On July 14, 2026, CMS released the Physician Fee Schedule (aka Part B rule) as proposals for 2027.
Find the early release version here (a typescript of 1500 pages).
Wait for the Federal Register version here on July 16.
See the FACT SHEET here. (Shared Savings programs get their own Fact Sheet here.)
See the PRESS RELEASE here.
See rapid coverage of the CMS RFI at WaPo here (7 pm). My coverage of WaPo here (10 pm).
##
Included are reforms to practice expense (will you miss IPCI?), an RFI about the big world of CPT and RUC, special rules on remote monitoring (some restrictive), and more about Software as a Medical Service, which was introduced at lengnth (both for general services and specifically for lab services) a couple weeks ago in the OPPS rule (here). They also worry about duplicate testing, an area where PET scans inflict a lot more financial damage than hematocrits.
Comment to September 14; the listed manager for CLFS topics is Patrick Sartini.
##
Lab Software / SaMS
The OPPS rule released about July 2 discussed general services for SaMS at length, before turning to the special case of laboratory tests. It proposed pulling about ten codes—some almost certainly selected in error—off the CLFS and paying them through similarly priced New Technology APCs.
The PFS rule largely reproduces the general argument about software-intensive laboratory analyses: because they are downstream analyses of previously generated data, CMS says they do not have to be performed by a CLIA-certified laboratory or under CLIA requirements. It is my understanding that this position is simply wrong (see here and here), but we will see whether CMS actually validated its legal interpretation before adopting it.
Under OPPS, codes removed from the CLFS would be assigned to New Technology APCs. Under the PFS, however, the same ten codes—again, some apparently selected in error—would be contractor-priced. Eek.
##
RFI on CPT and RUC
The RFI on CPT is has some daggers tucked inside its federalese. But perhaps not surprising in that Secr. Kennedy (and some in Congress) have been complaining about AMA for years. (Yes, you're probably thinking of Sen. Bill Cassidy MD, R-Louisiana, last October). (His press here, my blog here.) AMA may have been popping champagne when said Sen. Cassidy lost his primary on May 16. But the AMA critics are baaack.
Kicking off the RFI, CMS notes that federal regulation—45 CFR 162.1002—requires physician and other non-hospital services to be coded using HCPCS and CPT. But neither HIPAA nor the regulation specifies what the balance between the two systems must be. Why not, to take an extreme example, 100 percent HCPCS and zero percent CPT?
CMS then turns to the AMA Relative Value Scale Update Committee, or RUC. The RUC itself has no apparent place in statute; its role rests largely on administrative practice and tradition. CMS cites “longstanding concern” about reliance on the RUC. MedPAC has raised similar concerns for nearly 20 years, arguing that CMS has “over-relied on specialty societies with a financial stake in the process.”
CMS goes further, suggesting that reliance on CPT and the RUC may be “a potential contributor” to the development of a U.S. “sick-care system,” with too little emphasis on prevention and lifestyle modification. In that sense, the AMA processes may inhibit progress toward the Secretary’s Make America Healthy Again goals. CMS also cites a 2025 National Academies report offering alternatives to dependence on AMA processes for valuing primary care services.
CMS closes with five unusually pointed questions:
Is CMS adequately considering the harms associated with the AMA’s monopoly over CPT? These might include licensing costs, barriers to innovation, and improvements in patient care that have been delayed or abandoned.
Does the CPT code-development process meaningfully consider medical necessity? Could CMS learn from other populations, sites of care, or international health systems?
Could HHS revisit the regulation that formally incorporates CPT into the national coding standard? Could another system replace CPT, or could private competition supplement what is now effectively a monopoly?
What alternatives could replace—or make more objective—the current CPT and RUC processes? How would those alternatives affect innovation?
Could physician procedures be coded and paid using ICD-10-PCS rather than CPT? Could procedures be grouped into broader payment categories, as Medicare already does with MS-DRGs and outpatient APCs, rather than pricing services one CPT code at a time?
The questions are not merely technical. CMS is openly reconsidering whether two privately controlled AMA processes should continue to occupy such central positions in federal coding and physician-payment policy.'
###
There is also an RFI on duplicative tests, 857-873. One focus is poor interoperability and exchange between hospitals, lacking access to yesterday's Seattle MRI, the Los Angeles hospital must repeat the MRI.
# # #
CODA
Chat GPT discusses the harebrained list of inconsistent codes that CMS proposed to take off the CLFS because they are purely software.
### AI CORNEF - Software, Labs, and Sloppiness ###
The deeper problem is not just that the CMS extracted-codes list contains a few bad code selections. It suggests CMS has not developed a workable boundary between a stand-alone algorithmic service and an integrated laboratory test that happens to use software.
Several listed codes plainly describe wet-lab or specimen-based work, not merely reanalysis of an existing digital data set. For example:
0511U includes tumor-cell culture in a three-dimensional microenvironment and testing against 36 or more drugs.
0208U includes mRNA expression analysis from a fine-needle aspirate.
81416 is comparator-exome sequence analysis.
They mean 81417 but that's not what they wrote.)
Several image-analysis codes expressly depend on histology or immunohistochemistry features.
Those are not clean examples of software operating independently on previously generated data.
And once CMS accepts the idea that the algorithmic portion of an integrated test can be conceptually peeled away from the laboratory portion, the policy potentially reaches a huge share of modern oncology testing—not just ten codes. Tempus, Caris, Foundation, Exact Sciences, and many others combine specimen processing, sequencing, pathology, clinical variables, and proprietary computation. CMS would need to explain why one integrated test remains a CDLT while another becomes a separately payable “other diagnostic test.”
The policy could also create perverse operational incentives. A company might be rewarded for generating its own slides, sequencing data, or IHC results as part of an integrated laboratory service, rather than accepting standard external DICOM images or other digital inputs for analysis. Conversely, companies might redesign code descriptors and billing arrangements merely to preserve CLFS treatment or avoid contractor pricing. That is not technological neutrality; it is Medicare shaping laboratory architecture and corporate workflow through an unstable coding distinction.
The fact that the PFS proposal repeats the same peculiar list used in the OPPS proposal—despite the obvious wet-lab content—makes it look less like a carefully validated legal and technical classification and more like a text search for descriptors containing words such as “algorithmic” or “image analysis.” CMS itself says the ten codes were identified from their CPT descriptors, which may explain how the list was assembled, but it does not make the classification defensible.
