Karma's Most Wanted #7: UnitedHealth Group

By Matt Stone

The American healthcare system is a monolithic beast.

There is a strange disconnect in America where the most powerful decisions affecting human lives are made in rooms where no human suffering is visible at all.

No hospital smells. No IV pumps beeping in the background. No exhausted family members sleeping in plastic chairs next to a bed that costs more per night than a decent hotel in Manhattan.

Just fluorescent lights, spreadsheets, and the quiet confidence of a machine that has never once had to learn how to walk again after a stroke.

Somewhere deep inside the digital bowels of the modern healthcare system, a predictive model is making very confident guesses about how long it should take a human body to heal. It does this without fear, doubt, or the slightest awareness that it is gambling with the most fragile commodity on earth: time.

Recovery time. Therapy time. The delicate weeks when a patient is clawing their way back from surgery, stroke, or catastrophic illness.

And the machine is impatient.

If the model says the recovery window is over, the coverage stops. Not with drama or sirens or a committee of doctors arguing over charts. It stops quietly. A code flips in the system. A coverage authorization expires. A patient is suddenly on their own.

That is how power works in modern America. Not with shouting, but with paperwork.

You rarely see the system itself. What you see instead are the people it leaves behind.

And if you want to understand what that looks like in practice, you have to start with the patients.

Because the machine may be invisible.

But the consequences are not.

If you want to see how this works in the real world, you do not start with spreadsheets. You start with people. Three of them will do for now.

Story One

The Stroke Patient

One of the cases described in legal filings begins the way thousands of medical crises begin every day in America.

An elderly patient suffered a stroke.

Stroke recovery is not a quick process. When blood flow to the brain is interrupted, entire networks responsible for movement, speech, and cognition can be damaged. Rehabilitation is the slow and delicate work of rebuilding those pathways. Patients spend hours each day relearning how to stand, how to move their arms, how to speak clearly enough to be understood.

It is exhausting work. It is also time-sensitive. The first weeks after a stroke are often the most critical window for recovery.

The patient’s physicians recommended continued rehabilitation inside a skilled nursing facility. Therapists believed additional weeks of treatment were necessary to improve mobility and reduce the risk of long-term disability.

But inside the insurance system, another calculation was taking place.

A predictive algorithm estimated how long recovery should take based on statistical averages drawn from thousands of other patients. The model generated a projected discharge date.

When that date arrived, coverage ended.

The rehabilitation facility informed the family that the insurer would no longer pay for the patient’s stay. The doctors still believed the patient needed therapy, but medical recommendations do not control insurance coverage.

Coverage controls care.

The patient was discharged despite the physician’s recommendation that rehabilitation continue.

The family appealed the decision. Eventually the denial was reversed. But by the time the appeal succeeded, the patient had already left the facility and the interruption in therapy had already occurred.

Stroke rehabilitation is measured in weeks and months. Interruptions in therapy can permanently affect the outcome. Muscles weaken quickly when therapy stops. Speech improvements can stall. Recovery windows can close.

For that patient, the clock was not controlled by the neurologist overseeing the case.

It was controlled by an insurance algorithm.

Story Two

The Post-Surgery Discharge

Another case involved a patient recovering from major surgery who had been admitted to a rehabilitation facility to regain strength and mobility.

After significant operations, patients often require structured rehabilitation before returning home safely. Physical therapy helps rebuild muscle strength. Occupational therapy focuses on everyday tasks such as walking, bathing, and climbing stairs. Without those supports, complications become far more likely.

In this case, the patient’s doctors believed additional days of supervised therapy were necessary to stabilize the recovery process.

But the insurance model predicted a shorter rehabilitation timeline.

The algorithm estimated how long recovery should take based on population-level data. Once the predicted number of days had passed, the coverage authorization expired.

The patient and family were presented with two options.

Leave the facility.

Or pay out of pocket.

Rehabilitation care can cost thousands of dollars per week. For many families, the price of continuing treatment without insurance coverage is financially devastating.

So the calculation becomes brutally simple.

Drain savings to keep a loved one in care.

Or bring them home before they are medically ready.

Families across the country make that decision every day.

And they often make it under intense time pressure, after a single phone call informing them that coverage has ended.

Story Three

The Nursing Home Transfer

Investigative reporting has revealed another layer of the system, one that operates not through discharge decisions but through hospital transfers.

Some nursing homes participating in certain insurance programs were reportedly evaluated based on how frequently they transferred residents to hospitals. Internal metrics tracked hospitalization rates, and facilities that kept those numbers low could receive financial incentives.

The stated goal of these programs was to reduce unnecessary hospitalizations and improve care coordination. Many hospital transfers can indeed be avoided with proper on-site treatment.

But critics say the incentive structure created dangerous pressure.

Former employees and whistleblowers have alleged that facilities sometimes delayed sending residents to hospitals even when serious symptoms appeared.

In one documented case described in investigative reporting, a nursing home resident began experiencing symptoms consistent with a stroke.

The resident was not immediately transferred to a hospital.

The delay resulted in catastrophic brain injury.

UnitedHealth disputes wrongdoing and maintains that its programs are designed to improve patient care and prevent unnecessary hospital visits.

But critics argue that the system can create a troubling dynamic.

When financial incentives reward fewer hospital transfers, hesitation becomes part of the system.

And in medical emergencies, hesitation can change the outcome of a life.

To whistleblowers and investigators, the structure looked less like care coordination and more like cost management operating under a medical label.

Cost management disguised as care coordination.

The Numbers Behind the Machine

The lawsuits and investigations are disturbing enough on their own. But the statistics reveal something even more unsettling.

Only a tiny fraction of denied claims are ever appealed.

Some estimates suggest fewer than one percent of patients challenge insurance denials.

When appeals do occur, a significant percentage of those decisions are reversed.

Which means something extraordinary is happening.

Millions of medical decisions are effectively final not because they were correct, but because patients never had the time, knowledge, or resources to challenge them.

The algorithm does not have to be perfect.

It only has to be final.

The Real Scandal

The real scandal is not that an algorithm made mistakes.

Every system makes mistakes.

The scandal is that the American healthcare system quietly allowed those systems to become gatekeepers of care for millions of people.

Medicine used to revolve around a simple hierarchy.

Doctors made decisions. Hospitals delivered treatment. Insurance companies paid the bill.

That hierarchy has been quietly inverted.

Today insurance companies often decide what treatment will be allowed, and increasingly those decisions are guided by predictive models analyzing population-level data.

In other words, a statistical model can determine how long a human being is allowed to heal.

Why This F*cking Matters

UnitedHealth Group generated hundreds of billions of dollars in revenue last year. It is not just an insurance company anymore. It is a data infrastructure for modern healthcare.

When companies operating at that scale introduce predictive decision systems into medicine, those systems effectively become part of the healthcare system itself.

Which means the most important medical decisions in America are no longer happening only in hospitals.

They are happening inside algorithms.

And almost nobody outside the system knows it.

Because these cases did not happen in isolation. They happened inside a system that processes tens of millions of patients every year, a system increasingly guided by predictive models that calculate how long recovery should take and how much treatment a body is statistically worth.

The machine does not see the patient struggling to lift their arm again after a stroke. It does not hear the speech therapist coaxing a sentence out of a damaged brain. It only sees timelines, averages, and cost curves. And when those numbers say the recovery window is closed, the decision moves forward with quiet administrative certainty.

Somewhere in America tonight, another family will receive the same phone call.

Coverage has ended.

Therapy stops tomorrow.

Take the patient home.

It is worth noting, that less than 1% of claim decisions are even even appealed. And those that are, many see reversals in those decisions. People are too tired and too defeated to even fight back. We have to change that.

Sources

Christopher Weaver and Anna Wilde Mathews. “UnitedHealth Uses an Algorithm to Cut Off Rehab Care for Medicare Patients.” STAT News, November 14, 2023.
https://www.statnews.com/2023/11/14/unitedhealth-algorithm-medicare-advantage/

Christopher Weaver. “Inside UnitedHealth’s Effort to Control Rehab Lengths Using Predictive Software.” STAT News.
https://www.statnews.com

Chris McGreal. “UnitedHealth secretly paid nursing homes to reduce hospital transfers, investigation finds.” The Guardian, May 2025.
https://www.theguardian.com/us-news/2025/may/21/unitedhealth-nursing-homes-payments-hospital-transfers

Chris McGreal. “How UnitedHealth’s nursing home strategy left patients without hospital care.” The Guardian investigative series.
https://www.theguardian.com/us-news/ng-interactive/2025/dec/17/unitedhealth-nursing-homes

U.S. Department of Health and Human Services, Office of Inspector General. “Some Medicare Advantage Organization Denials of Prior Authorization Requests Raise Concerns About Beneficiary Access to Medically Necessary Care.” 2022.
https://oig.hhs.gov/reports/all/2022/medicare-advantage-prior-authorization-denials/

Kaiser Family Foundation. “Medicare Advantage in 2024: Enrollment Update and Key Trends.”
https://www.kff.org/medicare

Centers for Medicare & Medicaid Services. “National Health Expenditure Data.”
https://www.cms.gov/data-research/statistics-trends-and-reports/national-health-expenditure-data

Estate of Gene B. Lokken et al. v. UnitedHealth Group Inc. et al., U.S. District Court (class action lawsuit regarding the nH Predict algorithm used in post-acute care coverage decisions).

UnitedHealth Group annual reports and SEC filings documenting corporate revenue and scale.
https://www.unitedhealthgroup.com/investors