HEADER: CMS Proposes Big Shift for Software-intensive Lab Services
See pages 457-462 of the new Hospital Outpatient rulemaking for CY2027:
https://www.federalregister.gov/public-inspection/2026-13656/medicare-program-hospital-outpatient-prospective-payment-and-ambulatory-surgical-center-payment
###
What was Wet Lab and Dry Lab is now... Wet Lab and Not-A-Lab? ?!?
###
News in Brief
[ChatGPT]
CMS has proposed a major CY2027 OPPS policy shift for “software as a medical service” analyses performed on prior laboratory test data. In recent years, lab AI and lab-plus-algorithm services have often been channeled through MAAA, PLA, and CLFS coding/payment pathways. The new OPPS proposal partially reverses that direction.
CMS argues that once the original laboratory work is complete, downstream algorithmic analysis of stored data—such as genomic, transcriptomic, or digital pathology data—may no longer be a clinical diagnostic laboratory test at all.
CMS proposes to move 10 such HCPCS codes out of the CLFS and into OPPS New Technology APCs, with a new status indicator “O1” for separately paid software-as-a-medical-service items. The agency emphasizes that these analyses may be performed by non-CLIA software entities, creating a new Medicare boundary between the laboratory test and later AI/software interpretation.
[I'm not sure that they can be performed by non-CLIA entities, but I'll leave that to CLIA law experts.]
Deeper Dive
CMS’s CY2027 OPPS proposed rule includes a compact but potentially far-reaching section on what it calls “SaMS analyses performed on laboratory tests.” SaMS stands for software as a medical service. The target is not ordinary hospital software, and not simply AI used inside a lab workflow. Rather, CMS is focused on algorithmic analyses performed downstream of a prior laboratory test, especially where the raw or processed data already exist and the later service is essentially computational.
CMS gives the example of genomic sequencing. The sequencing itself may be performed once by a CLIA-certified laboratory. But after that, the same underlying sequence data can be reanalyzed many times by different algorithms to produce diagnostic, prognostic, risk, or treatment-related information. CMS’s key point is that this later algorithmic step may not require a CLIA laboratory at all. A non-CLIA software entity with the relevant program could perform the computation. That distinction drives the policy proposal.
This is why the proposal feels like a flip-flop from the recent coding and payment trajectory for lab AI. Over the past several years, the ecosystem has treated many lab-plus-algorithm products as laboratory services. AMA created MAAA codes for multi-analyte assays with algorithmic analysis. PLA codes have also been used for proprietary lab and algorithmic services. CMS has then often placed these items on the Clinical Laboratory Fee Schedule, using crosswalk or gapfill logic. In practical terms, the system has been willing to treat “lab data plus algorithm” as a lab test.
Now CMS is drawing a sharper line. The agency says it does not believe these secondary algorithmic analyses should be considered clinical diagnostic laboratory tests or priced under the CLFS when they do not require CLIA-regulated laboratory performance. CMS instead characterizes them as “other diagnostic tests” under Medicare law, not “diagnostic laboratory tests.” That is a significant conceptual move.
CMS also flags payment-policy concerns.
- The CLFS generally lacks beneficiary cost-sharing and is not budget-neutral in the same way OPPS is.
- CMS also says it often receives limited transparent cost information for algorithmic components because companies consider the details proprietary.
- Historically, CMS may have relied on visible lab methods in code descriptors—NGS, RT-PCR, methylation, and so on—but CMS now says that comparison is not appropriate when the service at issue is entirely computer-based.
For CY2027, CMS proposes to assign 10 existing HCPCS codes to OPPS New Technology APCs, choosing APC bands that approximate current CY2026 CLFS payment levels. These include oncology and digital pathology AI services, transcriptomic or genomic algorithmic analyses, tumor-response prediction, recurrence scoring, MSI/HRD assessment, and comparator exome analysis. The proposal also creates a new status indicator, “O1,” meaning software as a medical service, paid separately under OPPS.
The larger issue is not just these 10 codes. CMS also proposes that future SaMS analyses performed on lab-test data should go to New Technology APCs under OPPS rather than the CLFS. If finalized, this could reshape the business model for lab-adjacent AI, blur the old PLA/MAAA assumptions, and force companies to think much harder about whether their product is a laboratory test, a software diagnostic, or both.
##
##
SIDEBAR: The Road to Today...
The CMS OPPS announcement was inevitable once AMA CPT last winter began issuing lab-AI codes as Category III codes. CMS handles every Category III code in its OPPS rulemaking, so the AMA CPT maneuver back in February, loaded the gun for this OPPS proposal now in July.
SIDEBAR: Odd Code List Table
CMS lists 9 PLA codes plus CPT code 81416 (comparator exome). I would think they'd leave off wet lab exome 81416, but could have included 81417 (exome re-analysis dry lab) and 81427 (genome re-analysis dry lab.)
SIDEBAR: Typo 493.21, 493.2
CMS refers erroneously to "CFR 493.21," which does not exist. What CMS is saying there - in awful syntax - is that it thinks there are things that are performed on lab tests, which are software rather than CLIA lab services (as define at CLIA 493.2). These "things" will be called, by CMS, SaMS or Software as a Medical Service.
There's no claim by CMS that SaMS is a CLIA term, at 493.2, 493.21 (sic), or anywhere else.
 |
| What was Wet Lab and Dry Lab is now... Wet Lab and Not-A-Lab? |
Changing Codes Table 62. Two parts. Click to enlarge.
