See a June 2026 article on sentencing of a Burbank lab fraud case in the tens of millions of dollars. And now in 2026, we're hearing about rising denials on code 87798.
The MolDx system, like Noridian, has a quite strict LCD about pathogen panels, generally restricted them to 3-5 pathogens per patient (with some exceptions). As far as I know, Novitas didn't have that clear an LCD.
I thought I'd heard top-level MolDx speakers say that if you can't bill a CPT panel code of >5 pathogens, you can't or shouldn't bill 87798 (other pathogen) as a large panel x 10...or x 20... or x 40.
However, a data survey by Discoveries in Health Policy found that 87798-intensive labs were found in Noridian as well as Texas in 2024 - just that there were a lot more in Texas.
We searched the Medicare database for Part B 2024 for high billers of 87798 inTX and in CA-AZ (Noridian). Claude Opus 4.8 did a broad search of each of 15 labs under each MAC, stratified by scale (e.g. billing $10M for 87798 versus $1M versus $100,000).
Here's what Claude found in essay format, which I have heavily redacted (Claude had tons of facts and details I've stripped out here.)
- Note that this blog isn't using a specially trained $10M CMS AI contract - this is just a few minutes of work by a consumer-grade $20 AI plan.
###
The Building on XXX Boulevard
What fifteen Noridian billers of CPT 87798 reveal about the geography of the popup lab
A useful thing about studying laboratory fraud at the level of the Medicare claim is that the claim doesn't know it is supposed to look innocent. It records what was billed, by whom, from where, and once you line up fifteen labs that all billed the same code under the same MAC, the addresses start talking. The first batch I looked at — billers of CPT 87798 scattered mostly across Texas — taught a vocabulary of signals: the recent NPI, the absent website, the charge fingerprinted to the penny across "independent" labs. This second batch, fifteen laboratories that registered and billed 87798 under the Noridian jurisdiction of MolDX, speaks with a heavier accent. It is overwhelmingly a Southern California story, with a short Arizona subplot, and its defining feature is not dispersion but concentration. Several of these labs are, quite literally, in the same building.
As before, a caveat that is not boilerplate: what follows is a footprint inventory, not an accusation. Billing 87798 is lawful; some of these fifteen Noridian labs are clearly real laboratories. The exercise sorts the field by what each operation has left behind on the open web and in the federal registries, and asks which ones exist mainly as an electronic billing stream.
Of the fifteen, I put four in the established-lab tier, two in a middle tier with real websites but significant live flags. Then there are nine more: five with a CLIA certificate but little or no web presence, and four that are effectively invisible online. That distribution — a third clearly legitimate, a third deep in the shadows — is itself a headline.
The code, briefly, again
The mechanics that made 87798 attractive in Texas are unchanged here under Noridian, so I'll be brief. It is a "not otherwise specified" amplified-probe infectious-agent code, billed per target, reimbursed at a fixed Clinical Laboratory Fee Schedule rate that lands at almost exactly $34.39 for every one of these fifteen labs. There is no coinsurance to chase and no balance to bill. The submitted charge — which here ranges from $36 to $79.22 — is economically inert under a fixed schedule; it pays nothing extra, which is precisely why it functions as a behavioral tell rather than a price.
The only levers a biller controls are how many beneficiaries and how many targets per beneficiary, and the second of those, services per beneficiary, again does most of the diagnostic work. The established labs in this set cluster low: LEGIT LAB at roughly 6, ANOTHER at 13.5.
The outliers run the other way — XXX LAB in SMALL CITY, billed Noridian about 51 units per patient; another Noridian lab was paid 65 tests per patient. Few patients, enormous per-patient breadth, is the arithmetic of a big-panel invoice.
What legitimacy looks like here
It is worth fixing the honest end of the distribution first, because it calibrates everything else. Four labs carry the boring virtues of a real business. [Labs had NPIs or CLIA certifieates variably months or over a decade old]. None of this proves anything about the quality of their 87798 billing — AAA notably, presents publicly as a toxicology shop, so its unexpected infectious-disease volume is a pivot worth a glance — but these four Noridian labs have good evidence of actually existing.
The building on NNNN STREET, L.A.
Now the concentration. In the Texas set, the suspicious labs were scattered across metro areas and linked only by softer signals — a shared charge, a common phone area code. Here the link is a street address. 111 LAB and 222 Lab share a single building: NNNNN SOMETHING Boulevard in [VALLEY], adjacent suites. Two blocks west, 3333 operates from nnnn; around the corner, 4444 bills from xxxx Avenue [as does a lab curiously named MOLDX LAB, now vanished, not included in these 15.)
.And NNNN SOMETHING is not a quiet medical office — it also houses a stack of separately licensed home-health agencies. That combination, multiple labs and multiple home-health entities sharing one nondescript Valley address, is the kind of co-location that recurs in lab-billing schemes, where a single back office can stand up and operate several nominally unrelated providers.
The people signals rhyme with the geography. A recurring profile in this set is the single-officer corporation in which one individual is simultaneously chief executive, chief financial officer, secretary, and registered agent, reachable at an 818 or 310 — San Fernando Valley and West-LA — telephone number. Examples of odd executive name lists. [....] (JOHN SMITH), is listed in corporate records as the authorized person for thirty-one separate entities. A lab with a website is not automatically clean; a serial incorporator behind it is a question, not an answer.
Two of these — 5555 6666 — sit in my most-invisible tier not because of any one damning fact but because of an absence. Repeated searches surfaced no company website, no resolvable CLIA, and, for ONE GUY, no usable NPI-aggregator profile at all.
The aggregators index essentially every NPI ever issued; returning nothing, while the lab two suites away returns nothing either, is its own kind of evidence. The absence is the finding.
The two that should not file themselves so neatly
Two labs in this set reward a second, closer look for reasons beyond a thin web footprint.
XXXX is recorded not as a Certificate of Compliance or Accreditation but as a Certificate of Registration — the provisional status a laboratory holds before it has been surveyed for compliance.
Billing more than half a million dollars of a high-complexity molecular code while operating on an unsurveyed registration certificate, from a chain whose officer's mailing address resolves to an apartment, and at roughly fifty-plus units per beneficiary, is three independent flags converging on one lab. It is the highest-interest profile in the group.
The second is YYYYY.....
What the screen is for
The limits of this method are the same as before [for Texas] and worth restating, because they protect the innocent labs in the file as much as they flag the others.
A quiet business-to-business reference lab can legitimately have no consumer website. A provisional CLIA can belong to a real startup mid-accreditation. A shared building can be an ordinary commercial sublease. The footprint screen does not adjudicate; it triages, and it converts vague unease into specific, testable questions.
For every lab flagged here the next moves are concrete and available to anyone with the registries: pull the NPPES enumeration date and ownership, pull the QCOR CLIA history for the certificate type, issuance date, laboratory director and any sanction, pull the California or Arizona corporate filing and count the officer's other entities, and map the ordering-physician geography against a Valley address that may have none of its patients anywhere near it. They might come from Florida, for example.
What this set adds to the Texas one is the lesson of geography. The popup lab's tell is not that it bills aggressively — honest labs bill 87798 hard, too. It is that the billing is the only thing it does in public, and that when you plot the quiet ones on a map they stop looking like fifteen independent businesses and start looking like one operation wearing fifteen names. The building on XXXX STREET does not prove that case. It just tells you which afternoon to spend, and where.
Prepared from a web- and registry-footprint review of fifteen CY2024 billers of CPT 87798 enrolled under the Noridian MAC (MolDX); companion spreadsheet contains per-lab detail, links, billing figures, and risk-group tallies (4 established / 2 mid / 5 CLIA-but-thin / 4 near-invisible). This is footprint analysis, not an allegation of wrongdoing; enumeration and CLIA dates should be confirmed at NPPES (npiregistry.cms.hhs.gov) and CMS QCOR (qcor.cms.gov).
# # #
# # #
But:
Others complained of great difficulty in getting 87798 payments, here. Others complained of rising denials, only after 1/2026, here. Article compares 87798, 81408, here.
