Matrix and HealthFair pay $56.5M over in-home assessment diagnoses the records didn't support

On June 3, 2026, the Justice Department announced that Community Care Health Network, doing business as Matrix Medical Network, mobile-assessment company HealthFair, and HealthFair's founder agreed to pay a total of $56.5 million to resolve False Claims Act allegations that they caused invalid diagnosis codes to be submitted to Medicare Advantage for risk adjustment. The thing to take off your desk: every diagnosis in the case came out of a one-time assessment visit, and the government's problem with it was documentation, not medicine.

What Matrix and HealthFair were paid to do

Matrix contracts with Medicare Advantage organizations to run in-home health assessments and hand back diagnosis codes for the plans to submit to CMS. The MAOs paid Matrix roughly $350 to $450 per assessment. HealthFair did the same work from mobile health buses staffed by nurse practitioners and technicians; Matrix bought it in 2018 and wound it down by 2020.

As a first-tier entity, Matrix had to certify that the data it generated was accurate and truthful. The government's case is that the visits were built to find codes, not to treat patients. Matrix didn't provide treatment, prescribe, or refer for follow-up beyond telling people to see their own doctor.

How the settlement breaks down

The Matrix conduct ran from 2014 to 2019; the HealthFair conduct ran from 2015 to 2017. The two relators will receive $7.3 million and $3.6 million between them. Matrix also made factual admissions about its conduct and signed a five-year Corporate Integrity Agreement with HHS-OIG.

The conditions DOJ flagged at Matrix

The government named seven chronic conditions Matrix reported from its home visits: proliferative diabetic retinopathy, drug-induced polyneuropathy, rheumatoid polyneuropathy, atrial fibrillation, rheumatoid arthritis, chronic obstructive pulmonary disease, and simple chronic bronchitis. These are the diagnoses, the complaint says, that Matrix recorded even though three things were true:

• The assessment form didn't carry enough clinical information to support the diagnosis.
• The diagnosis didn't conform to CMS coding and reporting guidelines.
• No other provider who saw the patient diagnosed the condition in the visit year, the two years before, or the two years after.

That last point is the one to sit with. A complex condition that shows up at a single annual home visit and nowhere else in two years on either side is exactly the gap an auditor pulls on. Matrix also recorded several of these without doing the testing or imaging it takes to establish them.

How HealthFair built diagnoses the records didn't support

HealthFair's pattern was its own kind of cautionary list. The government alleges its providers diagnosed HIV/AIDS, metastatic cancer, and Myasthenia Gravis with no documentation establishing the condition. It says they coded morbid obesity, rheumatoid arthritis, coagulation defect, drug dependence, major depressive disorder, and COPD on patient attestation, claims history, past medical history, or a medication alone.

Two more are worth flagging for any coder who works cardiac charts. HealthFair is alleged to have coded congestive heart failure and heart arrhythmia that the patient's own EKG and echocardiogram results contradicted, and to have coded thrombophilia off nothing more than a separate atrial fibrillation diagnosis. A code the chart's own test results argue against is not a close call.

The "HCC lift" problem

The Matrix complaint spends real space on marketing. The company advertised its ability to capture diagnoses a member's regular doctors hadn't reported, and it sold that as "HCC lift" and an "increase in RAF score" the plan could expect from its assessments. None of that is illegal on its own. It becomes a problem when the diagnoses don't hold up in the record, because then the only thing the visit added was a payment the plan wasn't entitled to.

For coders, that's the line. Finding a real, supported condition another provider missed is good work. Adding a code because the program is built to add codes is the conduct that lands a first-tier entity in a five-year CIA.

Why this lands on your desk

You probably don't code home assessments. You still inherit their output. When a plan submits an assessment-sourced diagnosis, the chart behind it has to survive RADV the same as any other, and a one-visit diagnosis with no corroboration is a weak chart. The seven Matrix conditions are ordinary risk-adjustment diagnoses you touch every week, so the failure here isn't exotic, it's the routine stuff coded without support.

The clean version of this work is the opposite of one-way "add only" review. It's two-way: you add the supported codes and you delete the ones the record won't back, and you treat a diagnosis the test results contradict as a query, not a code. If you want the standard those decisions run on, the MEAT criteria reference lays out what Monitored, Evaluated, Assessed, or Treated actually requires before a diagnosis is reportable, and the problem-lists-alone breakdown covers why a carried-forward entry isn't support. This settlement sits next to the Kaiser $556M resolution as another read on how the government frames unsupported risk-adjustment coding.

What clean coding does differently

Run the questionable encounters back through the encoder to confirm the diagnosis still maps and the documentation is there, and keep the code book open for the guideline language when a home-visit code looks thin. The case is a settlement of allegations, and Matrix's admissions are specific to its conduct, but the documentation bar it describes is the same one every coder already answers to.