Multiple screenfuls of infographics.
The DOJ’s 334-Case Fraud Takedown:
A Map of Where Medicare and Medicaid Are Most Hackable
The DOJ page describes “case summaries,” but the scale is better understood as a panoramic audit of how health care fraud mutates across benefit designs. The cases are not random. They cluster around a few recurring pressure points: high-reimbursing products, vulnerable patients, loose documentation pathways, paid referrals, nominee owners, identity misuse, and the conversion of Medicare or Medicaid payments into cash, real estate, luxury cars, watches, jewelry, and shell-company transfers. The 334 summaries are therefore less a list of isolated bad actors than a map of where public health insurance is most mechanically exposed. The DOJ’s own summaries range from amniotic wound allografts and durable medical equipment to hospice, laboratory testing, behavioral health, COVID-19 tests, genetic testing, pharmacy claims, and fake licensure schemes.
The first striking theme is product arbitrage. Fraud follows the margin. In these summaries, the most spectacular dollar figures arise where a product or service can be billed repeatedly, at high unit prices, with thin real-time verification. Wound allografts are the poster child. The allegations often describe elderly Medicare patients, sometimes terminally ill or in hospice, receiving expensive skin substitutes that were unnecessary, not properly applied, applied to inappropriate wounds, or supported by fabricated records. One Arizona-related case alleges a $1.2 billion wound allograft scheme involving sham invoices, kickbacks, vulnerable patients, and approximately $614 million paid; another wound case cited more than $906 million billed and about $297 million paid.
A second theme is that fraud often lives in the gap between clinical need and billing documentation. The claim form says one thing; the patient’s clinical reality says another. Hospice cases allege patients who were not terminally ill, sometimes with beneficiaries induced or misled into enrollment. Laboratory cases allege genetic, respiratory, urinary, toxicology, or other tests ordered without real medical necessity, often through paid marketing or recycled beneficiary information. DME cases allege braces or supplies that were unnecessary, ineligible, or procured through kickbacks. This is the grammar of health care fraud: create a paper trail that looks like medicine, then bill as if medicine happened.
A third theme is industrialization. These are not merely “doctor billed wrong code” stories. Many summaries describe businesses built to manufacture reimbursable events: marketers acquire patients; call centers obtain or buy beneficiary information; clinicians or nominal prescribers sign orders; billing entities transmit claims; shell companies move the money; and consultants explain how to avoid scrutiny. The alleged COVID-19 test schemes are almost pure industrial fraud: beneficiaries did not request tests, yet their Medicare numbers were used to ship or bill for kits. In one Florida case, Medicare beneficiary identification numbers were allegedly purchased and used both for COVID-19 tests and for laboratory testing referrals.
A fourth theme is the exploitation of administrative transitions. Fraudsters watch policy changes the way investors watch markets. The California Medi-Cal pharmacy case is a vivid example: the complaint alleges that a temporary suspension of prior authorization for certain non-contracted generic drugs created an opening for hundreds of millions in claims. According to the complaint, cheap generic ingredients were packaged into high-reimbursing products, prescriptions were generated without real patient evaluation, and Medi-Cal paid more than $178 million on roughly $269 million in false claims.
A fifth theme is that fraud often targets the least visible patients. The summaries repeatedly involve elderly beneficiaries, people with substance use disorders, Medicaid patients, Native American health program enrollees, hospice patients, behavioral health patients, and residents of nursing or assisted-living settings. In one Arizona Medicaid case, the alleged victims were patients in AHCCCS’s American Indian Health Program, with claims for services not provided, substandard, medically unnecessary, or tainted by kickbacks.
The final pattern is personnel fraud: fake nurses, fake credentials, impersonated providers, forged signatures, and stolen identities. These cases are smaller in dollars but disturbing in a different way, because they attack the licensing assumptions beneath the health care system. A person who fakes nursing credentials is not just stealing wages; she is turning the provider credential itself into a billing instrument.
The genius-level TLDR is this: modern health care fraud is rarely a single false claim. It is usually a business model that discovers a payable code, a weak verification point, a vulnerable patient population, and a way to convert medical documentation into cash. Medicare and Medicaid are built on trust, speed, delegation, and documentation. The takedown shows what happens when those same virtues are reverse-engineered by people who understand the payment system almost as well as the people who run it.
SIDEBAR: Six Surprises Hidden in the 334 Case Summaries
1. The biggest fraud theories were not always about exotic medicine.
Some of the largest alleged schemes turned on ordinary administrative vulnerabilities: beneficiary numbers, provider enrollment, prior authorization gaps, routine orders, and electronic claims. The Medi-Cal pharmacy case is especially striking because the alleged opening was a temporary suspension of prior authorization during a payment-system transition, which allegedly enabled $269 million in false claims in just 11 months.
2. Wound allografts became the new “molecular testing” of fraud headlines.
A fraud-watcher might expect DME, home health, opioids, or genetic testing. But the summaries repeatedly spotlight amniotic wound allografts, including a $1.2 billion alleged scheme involving sham invoices, kickbacks, pass-through bank accounts, and elderly patients, some terminally ill or in hospice.
3. AI appears, but not where you might expect.
The AI hook was not a futuristic diagnostic algorithm; it was allegedly fake consent recordings. In one COVID-19 test-kit case, the defendant allegedly supplied laboratories with recordings in which Medicare beneficiaries supposedly agreed to receive tests, and the source of the recordings reportedly said they were AI-generated.
4. Hospice fraud had a macabre identity-theft variant.
The hospice cases were not limited to patients who were not terminally ill. One Los Angeles case allegedly involved buying the identifying information of already-deceased beneficiaries from a mortuary employee, then purporting to enroll them in hospice before death to make the hospice statistics look better.
5. The “patient” was sometimes less a patient than an entry in a production process.
Several schemes appear to have treated beneficiaries as raw material: names to be bought, induced, enrolled, called, shipped to, or billed against. In the Arizona AHCCCS case, the alleged target population was Native American Medicaid members in a fee-for-service program, with claims for services that were allegedly not provided, substandard, medically unnecessary, or tainted by kickbacks.
6. Credential fraud was its own quiet category.
Not all the surprising cases were billion-dollar billing machines. Some involved people allegedly posing as licensed health professionals: for example, an Idaho defendant allegedly used victims’ names and nursing licenses to obtain jobs at facilities including behavioral health, rehabilitation, skilled nursing, and hospice businesses.
# # # #
Appendix
Chat GPT adds, regarding the court cases:
Information. An information is a formal criminal charging document filed by prosecutors, rather than returned by a grand jury; it is commonly used when the defendant waives indictment or is expected to plead/resolve the case. In the Galbraith example, the document simply says “The United States of America charges” the defendant with health care fraud, then lays out the alleged Medicare hospice scheme.
Indictment. An indictment is a formal criminal charging document returned by a grand jury after prosecutors present evidence and the grand jury finds probable cause to charge the defendant. In the Lopez example, the caption identifies the “February 2026 Grand Jury,” and the charging language begins, “The Grand Jury charges.”
Complaint. A criminal complaint is usually an earlier-stage charging document supported by a sworn affidavit showing probable cause, often used to obtain an arrest warrant before indictment or information. In the Mareik example, the complaint says it is based on an attached affidavit, and the affidavit states it is offered to support a criminal complaint and arrest warrant and “does not purport to set forth all” the investigator’s knowledge.
Count: ~18 matters (19 filings if you split the two Puerto Rico actions).
Grouped by type:
Genetic / genomic testing
- CM (MDFL) — COVID test kits and genetic testing via beneficiary-info sales to labs.
- SM (MDFL/SDFL) — ASAP Labs, genetic and respiratory diagnostic testing kickbacks; $5M FCA + guilty plea.
- TB (SDFL) — $62M genetic-testing telemarketing/"doctor-chasing" scheme.
- DN (EDNC) — Andor Labs / Minerva Genetics; cancer genomic, toxicology, COVID; >$15M paid.
- AK(OR) — Oregon Clinical Laboratory, genetic testing to MA plans; >$15M.
- BE (NDTX) — Neva/Areahou Diagnostics; lab testing incl. genetic; $65M billed, $43M paid.
Clinical lab / toxicology / drug testing
7. LC (CDCA) — UTI tests, respiratory pathogen panels, oral tox screens; $9M billed (some specimens from deceased beneficiaries).
8. DD (CT) — reference-lab Medicaid enrollment fraud; related lab settled for medically unnecessary drug tests.
9. CB (DE) — medically unnecessary urine drug tests + tests never performed; $2.8M.
10. HK (EDMI) — false oral-fluid drug-test results; $150K civil.
11. WE (MDNC) — thousands of fictitious drug tests; $2.8M.
COVID-19 / respiratory lab billing
12. BL (MDFL) — ~152,000 COVID claims; $14.4M reimbursed.
13. BB (NDIL) — sold beneficiary info, AI-faked consent recordings; caused labs to bill ~$342M ($240M paid).
14. BI (EDLA) — RPP testing bundled with COVID; $51.7M billed, $28.4M paid.
15. SI (EDMI) — $566M COVID lab-test billing fraud.
16. BK(NDOK) — pharmacies billing Medicare/Medicaid for unrequested COVID tests.
17. MK et al. (NDTX) — COVID test-kit kickback scheme; >$73M billed.
18. CM (PR) — two filings: a civil settlement (services lacking medical necessity/ordering) and an indictment (repetitive/consecutive COVID-test billing, $114K).
Good distinction to draw, and it's actually the most analytically revealing cut you can make on this document. Three buckets, not two, because a meaningful slice fits neither:
| Bucket | What it means | Rough count |
|---|---|---|
| Pure electronic (phantom billing, no clinical event) | The service never happened to anyone. Fraud lives entirely in the claim/timesheet/record. | ~150–160 (roughly half) |
| Improper clinical event | A real test/device/drug/procedure occurred, but it was unnecessary, upcoded, diverted, or delivered by an unqualified/excluded person. | ~80–90 (~25–30%) |
| Neither | Pure theft/embezzlement, eligibility/identity fraud, PPP/SNAP/check fraud, abuse-or-neglect-only. | ~30–40 |
These are estimates, not a census — many are genuinely mixed, and reasonable people would shuffle 20–30 entries between buckets. But the central finding is robust: pure electronic fraud is the largest category, and it dominates because of the state Medicaid Fraud Control Unit caseload.
What drives the "pure electronic" half. The state cases — Louisiana, Missouri, Illinois, Minnesota, Pennsylvania, Mississippi, Washington, Ohio, the Carolinas — are overwhelmingly personal-care-attendant / timesheet fraud. The signature fact pattern is "billed for services that could not have been provided": the aide was at a second job, in another state, at Walmart or Jack-in-the-Box, or the recipient was hospitalized, incarcerated, out of the country, or dead. No clinical event of any kind — just a phantom hour entered into an EVV system. Your Shachar dead-beneficiary hospice example is the federal-scale version of exactly this: the BIN got billed, the corpse got no care. (Shachar is technically mixed — he also enrolled living non-terminal patients who did get unwanted hospice — but the dead-enrollment portion is the purest of pure.)
The clean federal "pure electronic" entries are the ghost-supplier schemes: Hilmi's $3.76B and the Operation Gold Rush DME shells (Parastatov/Smirnov $10.6B, Bendeliani, Danelia, Kimeridze, Zarbavel), the 12-clinic Florida "bust-outs," Muniz Rodriguez's nominee clinics, Grinberg's fictitious eyelid surgeries run out of optical shops, Ekpenyong billing for Pediasure he never bought, Whipple billing for care while in Hawaii and after patients died.
What's genuinely improper-clinical (the real-event quarter). Three sub-patterns:
- Real product, applied: the amniotic allograft cases (Rowan, Yukee, McMillan, Kamien, much of Tesar) — these grafts were physically slapped onto infected, healed, or terminally-ill patients' wounds. Most DME-delivered and skin-substitute matters live here.
- Real drugs, moved: the pill mills and diversions (Aquino's 1.08M doses, Moss's 5.6M, the Houston clinics, Dandy/Kirk, plus the Indiana nurse-diversion cluster). Pills genuinely changed hands.
- Real test, run unnecessarily / upcoded: Finkelstein's cardiovascular testing, the EEG case, Broussard's RPP panels, the Washington dental upcoding (ibuprofen billed as parenteral medicaments).
One conceptual wrinkle worth flagging for your taxonomy. There's a recurring sub-type that resists clean placement: a clinical event happened, but the wrong person did it — services rendered by an excluded provider (Meckler, Nala Care), by an unlicensed person under a licensee's number (Appiah/Steinway, Avant/Anyokwu), or by LPNs billed as RNs (Vassar-Cunningham, Isluv Robertson). Physically, care occurred; legally, the billable event never existed. I parked these in "improper clinical" because a human clinical interaction took place, but you could equally argue they belong with the phantom-billing group since the billed service (qualified-provider care) is fictional. Roughly 10–12 cases hinge on this judgment call, which is most of the wobble in the totals.
The headline, if you're writing this up: the modern takedown is mostly a billing-data crime, not a bad-medicine crime. The dollar value skews toward improper-clinical (allografts and DME, where a real expensive thing got delivered), but the case count skews heavily toward pure electronic fiction — phantom hours and ghost suppliers — because that's what EVV records, NPI-misuse, and dead-beneficiary matches catch cheaply at scale.
.
