topic went radio silent, at least on the CMS side, around the beginning of 2018.  (I summarized this in a February 2018 blog, here.  The CY2017 policy buzzwords for this topic included "EXCITE" and "PACER.")
Now in late 2018, Holland & Knight's Ethan Jorgensen-Earp has a great pick-up in a new essay on CMS device innovation policy.   Here.    He reports that CMS has published notice of a "proposed rule" (meaning a future publication) for streamlined Medicare coverage of breakthrough technologies.  It will be called CMS-3372-P and has a notification webpage, here.  All we know, is that this is a new rule in the CMS rule pipeline.  Fine print in the bottom of the notice suggested it is expected to appear in 03-2019.

See an earlier quote from Don May of Advamed in September, here, that CMS might make rules on its own, that are similar to the goals of H.R. 5997 (below).  That fits to the clue about CMS-3372-P.  See also: MedTechDive, September, here.

click to enlarge

See also:  House Bill 5997, Breakthrough Products Access

See also H.R. 5997, a bill to encourage access to FDA breakthrough products in the Medicare system.  Here.  Delbene press release June 2018 here.  AdvaMed June press release here.  September trade press here, including AdvaMed quotes.

Don't Confuse with Recent LCD Rules Update 

In early October, 2018, CMS released a wide set of updated LCD process rules, to be effective on or before January 1, 2019.  Here.  These were touted as "speeding the LCD process" but I don't see it at all. 

For example, LCD reconsideration requests under the current rules, must get a yes/no reply within 90 days.  

Under the "new rules," the MAC is now only allowed to respond that it has accepted the reconsideration and that...wait for it...that the MAC may or may not do an LCD, which may or may not be favorable, and may or may not occur at any unknown time in the future.  Not an improvement to my eye.   I believe the motivation was to avoid any risk of pre-public leaks of financially important information (see trial about CMS leaks, with criminal convictions,  here), but the net result of not knowing the LCD reconsideration review decision, perhaps ever, is unsatisfactory.

...All the MAC now has to do is reply, after receipt, that they may or may not do an LCD that may or may not be positive and may or may not occur at any unknown time int he future.   A perverse or overworked MAC would hardly even have to read the data to send back such a "Dear John" letter. and it would be a "true" letter, and would conform to the new rules.

I emphasis the LCD chapter because in some ways it short circuits the pending legislation, LCD Clarity Act.   Possibly - I don't know - the CMS regulation on breakthrough products might get a jump on HR 5997 or relate to it in some way.

Thursday, November 1, 2018

Very Brief Blog: Access CMS Final CY2019 Rules for Physician Fee Schedule, Outpatient Fee Schedule, ESRD/DME

On November 1, 2018, CMS released final rules for CY2019 for ESRD & DME, and for the Physician Fee Schedules.   Final Federal Register typeset versions will appear later in November.
  • Fact Sheet, Physician Fee Schedule, here.
    • Rule (inspection copy), here.
  • Fact Sheet, Outpatient & ASC Fee Schedule, here.
    • Rule (inspection copy), here.
  • Fact Sheet, ESRD/DME, here.
    • Rule (inspection copy), here.
For the lab industry, CMS will somewhat liberalize rules for hospital reporting under PAMA including use of hospital outpatient UB1450 claim information (see inspection copy, p. 706ff).  CMS noted it received conflicting comments on this issue.   Genomeweb summarizes the PAMA changes, here

A side note, CMS has proposed in July a huge bubble in the price of several key ISH/IHC reagents, but generally walked these back in November (here).

For physicians, CMS will substantially revamp E&M payments (deferred to 2021) and provides new codes for some digital health services.  For a trade journal report, Healthcare Dive, here.  For insights into the degree this helps digital health, MedCityNews, here.

CMS entertained comments on new approaches to DME gapfill pricing for new products that would be less adverse to innovators but drew no conclusions (see inspection copy, p. 440ff).