Sunday, November 5, 2017

Could the MolDx Program Be Even Better with Increased Transparency?

On a cross country flight this week, I was reading CMS's two 1200 page November policy documents on PFS and Outpatient 2018 policies (here, here).   I was struck by a section in the PFS policy about new approaches to coding and pricing biosimilars, aimed squarely at improving the marketplace for biosimilar development and encouraging market entry once the statutory exclusivity of the original product has expired.   The particular policy will provide more specific coding, and product-specific ASP price setting, and the text about Medicare's desire to support innovation and avoid underpayment is inspiring.  (See CMS text here).  The upshot, which you can read for yourself, is that CMS strongly takes a position both toward encouraging the first innovators (e.g. the first Avastin), but equally, to thereafter encouraging market entry by new companies in the same area.

The MolDx program, which will soon operate in 28 states, handles nearly 90% of CMS molecular test payments as of 2015.*   The program has strongly supported innovators and its coverage policies are all available on the CMS website as LCDs.   MolDx has published extensive documentation about its coverage process.  By 2016, the lion's share of CMS national molecular test payments are MAAA tests covered by MolDX, making it a very important program:

click to enlarge

If so much of Medicare's mopath payments flow through special MolDx rules (see rest of article), it is hard for the CMS Actuary, MedPAC, OIG, or other policymakers to predict what payments will be based on the CMS rules and CMS fee schedules held at Baltimore and on the website.   Even if you study and click around the MolDx website for a long time, you may not know what local payment rules you haven't found or how large their impact might be.

While I am a strong MolDx supporter, some stakeholders such as College of American Pathologists have had concerns about MolDx (see their most recent 2017 infographic here).   The MOLDX statement of work is online at a government RFP website (here). 

MolDx also has at least three programs that raise or lower the price of tests on the CLFS national fee schedule.  These policies are not on the CMS website and are much harder to find and understand than LCDs.   However, as CMS shows in its new innovation-oriented biosimilar rulemaking, payment rules are important important.  I describe the three unique MolDx payment policies here, and offer three practical ways that each policy might be - not so much changed, but made more transparent.

MolDx Payment Policy One:  Raising CPT Code Payments

**  NOTE ***

MolDX has several articles describing use of CMS modifier 22 - for extended services - to flag genetic tests performed with FDA approved kits.   See the "Approved Genetic Tests" webpage at MolDx.  (MolDx here, archive here.)  As of November 5, I count 6 genetic tests flagged with the -22 modifier:  Roche Cobas tests BRAF, EGFR, KRAS (81210, 81235, 81275), Qiagen tests EGFR and KRAS (81235, 81275) and bioMerieux test BRAF (81210).   

MolDX describes its authority for repricing FDA approved genetic tests here.  A discussion of the MolDX negotiated up-pricing of a Qiagen EGFR assay in Genomeweb, May 2014, here.

While no payment amounts are specified, CMS Open Data for 2015 and 2016 suggests that bonus payments ran as high as 2X ($600 for EGFR; 81275; Genoptix 2015; see my deep dive article on an OpenData view of the MolDX Mod-22 program, here.)

click to enlarge
More after the break.

Discussion and Proposal

(A) The MolDx website doesn't discuss price adjustment levels, which could vary from quarter to quarter or even from lab to lab performing the same test.  This makes it hard to plan capital equipment and other investments.  (B) The MolDx innovation payments on these CPT codes are beneficial to MolDx state labs, but could give them a large advantage over competing labs elsewhere (e.g. Texas, Minnesota, New York, Massachusetts, New Jersey, Maryland, etc.)    (C) It isn't clear how these particular FDA approved tests were chosen given the large number of FDA approved combination diagnostics by 2017.   (D) It isn't clear why the articles on Mod22 enhanced payments aren't on the CMS article website, where they would be far more accessible to policy researchers.   (E) Typically, quality bonus programs are handled by CMS centrally (PQRS, MACRA) rather than by local medical directors.  (F) The Center for Innovation CMMI can alter prices (as in a large scale Part B drug pricing proposal in 2016) but must follow statutory processes, among which the payment-altering programs must maintain quality at the same or reduced cost.   

The Modifier 22 based program could be discontinued in lieu of using specific Proprietary Laboratory Analysis Codes (PLA codes).  Within a few quarters, the Mod22 program could be phased out in favor of a PLA based program.  MolDx could set initial PLA code prices at just the same rates as the CPT-mod22 prices, and publish them on its website, just has it has published Tier 2 code prices.  By the following summer, the FDA specific PLA codes would reach public price-setting by CMS in a national public comment process.  Labs would know that other labs using the PLA codes were paid the same price, and the public would have input on the correct uptick for FDA tests.  It wouldn't be secret.  

The proposal is interesting but has the disadvantage that technically, the PLA codes would obligatorily replace the generic CPT codes for the product, raising complexities with private insurers and Medicaid.  An alternative would be to use CMS-specific G codes for the FDA test versions, allowing labs to still bill their commercial insurers with the normal CPT codes while getting the price uptick from CMS that they are now used to with MolDx.  However, the G codes would not get reset to market prices under PAMA, as PLA codes are.   

A third alternative is that if it is good to pay XX% more for FDA genetic tests in 28 states, CMS could codify this as a rule to pay that XX% more for FDA tests in the 50 states.[**]

MolDx Payment Policy Two:  Not Using CPT Codes for "Two or More Genes"

In 2015, MolDX issued a "Test Panel Edit Alert" stating that when 2 or more genetic CPT codes were used, they should be coded as unlisted code 81479 and a combination-specific Z code and price would be created for the lab by MolDx.  MolDx web page here, archive here.

Discussion and Proposal

This is essentially a Multiple Procedure Reduction, which has long applied to surgeries and has recently been applied to imaging and even professional interpretation of imaging (here).  CMS rulebooks state that MACs cannot apply multiple procedure reductions except where specified by CMS (here).   So this is a pretty big departure from policy.  ACLA recently argued to CMS that under PAMA, CMS must pay for CPT codes exactly as specified in PAMA law based on survey prices per CPT code (here) in the context of chemistry panel pricing.  Issues are (A) CMS has written point blank that MACs can't implement multiple procedure pricing except where specified by CMS; (B) labs don't know if their procedure reductions are the same as other labs; and (C) this rule results in very large numbers of 81479 unlisted code claims.   For example, a California lab Ambry in 2015 had 80% unlisted code claims.  (D) This impairs the usefulness of administrative claims records for publish health research.  And (E) it makes it hard or even impossible for OIG, RACs, ZPICs, or other organizations to audit automatically for improper payments due to carrier payment errors.

I believe that ACLA is correct that PAMA law requires payments based on CMS fee schedules, and this particular MolDX policy should probably not be used.  CMS could work on other coding and pricing solutions, possibly a legislative fix regarding multiple genetic services just as there are other legal fixes for multiple services in surgeries and imaging.   

Another answer would be that if MolDx multiple procedure payment reductions for genetic tests followed certain fixed internal rules in MolDx states, these rules should be published in articles, so labs know each other are treated the same.  (For example, if a surgeon bills for two procedures, he knows that public Medicare regulations define his payment, and he knows that his colleague across the street will get the same.)  After a year or so of experimentation within the MolDx test kitchen, the approach should be brought forward in annual CMS rulemaking for nationalization and public comment.  (This would also deal with the question of whether CMS had legal authority to make a particular adjustment). 

The reader may have noticed that Policy 1 and Policy 2 conflict with each other, and without additional information, a RAC or OIG would have a hard time determining correct billing.  If you bill the Roche EGFR and Roche KRAS tests together, do you follow Rule 1 (requiring you to bill both with 22 modifier) or Rule 2 (requiring that the two genes billed with unlisted code 81479?)  MolDx may know the answer, but how would an OIG or RAC know which rule to use?   

MolDx Payment Policy Three:  Irregular Use of CCI Edits (Or Local Creation of CCI-like Edits)

There are tens or hundreds of thousands of CCI edits (here), which block payment of two codes together.   For example, they block payment of "MRI of chest" at the same time as "MRI of chest & abdomen."   Some edits are "Type 0" meaning they are locked prohibitions on co-payment.  Others are "Type 1" edits, meaning they can be overcome by the use of the 59 modifier in exceptional circumstances.  (See OIG report here).   For example, if a podiatrist does a bunionectomy and also has a 20 minute office visit discussing care of diabetic foot problems, both the surgery and office visit are payable, but require a 59 modifier.

There are at least two examples of puzzling use of CCI or similar edits.  
  • CCI has an public, published CCI edit against co-use of BRCA sequencing code 81211 and BRCA dup del code 81213 (see article based on CMS Open Data research, here).   However, at least some MolDx states didn't seem to enforce this in CY2015.    So this would be not applying a CCI edit that is published by CMS staff.  (Imagine Captain Kurtz in Heart of Darkness.)
    • It's hard to know if this OpenData finding is intentional on the part of MolDx or not.  A FOIA request was processed but yielded no correspondence between MolDx and any labs about correct coding of 81211 in the past 3 years.
  • On the other hand, MolDx publishes a table of statutorily excluded codes (here).   One of the excluded codes is 81433, the Dup Del code paired with breast cancer sequencing services 81432.   Thus, 81433 is treated exactly like a code in the CCI book with a Type 0 edit, yet this edit is not in the CCI edit library.   

Analysis and Proposal

Either (for one example) 81433 is statutorily excluded under Medicare laws or it is not; this finding can't vary by state or by MACs.  

If MolDx thinks a code pair is statutorily excluded, it should update the CMS CCI staff.  If MolDx thinks it must implement an edit immediately - well, I can't say.  But CCI edits are updated quarterly and have major updates annually.  MolDx should not implement an idiosyncratic CCI-like exclusionary edit indefinitely for years.  Such edit proposals should be brought to CMS attention, and if it makes sense to a group of policy authorities at CCI, it goes into the CCI edit book; if it fails and is voted down as incorrect by CMS, it shouldn't be applied.

In exactly the same way, if MolDx is processing tens of thousands of claims over the head of a Type 1 CCI edit, the edit might be wrong, and CCI should review it and remove it if the MACs are finding it is always over-ridden.   (In the particular case mentioned, of 81211+81213, I think when used together 81211+81213 can be replaced by 81162, justifying a Type 0 edit, but the decision is CCI's not mine.)

Policy Footnote

While this essay enumerated a range of CMS published policies that are implemented with different local alternatives by MolDx, the full range of flexibility in creating local alternatives is not clear.  CMS does allow contractors some discretion on some topics.  But some CMS rules are (A) manualized policy, others are (B) federal regulations, and others are based in (C) statute (law from Congress).  It isn't always clear from the pages of Medicare manuals whether a particular rule has which source of authority.  Courts have slapped CMS's and MACs' hands for departures from its statute (for example Hays v HHS forbidding the ad hoc practice of least costly alternative pricing.)   Some CMS rules have specific bases in statute (such as the professional services multiple procedure reduction), meaning that neither can CMS vary the law nor can CMS delegate to its contractors the authority to vary the law.  Stated another way, CMS staff may have given MolDx privileges to bend certain rules, but be allowed only within the authority of CMS itself; where the rules depend on statute, CMS doesn't have the authority to delegate authority to act differently.

In addition, the recommendations show a bias toward national policymaking, just as seen in the opening paragraph about biosimilars.  CMS did not authorize some of its contractors to code and price biosimilars differently, or variably use mod 22 (extended service) and mod 52 (reduced services) to change the per item price of biosimilars upward depending on the biosimlar if they were underpriced or dock the price of cancer drugs if they were overpriced.

Footnote to Policy Footnote

ACLA and its attorneys have argued that if PAMA becomes effective January 1, 2018, CMS must pay for all lab codes by statute exactly as written on the CLFS fee schedule (here).  If correct, this could affect all three of the MolDX special pricing policies shown above.


* What is MolDx?   MolDx represents a CMS subcontract to Palmetto GBA, in Columbia SC.  By agreement among several MACs, MolDx is able to hire extra staff to review and set molecular test policies.  It runs a very elaborate website.   The MACs in the MolDx "consortium" follow harmonized policies.   MolDX Percentage of Medicare genomic tests:  The percentage reached almost 90% in 2015, but will be somewhat less now from 2016 forward to the high growth of the Cologuard test, 81528, billed in Wisconsin, which is not a MolDx state.  It would still be true that "except for 1 CPT code," MolDx handles about 85-90% of molecular claims.

** Question left as an exercise for the reader.  If the CMS coverage group can set up a program MolDX, and allow it to pay 2X prices for a Roche cobas test under a CPT code, could the CMS coverage group next set up a program NucMedX, and pay 2X prices for a Siemens (versus GE) PET scan?  You will not be graded on choosing yes or no, but you will be graded on your justification.   Pretend it's a final exam in your law school health policy class.  Limit 600 words, 20 minutes.

A regular MAC has its Statement of Work online, as part of the RFP for its MAC in the federal procurement process.  MolDX has never shown up as an RFP, so there's no public statement of work.  I submitted FOIA requests for the MolDX SOW to both Palmetto and CMS.  Palmetto declined to provide the information under FOIA.   CMS promptly sent me a registration number on my request, which can be tracked online like a FedEx package, but CMS also commented that FOIA requests can be very slow.   Apparently there is a statutory hope that FOIA requests can be fulfilled in 20 days but it's not enforceable.