Finalist: Casey Ross and Bob Herman of STAT

By Casey Ross and Bob Herman 

An algorithm, not a doctor, predicted a rapid recovery for Frances Walter, an 85-year-old Wisconsin woman with a shattered left shoulder and an allergy to pain medicine. In 16.6 days, it estimated, she would be ready to leave her nursing home.

On the 17th day, her Medicare Advantage insurer, Security Health Plan, followed the algorithm and cut off payment for her care, concluding she was ready to return to the apartment where she lived alone. Meanwhile, medical notes in June 2019 showed Walter’s pain was maxing out the scales and that she could not dress herself, go to the bathroom, or even push a walker without help.

It would take more than a year for a federal judge to conclude the insurer’s decision was “at best, speculative” and that Walter was owed thousands of dollars for more than three weeks of treatment. While she fought the denial, she had to spend down her life savings and enroll in Medicaid just to progress to the point of putting on her shoes, her arm still in a sling.

Health insurance companies have rejected medical claims for as long as they’ve been around. But a STAT investigation found artificial intelligence is now driving their denials to new heights in Medicare Advantage, the taxpayer-funded alternative to traditional Medicare that covers more than 31 million people.

Behind the scenes, insurers are using unregulated predictive algorithms, under the guise of scientific rigor, to pinpoint the precise moment when they can plausibly cut off payment for an older patient’s treatment. The denials that follow are setting off heated disputes between doctors and insurers, often delaying treatment of seriously ill patients who are neither aware of the algorithms, nor able to question their calculations.

Older people who spent their lives paying into Medicare, and are now facing amputation, fast-spreading cancers, and other devastating diagnoses, are left to either pay for their care themselves or get by without it. If they disagree, they can file an appeal, and spend months trying to recover their costs, even if they don’t recover from their illnesses.

“We take patients who are going to die of their diseases within a three-month period of time, and we force them into a denial and appeals process that lasts up to 2.5 years,” Chris Comfort, chief operating officer of Calvary Hospital, a palliative and hospice facility in the Bronx, N.Y., said of Medicare Advantage. “So what happens is the appeal outlasts the beneficiary.”

The algorithms sit at the beginning of the process, promising to deliver personalized care and better outcomes. But patient advocates said in many cases they do the exact opposite — spitting out recommendations that fail to adjust for a patient’s individual circumstances and conflict with basic rules on what Medicare plans must cover.

“While the firms say [the algorithm] is suggestive, it ends up being a hard-and-fast rule that the plan or the care management firms really try to follow,” said David Lipschutz, associate director of the Center for Medicare Advocacy, a nonprofit group that has reviewed such denials for more than two years in its work with Medicare patients. “There’s no deviation from it, no accounting for changes in condition, no accounting for situations in which a person could use more care.”

Medicare Advantage has become highly profitable for insurers as more patients over 65 and people with disabilities flock to plans that offer lower premiums and prescription drug coverage, but give insurers more latitude to deny and restrict services.

Over the last decade, a new industry has formed around these plans to predict how many hours of therapy patients will need, which types of doctors they might see, and exactly when they will be able to leave a hospital or nursing home. The predictions have become so integral to Medicare Advantage that insurers themselves have started acquiring the makers of the most widely used tools. Elevance, Cigna, and CVS Health, which owns insurance giant Aetna, have all purchased these capabilities in recent years. One of the biggest and most controversial companies behind these models, NaviHealth, is now owned by UnitedHealth Group.

It was NaviHealth’s algorithm that suggested Walter could be discharged after a short stay. Its predictions about her recovery were referenced repeatedly in NaviHealth’s assessments of whether she met coverage requirements. Two days before her payment denial was issued, a medical director from NaviHealth again cited the algorithm’s estimated length of stay prediction — 16.6 days — in asserting that Walter no longer met Medicare’s coverage criteria because she had sufficiently recovered, according to records obtained by STAT.

Her insurer, Security Health Plan, which had contracted with NaviHealth to manage nursing home care, declined to respond to STAT’s questions about its handling of Walter’s case, saying that doing so would violate the health privacy law known as HIPAA.

Walter died shortly before Christmas last year.

NaviHealth did not respond directly to STAT’s questions about the use of its algorithm. But a spokesperson for the company said in a statement that its coverage decisions are based on Medicare criteria and the patient’s insurance plan. “The NaviHealth predict tool is not used to make coverage determinations,” the statement said. “The tool is used as a guide to help us inform providers, families and other caregivers about what sort of assistance and care the patient may need both in the facility and after returning home.”

As the influence of these predictive tools has spread, a recent examination by federal inspectors of denials made in 2019 found that private insurers repeatedly strayed beyond Medicare’s detailed set of rules. Instead, they were using internally developed criteria to delay or deny care.

But the precise role the algorithms play in these decisions has remained opaque.

STAT’s investigation revealed these tools are becoming increasingly influential in decisions about patient care and coverage. The investigation is based on a review of hundreds of pages of federal records, court filings, and confidential corporate documents, as well as interviews with physicians, insurance executives, policy experts, lawyers, patient advocates, and family members of Medicare Advantage beneficiaries.

It found that, for all of AI’s power to crunch data, insurers with huge financial interests are leveraging it to help make life-altering decisions with little independent oversight. AI models used by physicians to detect diseases such as cancer, or suggest the most effective treatment, are evaluated by the Food and Drug Administration. But tools used by insurers in deciding whether those treatments should be paid for are not subjected to the same scrutiny, even though they also influence the care of the nation’s sickest patients.

In interviews, doctors, medical directors, and hospital administrators described increasingly frequent Medicare Advantage payment denials for care routinely covered in traditional Medicare. UnitedHealthcare and other insurers said they offer to discuss a patient’s care with providers before a denial is made. But many providers said their attempts to get explanations are met with blank stares and refusals to share more information. The black box of the AI has become a blanket excuse for denials.

“They say, ‘That’s proprietary,’” said Amanda Ford, who facilitates access to rehabilitation services for patients following inpatient stays at Lowell General Hospital in Massachusetts. “It’s always that canned response: ‘The patient can be managed in a lower level of care.’”

Brian Moore, a physician and advocate for patients denied access to care at North Carolina-based Atrium Health, recalled visiting a stroke patient who was blocked from moving to a rehabilitation hospital for 10 days. “He was sitting there trying to feed himself. He was like, ‘I just never thought when I signed up for Medicare Advantage that I wouldn’t be able to get the care I need,’” he said. “He was drooling and crying.”

The cost of caring for older patients recovering from serious illnesses and injuries, known as post-acute care, has long created friction between insurers and providers. For decades, facilities like nursing homes racked up hefty profit margins by keeping patients as long as possible — sometimes billing Medicare for care that wasn’t necessary or even delivered. Many experts argue those patients are often better served at home.

The enactment of the Affordable Care Act in 2010 created an opportunity for reform. Instead of paying for care after the fact, policy experts proposed flipping the payment paradigm on its head: Providers would be paid a lump sum upfront, incentivizing them to use fewer resources to deliver better outcomes.

At the time, most Republicans in Congress were wringing their hands over the new law and its subsidies to help low- and middle-income Americans pay for health insurance. Tom Scully, the former head of the Centers for Medicare and Medicaid Services under George W. Bush, shared those concerns. But he also saw something else: a potential billion-dollar business.

Scully drew up plans for NaviHealth just as the new law was taking effect. Its payment reforms aligned perfectly with the Medicare Advantage program he had played a pivotal role in creating during the Bush administration.

Scully knew how those insurance plans worked. He also knew they were taking a financial beating in post-acute care.

“Look, I love the nursing home guys, but there were a lot of patients coming out of hospitals spending 20 days in a nursing home in MA,” because that’s what Medicare’s rules allowed, Scully said on a podcast in 2020. “It was just like Pavlov’s bell.”

As a well-connected partner at the private equity firm Welsh, Carson, Anderson & Stowe, Scully heard of a small shop called SeniorMetrix that was working on this type of post-acute data and analytics. The firm quickly won him over. “They had an algorithm,” Scully said on the podcast. “I saw it and said, ‘This is it.’”

He wrote a $6 million check to buy the company, which he rebranded to NaviHealth. Scully then raised $25 million from wealthy friends and companies, including the health system Ascension and the rehabilitation hospital chain Select Medical, and coaxed another $25 million from Welsh Carson.

NaviHealth started making its sales pitch to Medicare Advantage plans: Let us manage every piece of your members’ care for the first 60 to 90 days after they are discharged from the hospital, and we’ll all share in any savings.

The sweetener was the technology. One of the company’s core products is an algorithm called nH Predict. It uses details such as a person’s diagnosis, age, living situation, and physical function to find similar individuals in a database of 6 million patients it compiled over years of working with providers. It then generates an assessment of the patient’s mobility and cognitive capacity, along with a down-to-the-minute prediction of their medical needs, estimated length of stay, and target discharge date.

In a six-page report, the algorithm boils down patients, and their unknowable journey through health care, into a tidy series of numbers and graphs.

The product was a revelation to insurers, giving them a way to mathematically track patients’ progress and hold providers accountable for meeting therapy goals.  By summer 2015, NaviHealth was managing post-acute care for more than 2 million people whose insurance plans had contracted with the company.   It was also working with 75 hospitals and clinics seeking to more carefully manage contracts in which they shared financial responsibility for holding down costs. At the time, spending on post-acute care accounted for $200 billion annually.

That same year, Scully sold NaviHealth to the conglomerate Cardinal Health for $410 million — roughly eight times the investment. In 2018, another private equity firm, Clayton, Dubilier & Rice, upped the ante and paid $1.3 billion to take over NaviHealth. Then in 2020, UnitedHealth — the largest Medicare Advantage insurer in the country — decided to make the hot commodity its own, buying NaviHealth in a deal valued at $2.5 billion.

In an interview with STAT, Scully said the concept behind NaviHealth is “totally correct,” because it roots out wasteful spending. And he did not believe the algorithms restricted necessary care. But when presented with reporting that showed NaviHealth was at the center of voluminous denials and overturned appeals, Scully said he wasn’t in a position to comment on what may have changed since he sold his stake.

“The NaviHealth decision tool as I knew it — again, this is eight years ago — has a place and is valuable. If [it] overdoes it and is inappropriately denying care and sending people to the wrong site of service, then they’re foolish, and they’re only hurting themselves reputationally,” Scully said. “I have no idea what United’s doing.”

Providers told STAT that as NaviHealth was changing hands and enriching its investors, they started noticing an increase in denials under its contracts — that the pendulum had now swung too far in the other direction in an effort to prevent overbilling and make sure patients weren’t getting unnecessary services.

Patients with stroke complications whose symptoms were so severe they needed care from multiple specialists were getting blocked from stays in rehabilitation hospitals. Amputees were denied access to care meant to help them recover from surgeries and learn to live without their limbs. And efforts to reverse what seemed to be bad decisions were going nowhere. Atrium Health’s Moore, who leads a team that specializes in reviewing medical necessity criteria, started taking a deeper look at the denials.

“It was eye-opening,” he said. “The variation in medical determinations, the misapplication of Medicare coverage criteria — it just didn’t feel like there [were] very good quality controls.”

He and many other providers began pushing back. Between 2020 and 2022, the number of appeals filed to contest Medicare Advantage denials shot up 58%, with nearly 150,000 requests to review a denial filed in 2022, according to a federal database.

The database fails to capture countless patients who are unable to push back when insurers deny access to services, and only reflects a portion of the appeals even filed. It mostly tracks disputes over prior authorization, a process in which providers must seek insurers’ advance approval of the services they recommend for patients.

In comments to federal regulators and interviews with STAT, many providers described rigid criteria applied by NaviHealth, which exercises prior authorization on behalf of the nation’s largest Medicare Advantage insurers, including its sister company UnitedHealthcare as well as Humana and several Blue Cross Blue Shield plans.

“NaviHealth will not approve [skilled nursing] if you ambulate at least 50 feet. Nevermind that you may live alon(e) or have poor balance,” wrote Christina Zitting, a case management director for a community hospital in San Angelo, Texas. She added: “MA plans are a disgrace to the Medicare program, and I encourage anyone signing up…to avoid these plans because they do NOT have the patients best interest in mind. They are here to make a profit. Period.”

Federal records show most denials for skilled nursing care are eventually overturned, either by the plan itself or an independent body that adjudicates Medicare appeals.

But even patients who win authorization for nursing home care must reckon with algorithms that insurers and care managers like NaviHealth use to help decide how long they are entitled to stay. Under traditional Medicare, patients who have a three-day hospital stay are typically entitled to up to 100 days in a nursing home.

With the use of the algorithms, however, Medicare Advantage insurers are cutting off payment in a fraction of that time.

“It happens in almost all these cases,” said Christine Huberty, a lawyer in Wisconsin who provides free legal assistance to Medicare beneficiaries. She said Medicare Advantage patients she represents rarely stay in a nursing home more than 14 days before they start receiving payment denials.

“But [the algorithm’s report] is never communicated with clients,” said Huberty, who often only finds the report after filing a legal complaint. “That’s all run secretly.” NaviHealth said the findings of the algorithm, if not the report itself, are routinely shared with doctors and patients to help guide care.

A director at one post-acute facility said denials from UnitedHealthcare and NaviHealth are now the norm for many of their patients, even if they are clearly sicker than what the algorithm projects.

“They are looking at our patients in terms of their statistics. They’re not looking at the patients that we see,” said the director, who asked not to be named to avoid jeopardizing relationships with Medicare Advantage plans.

And when insurers deluge providers with denials, “they’re hoping that their endurance is greater than ours,” the director said.

NaviHealth has not published any scientific studies assessing the real-world performance of its nH Predict algorithm. And to the extent it tests its performance internally, those results are not shared publicly.

Additionally, regulators do not monitor these algorithms for fairness or accuracy, but the industry-wide blowback has forced the government to consider acting. Federal Medicare officials proposed new rules in December that say Medicare Advantage insurers can’t deny coverage “based on internal, proprietary, or external clinical criteria not found in traditional Medicare coverage policies.” Insurers also would have to create a “utilization management committee” that reviews their practices every year.

But even these proposals would still allow insurance companies to “create internal coverage criteria,” as long as they are “based on current evidence in widely used treatment guidelines or clinical literature that is made publicly available.”

Major lobbying groups for health insurance companies — America’s Health Insurance Plans, the Better Medicare Alliance, and the Alliance of Community Health Plans — did not make anyone available for interviews. Instead, the groups referred to comments they sent to Medicare supporting some, but not all, of these government proposals. AHIP, for example, urged Medicare “to not adopt policies that would place limits on plan flexibility to manage post-acute care.” Final regulations are due this spring.

If concerns about the algorithms have begun to surface in legal filings and public letters to Medicare, they remain almost entirely out of sight for patients like Dolores Millam, who fell and broke her leg on a summer day in 2020.

After surgery, she began her stay in a Wisconsin nursing home on Aug. 3. Like many older patients, Millam arrived with a complicated medical history, including coronary artery disease, diabetes, high blood pressure, and chronic pain, according to court records. Her doctor had ordered that she stay off her leg for at least six weeks.

Nevertheless, an algorithm used by her insurer, UnitedHealthcare, predicted she would only need to stay for 15 days, until about Aug. 18, according to records obtained by STAT.

Just a couple days after that date, Millam received notice that payment for her care had been terminated. It was 4 p.m. on a Friday.

“I must have made — I’m not kidding — 100 phone calls just to figure out where she could go [and] why this was happening,” said Millam’s daughter, Holly Hennessy, who also received the notice.

She said she couldn’t fathom UnitedHealthcare’s conclusion that her mother — unable to move or even go to the bathroom on her own — no longer met Medicare coverage requirements.

“You try to call and reason with somebody and get explanations, and you’re talking to somebody in the Philippines,” Hennessy said. “It’s simply a process thing to them. It has nothing to do with care.” UnitedHealthcare declined to discuss its handling of Millam’s care, asserting that doing so would violate federal privacy rules.

When she received the denial, Millam could not put weight on her left leg and was being moved with a Hoyer lift, a large, freestanding harness used to transport patients who can’t use their legs. She also required 24-hour care to help with dressing, eating, and other basic tasks, according to court records.

In a note filed after payment was denied, a speech therapist wrote, “Pt. is not yet safe to live independently. She will need assistance with medication administration and supervision with ADLS [activities of daily living] due to memory deficits making her unsafe.”

Hennessy said she had no choice but to keep her mother in the nursing home, Evansville Manor, and hope the payment denial would get overturned. By then, the bills were quickly piling up.

Medicare rules call for a five-stage appeal process. The first appeal goes directly to the insurer. If denied, the patient can ask an outside entity known as a “quality improvement organization” to reconsider.

Hennessy and her mother were denied at both levels, forcing them to consider an appeal to a federal judge, a process that takes months and requires filling out reams of paperwork. Somewhere in her blitz of phone calls, Hennessy heard about the Greater Wisconsin Agency on Aging Resources, which agreed to take up her case.

In late October, Millam returned home from the nursing home after a nearly three-month recovery. The bill was almost $40,000. A few days later, her appeal came before a judge.

Hennessy, who was driving to Florida at the time, recalls pulling over for the hearing, which was held via Zoom.

The judge only asked a handful of questions of the family and representatives from the nursing home. If there was any participation from UnitedHealthcare, its opinions were not mentioned in the official record. Court documents only reference a finding from the quality improvement organization, Livanta, which had asserted that Hennessy’s mother had no “medical issues to support the need for daily skilled nursing care” when the payment denial was issued in early August.

The final ruling, issued on Nov. 25, found instead that it was the insurer that hadn’t given any good reason to deny care for a patient who was still “a safety risk.” The judge said her treatment should be paid for in full.

In the months afterward, Hennessy herself crossed the age threshold into Medicare eligibility. She said a friend who sold Medicare Advantage plans had always expected to get her business when she turned 65.

“I just told him, ‘I can’t do it. I’ve lived this nightmare,’” Hennessy recalled. The conversation ended their friendship, until the neighbor called back a couple years later following a struggle with his own Medicare Advantage insurer over a knee replacement.

“He called me to apologize for having gotten so bent out of shape,” Hennessy said. “I’ve still got friends who say, ‘Oh, I’ve got UnitedHealthcare Advantage, and it’s wonderful.’”

“Well, it is,” she said. “Until you need the big stuff.’”

This story is part of a series examining the use of artificial intelligence in health care and practices for exchanging and analyzing patient data. It is supported with funding from the Gordon and Betty Moore Foundation.

By Casey Ross and Bob Herman

Frustration was boiling into open conflict within NaviHealth, a company that uses computer predictions to help control the cost of caring for millions of older and disabled Americans on privatized Medicare plans.

The source of the outrage was not a customer or a salesperson, but an algorithm — specifically one that was being used to predict the amount of care needed by seriously ill patients. In 2021, employees raised alarms that efforts to bypass the algorithm — and pay for longer rehab stays — were getting slapped down at higher levels of the organization. Multiple cases involved patients who were still so sick they needed daily infusions to treat infections.

“It’s still happening,” one employee complained in internal communications obtained by STAT. Another added: “I had one that I had to communicate two times that was an IV and very clear on the continued stay — and it was still [denied].”

STAT previously revealed that NaviHealth’s algorithmic denials cut off insurance payments for Medicare Advantage patients struggling to recover from severe diseases and injuries, forcing them to either forgo care or pay thousands of dollars out of pocket. But a new investigation shows the denials weren’t just roiling patients and physicians — even clinical staffers within NaviHealth became increasingly distressed by the way their bosses were letting an algorithm override their discretion.

The tensions emerged after NaviHealth was acquired by Optum, a division of UnitedHealth Group, which also owns the nation’s largest Medicare Advantage insurer, according to three former NaviHealth employees. Attempts to extend care past a predicted discharge date, or authorize treatment in a more expensive facility, resulted in pushback from managers. If employees did it repeatedly, managers questioned whether they needed to be retrained.

“It was very much about following the algorithms and basically not using your clinical judgment,” one former NaviHealth medical reviewer said. “That was very different from before we were owned by Optum.”

The company’s adherence to the algorithm, even when clinicians disagreed, is a powerful example of the potential dark side of artificial intelligence and its incursion into health care. As large corporations gain control of increasingly powerful and largely unregulated AI tools, patients may find themselves at the mercy of machines that operate behind the scenes of their care, making money for big businesses even if following their advice hurts everyday people.

STAT’s reporting is based on a review of internal communications at NaviHealth, interviews with five former employees, patients’ family members, lawyers, and experts in insurance coverage and mathematical modeling. The former employees were granted anonymity because they feared professional and legal retaliation.

NaviHealth managers have said the algorithm is not used to make coverage determinations. In a statement, the company repeated what it told STAT in March: “The tool is used as a guide to help us inform providers, families, and other caregivers about what sort of assistance and care the patient may need both in the facility and after returning home. Coverage decisions are based on [Centers for Medicare and Medicaid Services] coverage criteria and the terms of the member’s plan.”

But a deeper look inside NaviHealth reveals the algorithm is, in fact, a focal point for determining coverage. It puts the process in motion, sets the deadline for coverage decisions, and provides information that can influence the outcome. In some cases, the algorithm was used in coverage decisions that may violate Medicare’s rules, which are designed to protect the safety and dignity of beneficiaries. The investigation revealed that even in instances when staffers argued that patients needed more time in rehab, NaviHealth’s physician medical reviewers deferred to the algorithm, fueling internal dissent that the denials were inappropriate and contrary to clear medical evidence.

NaviHealth also directed frontline care coordinators to time their reviews of patients’ progress so they could be discharged on the exact date projected by the algorithm, creating a kind of assembly line that staffers had to stop to get patients more care. The staffers said they didn’t always know what happened to patients next, or how they fared. But they worried that the process was designed to save money — not lives.

“It messes with you when you’re discharging people because they’re just a number to them,” a former NaviHealth patient chart reviewer said. “You’re like, ‘I’m doing that. Like, I’m part of this gear that’s turning, that’s grinding out savings.’”

NaviHealth said in its statement that it offers to speak with patients’ caregivers before payment denials are issued, and a senior clinical manager added that patients can appeal if they disagree with the company’s decision.

But the concerns voiced by NaviHealth’s former employees echo the experiences of family members on the receiving end of the denials — people forced to watch their loved ones suffer after being cut off from care prematurely.

“I felt like we were targeted,” said Gloria Bent, 78, the widow of Gary Bent, a retired physics professor at the University of Connecticut who needed rehab care after undergoing brain surgery to remove a cancerous lesion. She said she successfully appealed two of NaviHealth’s attempts to cut off payment for her husband’s stay in a skilled nursing facility in July 2022. But the company’s third denial stuck, and her husband was sent home with a fever and throbbing pain in his head and neck.

She gave him Tylenol, and tried to keep him comfortable.

But “the next morning, we could not rouse him,” Bent said. “When we were able to rouse him, he did not know who he was, where he was, or who I was.” She called for emergency help, and her husband was taken to a hospital, where he spent three weeks recovering from bacterial meningitis. He died in March 2023 at age 82.

NaviHealth declined to comment on Gary Bent’s care, citing the need for his wife to sign a privacy waiver. Earlier this year, she told her story under oath before a Senate subcommittee that launched an investigation of denials by major Medicare Advantage insurers.

Former staffers said UnitedHealth’s $2.5 billion acquisition of the company in May 2020 significantly increased the number of clinical employees. As a result, Optum sought to standardize their training and responses to questions that arose about coverage for patients’ care. Those standards, clinical staffers said, often favored authorizing the lowest-cost type of care and adhering to the algorithmically projected discharge date once a patient started getting rehab care.

If employees deviated from that formula, they came under scrutiny from supervisors who wanted to know why — and reminded them how they were expected to handle certain types of cases.

Medicare Advantage has become highly profitable for insurers as more patients over 65 and people with disabilities flock to plans that offer lower premiums and prescription drug coverage, but give insurers more latitude to deny and restrict services.

The company’s algorithm is used at multiple stages of care, starting when patients are still in the hospital. There, it’s used to measure their levels of physical and cognitive function, and to gauge the amount and type of care they may need after the hospital. Then it is run again when they arrive at a skilled nursing facility to update their clinical status, estimate the number of therapy hours they will need, and calculate an anticipated length of stay and suggest what kind of assistance they may need after leaving the facility.

Several former employees said the algorithm was helpful in understanding patients’ clinical status and likely care needs. It analyzed a wide array of data about them: primary diagnosis, additional medical problems, functional status, age, and living situation, and then compared them to similar patients in a database of 6 million people who previously received care.

The problem wasn’t the algorithm, they said. It was how it was being applied.

Under Optum, NaviHealth’s predictions about when a given patient would be ready to leave a facility weren’t used as a guidepost. They were a target discharge date the company tried to hit.

“They definitely wanted you to be conscious of the predicted discharge date,” another former NaviHealth care coordinator said. The ex-employee supported the use of the algorithm because it had the intent of more quickly getting patients home, which is where they generally want to be and carries lower risks of infections and other complications.

However, the former worker acknowledged NaviHealth and its insurance company clients — which include NaviHealth’s sister company, UnitedHealthcare, and Humana — have a clear financial incentive to get patients out of rehab as fast as possible. That’s especially the case with Medicare Advantage, a product that has the highest underwriting profit margins of any type of health insurance. “All sides of the company want to make money,” the former care coordinator said.

Under Optum, NaviHealth’s predictions about when a given patient would be ready to leave a facility weren’t used as a guidepost. They were a target discharge date the company tried to hit.

“They definitely wanted you to be conscious of the predicted discharge date,” another former NaviHealth care coordinator said. The ex-employee supported the use of the algorithm because it had the intent of more quickly getting patients home, which is where they generally want to be and carries lower risks of infections and other complications.

However, the former worker acknowledged NaviHealth and its insurance company clients — which include NaviHealth’s sister company, UnitedHealthcare, and Humana — have a clear financial incentive to get patients out of rehab as fast as possible. That’s especially the case with Medicare Advantage, a product that has the highest underwriting profit margins of any type of health insurance. “All sides of the company want to make money,” the former care coordinator said.

NaviHealth’s care coordinators could still recommend that patients remain in rehab. But that didn’t always mean higher-level physician medical reviewers would take their advice. In 2021, NaviHealth staffers operating in at least two states noticed an unusual pattern of denials: Seriously ill patients they said should stay in rehab were getting cut off from payment.

Multiple employees raised alarms, noting in an internal chain of communications reviewed by STAT that they had clearly marked the patients were not ready to be discharged. In two instances flagged in the communications, patients still required additional days of IV antibiotics when the algorithm said they could be sent home. The federal Centers for Medicare and Medicaid Services defines IV treatment as a skilled service, which means it would be covered under traditional Medicare.

“These IVs are pretty clear cut-and-dried with CMS,” one employee opined in the communications, complaining their recommendation for a patient to keep getting care was overturned.

Providers have reported similar denials involving patients on IV medicine. In testimony filed as part of the Senate investigation, LeadingAge, a group that represents nursing homes and other providers, flagged a case in which NaviHealth’s algorithm predicted a patient would be discharged after 13.2 days, even though the individual was prescribed to be on IV medicine for another 16 days past the 13-day mark.

“Under traditional Medicare … the patient would have received coverage for all the days that they required IV medication,” the organization wrote. “But the plan did not cover it.”

Casey Schwarz, a top attorney at the Medicare Rights Center, a nonprofit that helps Medicare beneficiaries with their coverage, cautioned that it’s difficult to say whether these denials violated Medicare law without knowing all the details of the cases. But patients would be entitled to that care if they met all of Medicare’s criteria for skilled nursing care.

“In situations like the one you’re describing, we have helped people file appeals of that decision, and they’ve been successful,” Schwarz said.

In one case reviewed by STAT, a former NaviHealth staffer said the family wasn’t comfortable administering an IV at home. When family members indicate they cannot safely administer IV medicines once a patient leaves a nursing home, it is harder to justify a patient’s discharge, experts in medical necessity criteria said.

“That’s a reason to not discharge them,” said Prashant Vaishnava, a physician and former medical director at EviCore, another firm that helps insurers manage the cost of care. “If there’s not a way to continue the IV antibiotics at home, or there are significant barriers for them to do so, that seems like a deviation from what our coverage criteria are or should be. We’re doing a disservice to the patient at that point.”

As the denials piled up, employees pressed for explanations. In addressing complaints that one physician reviewer repeatedly overturned recommendations for continued care, a NaviHealth supervisor told employees the reviewer “isn’t necessarily reading and looking at everything we’re doing.” The supervisor told the employees to keep doing their best to document patient cases, adding that poor decisions were inevitable in any process for handling such a large volume of complicated cases.

But the likelihood of errors increases significantly if physician reviewers are not reading the recommendations of frontline care coordinators and the clinical notes their opinions are based on, Vaishnava said. Those notes might contain details about how much pain a patient is reporting, or whether they’ve made good progress in physical therapy sessions.

“The medical reviewer is responsible for all of the information that’s in front of them, whether it’s 300 pages or whether it’s two pages,” he said. “There is a human being reflected in those pieces of paper.”

A NaviHealth senior clinical manager said the company was not aware of the complaints about denials issued to patients on IV antibiotics, or the complaints that a physician medical reviewer was not reading care coordinators’ reports. The manager said physician medical reviewers are trained to read all of the documents relating to a patient’s case and make decisions based on Medicare’s coverage rules.

But by inserting the algorithm into that process — along with its target discharge date — NaviHealth is empowering a tool with only a limited vantage point of the patient and their care needs. Its projected discharge date is based on care received by similar patients in its database. However, NaviHealth has not disclosed exactly where the data came from, what kinds of patients are included, or how they are stratified by age, race, gender, and income level. The company also couldn’t point to any peer-reviewed studies conducted to assess the algorithm’s accuracy in predicting discharge dates as measured against the outcomes of patients.

NaviHealth’s senior clinical manager said the company’s database has been built over many years with reports that providers generate about the type and length of care patients received, and their outcomes.

Experts in mathematical modeling said the lack of clarity on the underlying reference data makes it impossible to gauge the validity of the algorithm’s predictions. Furthermore, because its projected length of stay is based on an average, patients who are sicker than average could get cut off from care prematurely.

“The people who are the neediest are going to suffer,” said Ruth Etzioni, a biostatistician at Fred Hutchinson Cancer Center in Seattle, who added that patients’ conditions often change significantly, and suddenly, during the course of their care. “We cannot know at the admission what will happen later.”

Gloria Bent said NaviHealth called to notify her of her husband’s discharge date, July 4, 2022, before he had even been evaluated by the clinical staff at his skilled nursing facility in Connecticut. The date was only halfway into the eight weeks of rehabilitation Gary Bent’s doctors said he would need after his brain surgery.

The call was jarring, Bent said. Not only was she being told her husband might not get the care he needed, the message was coming from NaviHealth, a company she knew nothing about. The couple’s Medicare Advantage policy was with UnitedHealthcare — not NaviHealth — and she had read that under Medicare rules, her husband was entitled to get 100 days of skilled nursing care.

The person on the phone did not explain the involvement of an algorithm or discuss how the discharge date was calculated. Instead, Bent said, the NaviHealth staffer told her that she should make arrangements to put her husband in long-term care by July 4. If she preferred to care for him herself, she needed to consider moving out of her home within three weeks, because it was not wheelchair accessible.

“I was stunned,” Bent said. She relayed the information to the staff of the nursing facility, who explained it probably wasn’t the last time Bent would hear from NaviHealth. “And they were absolutely right,” Bent recalled. During her husband’s stay, she received two more calls from NaviHealth employees who floated different discharge dates.

NaviHealth did not respond directly to Bent’s assertions about her husband’s care. But the company said its policy is to run the algorithm after a patient has been evaluated by a facility’s staff, and then work with the family to make arrangements to safely discharge the patient.

The company backed off from the initial July 4 date, but started issuing payment denials on July 23, about six weeks after her husband began his recovery. Bent said he was becoming more mobile and engaged in physical therapy sessions, but wasn’t ready to go home. She said she beat back two of NaviHealth’s payment denials by appealing to a Medicare contractor that adjudicates coverage disputes.

A third denial came just a few days after she won the second appeal — and this time, she lost. Bent said the chain of events that unfolded next is still a traumatic blur. Her husband arrived home five hours later than expected, complaining of pain in his head and neck. He was running a fever and couldn’t move himself from gurney to wheelchair.

“It was a long, hard night,” Bent said. When her husband woke up disoriented the next morning, unable to recognize his wife and daughter, a call to local EMTs for transport back to the hospital was the only option.

After he was hospitalized again, Gary Bent eventually returned home, and died several months later. But Gloria Bent still laments that his dying days were not lived better, free from the dictates of an algorithm, financial fears, and seemingly arbitrary deadlines for his recovery.

“The layer of trauma that comes from dealing with these companies just piles on,” she said. “We’re still trying to get out from under the emotional fatigue of his last 10 months — and it didn’t have to be that way.”

This story is part of a series examining the use of artificial intelligence in health care and practices for exchanging and analyzing patient data. It is supported with funding from the Gordon and Betty Moore Foundation.

By Casey Ross and Bob Herman

The nation’s largest health insurance company pressured its medical staff to cut off payments for seriously ill patients in lockstep with a computer algorithm’s calculations, denying rehabilitation care for older and disabled Americans as profits soared, a STAT investigation has found.

UnitedHealth Group has repeatedly said its algorithm, which predicts how long patients will need to stay in rehab, is merely a guidepost for their recoveries. But inside the company, managers delivered a much different message: that the algorithm was to be followed precisely so payment could be cut off by the date it predicted.

Internal documents show that a UnitedHealth subsidiary called NaviHealth set a target for 2023 to keep rehab stays of patients in Medicare Advantage plans within 1% of the days projected by the algorithm. Former employees said missing the target for patients under their watch meant exposing themselves to discipline, including possible termination, regardless of whether the additional days were justified under Medicare coverage rules.

The stringent performance goal was part of a broader effort to reduce expensive nursing home care for frail patients with privatized Medicare plans, the internal documents show. The strategy was conceived and executed by former top Medicare officials whose policies became a blueprint for UnitedHealth to reap hundreds of millions of dollars annually by shredding the government’s safety net with payment denials backed by an algorithm.

The federal government has largely allowed these practices to proliferate. But in responses to questions about UnitedHealth and NaviHealth, the Centers for Medicare and Medicaid Services told STAT that it “is looking into these allegations and may take necessary enforcement or compliance actions.”

STAT previously reported UnitedHealth began limiting employees’ discretion to deviate from the algorithm after it bought NaviHealth in 2020. The newly obtained documents show that, since then, executives have sought to almost entirely subordinate clinical case managers’ judgment to the computer’s calculations. That has resulted in inflexible coverage decisions that legal experts say may violate longstanding case law and regulations that govern Medicare benefits.

Three former case managers said the individual stories behind the algorithmic denials were haunting: An older woman found in the laundry room by her grandson after a stroke, her right side paralyzed, was allotted 20 days of rehab by the algorithm, when the average for severely impaired stroke patients is almost double that. A 78-year-old legally blind man who needed care for a failing heart and kidneys, and then fell in the nursing home, was granted 16 days. Another older man nearing his discharge date after knee surgery was expected to learn how to “butt bump” up and down stairs. If case managers disagreed, and tried to extend a patient’s stay, they ran the risk of missing their targets.

The final call on whether a patient receives more care — or a payment denial — is left up to one of NaviHealth’s physician medical reviewers. The case managers working underneath them are in charge of assembling the medical records, running the algorithm, and deciding whether to press for additional care or recommend to the medical reviewer that the patient get a denial. With a 1% target over their heads, former case managers said, making that decision too often meant choosing between their job performance and their conscience.

“By the end of my time at NaviHealth I realized — I’m not an advocate, I’m just a moneymaker for this company,” said Amber Lynch, an occupational therapist and former NaviHealth case manager who said she was fired earlier this year for failing to meet performance goals. “And that is not why I went into health care. I went into health care to help people, not to say, ‘Well, we’ve got all the money, see you later.’”

UnitedHealth’s emphasis on adhering to the algorithm, called nH Predict, is a cautionary sign to an industry becoming increasingly infatuated with data-driven decision-making. Some technology leaders have warned apocalyptically of a future in which sentient artificial intelligence systems imperil humans and drive them into extinction. But in health care, less sophisticated technology is already threatening basic human safety. Algorithms and AI tools are being empowered to help make consequential decisions about the treatment of patients who have a limited understanding of these products — if they’re even aware of them — and almost no ability to question their conclusions.

In a statement responding to STAT’s findings, UnitedHealth said NaviHealth has helped countless patients, families, and caregivers by giving them information about their care they wouldn’t have received otherwise. “The assertions that NaviHealth uses or incentivizes employees to use a tool to deny care are false,” the statement said. “Adverse coverage decisions are made by medical directors and based on Medicare coverage criteria, not a tool or a performance goal tied to any single quality metric.”

The company also said it does not base reviews of employees solely on their adherence to the algorithm’s projected length of stay. “We look at a wide variety of factors in considering a clinician’s performance that focus on the overall quality of their work and, most importantly, providing outstanding support to the people they serve,” the statement said.

By restricting the days patients spend in nursing homes, UnitedHealth saves hundreds of millions of dollars every year on their care, increasing profits for one of its most lucrative health insurance products. NaviHealth doesn’t just manage nursing home care within UnitedHealth’s plans, but also serves large Medicare Advantage insurers such as Humana and many regional health plans — potentially affecting the care of more than 15 million people, or half of all Medicare Advantage enrollees. Its strategy is integral to a growing national business at UnitedHealth to deliver more care in patients’ homes using a network of providers it has spent billions of dollars acquiring in recent years.

With the algorithm on its side, the insurance giant could squeeze unpredictable recovery times into a predefined window. The reality is that people’s pace of recovery from serious illnesses, trauma, and operations is widely variable. Yet UnitedHealth’s enrollees had to get better on the company’s schedule, not on their own, even though many were struggling after cancer surgery, paralysis, smashed hips and shoulders, and other injuries.

“You can maximize your profitability if you’re able to get people out of expensive care settings as quickly as you possibly can,” Spencer Perlman, a longtime health care markets analyst and managing partner at Veda Partners, an investment adviser and consulting firm.

Medicare Advantage was designed to cut out unnecessary care, including situations where patients stay in nursing homes for long stretches when they don’t need to be — a problem that has plagued Medicare for decades. Perlman said there is still “undeniably waste” in traditional Medicare. But the pendulum has swung so far in the other direction that capturing cost savings within the $460 billion Medicare Advantage program, even at the expense of sick beneficiaries, has become an explicit goal for the industry.

“If UnitedHealth is using [NaviHealth’s] algorithms as gospel … that’s not clinical decision-making,” Perlman said. “That’s aggregating data and using an algorithm to make a decision that has nothing to do with the individual themselves, which is problematic and antithetical to Medicare policy.”

The government has attempted to prevent stinting on care by capping profits Medicare Advantage insurers can retain. Health care companies have gotten around that by becoming “vertically integrated,” which is when insurers like UnitedHealth also own health care providers, like physicians and post-acute companies. If money has to go out the door, the thinking goes, it may as well go to another part of the company. “There’s no profit cap on the provider side, so you can have greater profitability by paying yourself,” Perlman said.

STAT’s latest reporting is based on a review of internal NaviHealth documents and communications, interviews with three former employees, as well as with lawyers, policy experts, and people working in the post-acute care setting. Two of the former employees requested anonymity because of concerns about professional and legal consequences of speaking publicly about confidential matters.

The former employees, all licensed therapists with extensive experience with rehab care, said they joined NaviHealth for multiple reasons. Their jobs as case managers allowed them to work remotely, instead of being tied to a clinic or office. It also offered the opportunity to work directly with patients, whom they regularly spoke with in person or by phone, to help guide them through their rehab stays.

But they said their ability to advocate for patients, and apply their clinical judgment, diminished after UnitedHealth’s Optum division purchased the company in May 2020 in a deal valued at $2.5 billion. Managers seemed to have new marching orders. There was less leeway to deviate from nH Predict and an intense focus on meeting performance targets that kept getting narrower. When patients appealed payment denials, a process that typically added several days to rehab stays, the managers wanted to know why. What went wrong? Why weren’t they ready to go home?

If the answer was obvious to the frontline case managers — the patients were still severely ill and needed more care — it wasn’t acceptable to their supervisors. They wanted to know what could have been done differently to get them out by the date predicted by the algorithm.

“Things changed after Optum took over,” said one former case manager. “Instead of the [algorithm] being a tool that was used to anticipate a length of stay, it became a tool that you’d better make it happen or there’s consequences.”

Behind the algorithm: Toyota’s approach

In early 2020, just before it acquired NaviHealth, Optum recruited the perfect candidate to take charge. Patrick Conway was not just a seasoned insurance executive. He also had served as Medicare’s top clinician, and was a leading advocate for rooting out unnecessary care during the Obama administration.

Conway espoused a philosophy drawn from an unlikely source: the automotive industry’s Toyota Production System. It has gained popularity in health care, and is known as lean management. The Toyota approach focuses on improving quality and productivity by eliminating low-value work. Instead of “Do no harm,” its inventor, the Japanese industrial engineer Taiichi Ohno, started with a much different premise: “First reduce waste.”

Toyota pioneered lean management to crank out sedans for less money and with fewer defects. In following this strategy, NaviHealth could achieve similar results for chronically ill patients recovering from hip surgeries and strokes.

By the time Optum bought NaviHealth, the company had already built a version of an algorithmic assembly line for managing Medicare Advantage members. NaviHealth was founded by Tom Scully, who ran Medicare under former President George W. Bush and played a pivotal role in creating Medicare Advantage. He sold the company in 2015, part of a chain of deals that led to its acquisition by Optum.

Conway took the reins of Optum Care Solutions, the business unit overseeing NaviHealth, after two years as CEO of Blue Cross Blue Shield of North Carolina. As an evangelist of the so-called value-based care school of thought, he aligned with Scully on the core mission to use technology as a lever to move nursing care out of costly institutions and into patients’ homes.

From his perch at Optum, Conway now had the chance to make the assembly line run faster and smoother — and implement it on a national scale.

In one training document, the instructions to employees were clear and precise: It directed case managers to discuss the algorithm’s predictions twice with both patients and clinicians overseeing their care in nursing homes. It directed them to discuss its tentative discharge date on the first visit, so patients were not “surprised” when they received a payment denial. A second conversation should take place 24 to 48 hours before the denial was issued.

Each conversation, every interaction, had to be logged and audited by managers, to ensure the process of moving patients through the nursing home was proceeding apace. Conway explained that keeping to the schedule was better for patients.

“People in general would rather be in their home, if possible, so how do we get them home and have new care modalities centered around them that keep them healthy and at home?” Conway asked in a podcast interview posted on NaviHealth’s website. “If they go to a nursing home, how do we get them out as soon as possible?”

UnitedHealth officials declined STAT’s requests to interview Conway for this story.

At the very top of UnitedHealth, executives enthusiastically embraced Conway and his strategy. Speaking at a banking conference in 2021, UnitedHealth CEO Andrew Witty asked, “How do we avoid people ending up in long-term care institutions and the like?” After noting that investors should “look at the businesses that Patrick Conway is building up within our Optum at Home portfolio,” Witty said these types of projects are “the next wave of Optum growth, but very much at large scale.”

Wyatt Decker, another high-ranking UnitedHealth executive, also promoted the original NaviHealth deal to Wall Street in 2020. He said then that NaviHealth represented a chance to overhaul post-acute care, which “has not been well-managed, both in terms of the patient experience and expenses.”

UnitedHealth declined to comment on how much money it saves annually through the use of NaviHealth. In a paper published in 2022, Conway and co-authors from Optum reported that NaviHealth achieved average savings of more than 20% per episode of care. Based on the size of the company’s Medicare Advantage enrollment, and average skilled nursing costs and national utilization rates, STAT conservatively calculated the annual savings to be several hundred million dollars a year.

Under Conway’s leadership, NaviHealth tightened employee performance goals. In 2022, the company set a target for case managers to keep patients’ stays in nursing homes within 3% of the days projected by the algorithm, according to confidential internal documents. Earlier this year, the target was narrowed to less than 1%.

The new target was listed in a document outlining a case manager’s performance goals for 2023, stating that the employee was expected to “ensure appropriate utilization” by maintaining a variance of “<1% overall” from the algorithm’s estimated length of stay. The variance is calculated on an aggregate level, so that if 10 patients were allotted 100 days by the algorithm, they would collectively need to stay no more than 101 days for the case manager to meet the target. The document explained that the 1% included “short stays,” which occur when patients return to the hospital quickly or opt to recover at home. The case manager’s goal for 2022 was less than 3%, but short stays were not mentioned in that year’s document.

The 1% target was also listed in a presentation shared with a larger group of employees this year. The presentation contained a detailed breakdown of employee performance metrics. In addition to length of stay, it calculated the average number of patients they were managing per day, the percentage of patients they spoke with in person versus by phone, and a month-by-month breakdown of how they fared against the 1% target.

Early in the year, the case managers included in the presentation met the target, but the numbers crept above the line in the summer.

A spokesperson for NaviHealth said the 1% target is based on company-wide data and is one of several metrics used to assess employee performance. The spokesperson said if case managers missed the target, supervisors would take a closer look at the possible reasons, and it wouldn’t be applied to the care of an individual patient, or used to judge an employee’s performance in isolation.

However, the former case managers said supervisors were laser-focused on their length-of-stay numbers. “It was definitely held to me, ‘Amber, you have to do less than 3.’ Then, by the end of the time I was working there, it became less than 1%,” said Lynch, who was fired in the spring. “And I’m like, ‘That’s insane. It doesn’t take into account other comorbidities that may flare up. Or what if they have hospital-acquired pneumonia? Or what if they have Covid?’”

Lynch, who lives in Georgia, said the company’s metric-driven approach plunged her into a moral crisis. When she visited rehab facilities, it became hard to meet the gaze of nursing home managers and patients who knew her presence meant the impending issuance of a payment denial.

“I would hear this so much: ‘You’re going to cut me, you’re going to cut my funding,’” Lynch said. Some family members screamed at her. She got used to telling them she was just a messenger, a middleman, for NaviHealth. She explained the final decision on whether to issue a payment denial fell to one of the company’s physician medical reviewers.

Lynch and the other case managers interviewed by STAT said most of their requests to extend a patient’s stay were denied — if case managers dared to ask for extensions at all. Lynch recalled one patient recovering from a knee surgery who wasn’t able to negotiate the stairs in his home when the projected discharge date arrived. Instead of extending his stay — which Lynch said Medicare rules didn’t seem to support — a manager instructed her to ask the nursing home if it had trained the man to “butt bump” the stairs to get up and down.

“I felt terrible asking the director of rehab that,” Lynch recalled. “She looked at me like I had two heads, and she’s like, ‘He is 78 years old. He’s not going to do that. He’s not safe to climb the stairs yet. He’s not doing it. We’re not going to have him butt bump the stairs.’”

Lynch said she was eventually fired for failing to meet performance goals, including the length-of-stay target, and for missing deadlines for filing documentation. She said her daily patient load caused her to fall behind. The company put her on a performance plan and gave her opportunities to meet the goals. But she said meeting the numbers meant disregarding the needs of patients.

“It takes the person out,” she said of Optum’s metric-driven management. “It’s all about money and data points. It takes the dignity out of the patient, and I hated that.”

A recipe for bias — and premature denials

In a makeshift home office framed by clothes and scattered papers, a NaviHealth case manager pulled up a patient’s file on a recent morning. The patient, a man in his 80s, was recovering from a broken shoulder and suffered from a long list of medical problems, including heart failure, kidney disease, diabetes, and trouble swallowing.

As he lay in a nursing home miles away, the computer got to work on his care.

The case manager typed in the primary diagnosis and other medical problems, and clicked through a series of prompts and drop-down menus, tapping out answers to a series of questions: How much trouble did the patient have eating with a spoon? Could the patient read and understand a newspaper? How much help did he need getting to the toilet?

Within minutes, the computer spit out a series of numbers, including an estimated length of stay of 17 days. The case manager, having seen the patterns play out countless times over several years, was instantly skeptical.

Had the software really accounted for everything? Its predictions were based on similar patients in a 6 million-person database compiled by NaviHealth. But who were the patients in the database? Had they gotten the right amount of care? Did their outcomes translate to this elderly man getting care years later, in a different place, for a slightly different mix of problems?

A NaviHealth spokesperson said the database includes information about the care of other Medicare Advantage members, including details about their location and level of “social vulnerability.”  The company declined to further elaborate on the exact sources or contents of the database.

The elderly man was just one of 27 patients on the case manager’s list, a docket that during one recent week had swelled to 40. For each patient, documents had to be gathered and meetings held. NaviHealth’s supervisors had also been emphasizing the need to personally engage with each Medicare Advantage member, or their family.

“Do you think I was able to process everything correctly and call everyone correctly the way I was supposed to?” the case manager asked. “No. It’s impossible. So who loses in that situation? Members. No one can be that fast and that effective and capture all of the information that’s needed.”

There is also a deeper flaw involved in using a prediction based on the length of stays of prior patients to make care and coverage decisions about future ones: The measure itself is inherently imprecise, and fraught with bias.

“Length of stay is not some biological variable,” said Ziad Obermeyer, a physician and professor at University of California, Berkeley, who researches algorithmic bias. He noted that patients’ stays are influenced by factors that stretch far beyond their clinical condition, such as the extent of family support, the accessibility of their homes, and their ability to afford out-of-pocket costs.

“People are being forced out of the [nursing home] because they can’t pay or because their insurance sucks,” Obermeyer said. “And so the algorithm is basically learning all the inequalities of our current system.”

Such inequalities are only reinforced, and perpetuated, if the algorithm is used as a blunt instrument to make such a precise cut in the care of a patient. Just as when a butter knife is used to slice a tomato, the result is a mess, as it was when NaviHealth’s algorithm intervened in the care of Michelle Freed’s 69-year-old mother.

Freed said her mother, who suffered an acute back injury while working out with a personal trainer, was sent to a skilled nursing facility in northern California following surgery that led to an infection, sepsis, a coma, and then a halting climb back. The arrival of the first payment denial, when her mother was still receiving intravenous antibiotics, sent her family into a panic — and a never-ending phone loop.

“I was livid,” said Freed, whose mother had a Medicare Advantage plan with UnitedHealth. Perhaps it was her level of anger, or the force of her arguments, but Freed, who is training to be a nurse practitioner, persuaded a case manager from NaviHealth to level with her about the algorithm and her mother’s care.

“She’s like, ‘You don’t get it, we have a very narrow biplane to ride in,’” Freed recalled the case manager telling her. “‘You have to stay in your lane or else.’” She took it to mean that, regardless of the obstacles or complications, her mother’s care had to follow a certain path, on a predetermined timeline.

Freed said the nursing home successfully appealed the first payment denial, and prolonged her mother’s stay, but the denials kept coming, until her mother was sent home against her family’s wishes in October. She was still using a feeding tube 15 hours a day and couldn’t go to the bathroom on her own. Home health came a few times a week for short stints of therapy, but Freed said her family had to hire a private caregiver, at their own expense, to manage her daily needs. Even with the extra help, her mother’s inability to change position resulted in pressure sores and other problems, and eventually another trip to the emergency department.

“UnitedHealth and NaviHealth have failed my mom,” Freed said, noting that the companies also refused to pay for her mother’s stay in a hospital-based rehab unit. “Instead of doing what’s right by the patient and our family … they just outright denied us.”

Bending Medicare’s rules to fit the technology

Optum’s executives knew nursing homes would resist the data-driven approach to patient care. But their solution wasn’t to compromise, or find common ground. It was to fight back hard.

One training document updated soon after the company’s May 2020 acquisition of NaviHealth delivered stark instructions:

• If a nursing home balked at discharging a patient with a feeding tube, case managers should point out that the tube needed to provide “26 percent of daily calorie requirements” to be considered as a skilled service under Medicare coverage rules.
• If a nurse took a broader tack, and argued a patient was unsafe to leave, case managers were instructed to counter, in part, that the algorithm’s projections about a patient’s care needs, and readiness for discharge, are based on a “severity-adjusted” comparison to similar patients around the country. “Why would this patient be any different?” the document asks.

A NaviHealth spokesperson said the document was designed not as absolute but to elicit data and evidence from nursing homes to justify a patient’s continued stay. Case managers told STAT, however, that the talking points glossed over the complexity of patients who come with different clinical problems, life situations, complications, and abilities to cope and recover from the debilitating illnesses.

Overriding these important distinctions through adherence to an algorithm may violate more than three decades of case law, Medicare experts say.

In 1987, a judge ruled in Fox v. Bowen that any patient in a nursing home has a clear right to receive “an individualized assessment of his need for daily physical therapy based on the facts and circumstances of his particular case.” The federal government, therefore, “cannot permit … intermediaries to use blanket rules not supported or authorized by any applicable law or regulations to deny what otherwise might be meritorious claims.”

NaviHealth says its physician reviewers meet the standard of an individualized assessment. But Toby Edelman, a senior policy attorney at the Center for Medicare Advocacy who has represented Medicare patients in post-acute settings for 46 years, asserted that NaviHealth’s use of algorithms, enshrined in performance goals, would run afoul of that ruling.

“Fox v. Bowen says you can’t have rules of thumb. Well, these algorithms are much more than rules of thumb,” Edelman said. “It’s certainly not individualized if how many days you get [in a nursing home] is in the algorithm.”

Some of the AI-guided decisions also may sidestep other Medicare coverage rules. For example, Medicare patients don’t have to be making progress to keep getting rehab, even if that’s what plans are telling members.

“As long as it is helping you maintain your status, you’re supposed to be able to get the rehab,” said Charlene Harrington, a professor emeritus at the University of California, San Francisco, who has studied nursing homes. Indeed, Medicare’s rules state that skilled nursing care may be necessary “to maintain the patient’s current condition.”

Medicare has acknowledged the predicament for patients, telling health insurers explicitly that starting in 2024, they cannot rely on their own technology to deny care. “Organizations must ensure that they are making medical necessity determinations based on the circumstances of the specific individual … as opposed to using an algorithm or software that doesn’t account for an individual’s circumstances,” regulators said in April. Insurers also will have to publicly disclose the inner workings of their tools.

Meena Seshamani, a top official within CMS, the federal agency that regulates Medicare Advantage plans, told STAT that insurers must follow clinically appropriate guidelines. “We want to make sure that people have access to the care that they need,” she said in a brief interview, in which she did not respond directly to specifics about UnitedHealth and NaviHealth.

CMS also responded to STAT’s questions in writing. “An algorithm or software tool can be used to assist MA plans in predicting a length of stay, but that prediction alone cannot be used as the basis to terminate post-acute care services,” CMS said. The agency added that a qualified clinician must reassess a patient’s condition before cutting off services.

For companies that break the rules, Medicare could issue fines, which usually are inconsequential to a health plan’s finances, or halt Medicare Advantage insurers from enrolling more people, which rarely happens.

In addition, CMS officials sent a memo last month to Medicare Advantage plans, obtained by STAT, that said the government would start holding “strategic conversations” with them this month to explain how they can comply with coverage criteria and the use of their technology. Audits will start in January.

Good luck appealing, ‘they want you out’

Patients had one way to quickly push back against a payment denial: They could appeal to an outside contractor that adjudicates coverage disputes for Medicare. But even if they won, NaviHealth made sure their victories were short-lived.

One former case manager said a supervisor directed her to immediately request updated medical records for patients who won their appeals. That would enable the company to restart its review process.

“And 99.9% of the time, we’re going to turn right back around and issue another [denial],” the former case manager said. “Well, you won, but OK, what’d that get you? Three or four days? You’re going to get another [denial] on your next review, because they want you out.”

Internal communications show NaviHealth supervisors registered concern about rising numbers of appeals. In one recent meeting, a supervisor presented data to case managers showing that appeals had added tens of thousands of days onto rehab stays in the first several months of 2023. The numbers were broken down between Livanta and Kepro, the contractors that adjudicate the cases for Medicare, showing how often they sided with patients.

The takeaway, one case manager involved in the conversation said, was unmistakable: Appeals were bad for business.

“It’s adding days to stays, it’s costing them money, and it’s negating our model, because our model is supposed to be high touch,” the case manager said.

A spokesperson for NaviHealth said the issuance of repeated denials to the same patient can happen for multiple reasons. In some cases, she said, a lingering barrier to discharge, such as the installation of a ramp for mobility-impaired patients, gets resolved; in others, the company simply disagrees with the finding by Medicare’s contractor that the patient meets coverage criteria.

Appeals are rising across the Medicare Advantage industry. The number of expedited appeals filed to protest termination of rehab services by Medicare Advantage plans more than doubled from roughly 96,000 in 2020 to more than 227,000 in 2022, according to data reported by Medicare’s contractors. The exact reason for the increase is unclear, but it tracks with increasing enrollment in Medicare Advantage plans, and the advent of care management firms using algorithms to scrutinize rehab care. NaviHealth was the largest among them.

To diagnose the problem, NaviHealth is asking case managers to confront patients about the reasons for their appeals.

But even asking for a rationale felt borderline inappropriate, if not absurd, the case managers said. Patients with broken legs and shoulders wanted to be able to dress themselves and go to the bathroom alone. Patients with feeding tubes wanted to feed themselves. Each one wanted to recover, to have dignity, and to push back when dignity was denied.

“That is their right to appeal if they disagree” with a denial, one former case manager said. “That should be all the reason you need. I will not call them and ask, ‘Why did you appeal?’ That’s almost like a threat.”

The NaviHealth spokesperson said the company is not trying to intimidate patients, or interfere with appeal rights. Instead, it might ask about the reasons for appeals to understand the unresolved issues that led to them, so the company can deliver better services.

But patients who lose their appeals often end up back at home or in long-term facilities. Home care providers say many are not ready for that transition.

“We do get patients who come home too soon,” said Jennie George, the assistant director of home care at Citizens Memorial Healthcare in Bolivar, Mo. “They maybe did or didn’t try to appeal, but that appeal was denied. We’ll go out, and we’ll see them, and it’s evident that they still need to be in the facility.” George witnessed this personally with her grandmother who was on a Humana Medicare Advantage plan and received denials for a nursing home stay after a hip replacement surgery.

Lynch said the company wasn’t just tracking appeals on an aggregate level, but down to the individual case manager. She said the message from her supervisor was direct: “‘Your appeals are high. We expect some appeals, but not this much,’” Lynch recalled the supervisor telling her. “‘Are you asking why they’re appealing?’ And I said, ‘Well, often it’s because they didn’t feel it was safe to go home, they felt like they needed more therapy.’”

Even if she agreed, Lynch said, the focus on troubleshooting appeals had more to to do with her performance than the needs of the patient. “My impression [from NaviHealth’s management] was if you had educated more, or if you had talked to them more, or shared this about the [nH] Predict or whatever, then maybe they would feel like all avenues have been explored.”

Freed said she came to expect payment denials for her mother on Thursdays. They came in a loop. NaviHealth would issue the denial, the nursing home would appeal, and the denial would get overturned. Then a new denial would get issued the following week, giving her family only a fleeting respite between bureaucratic battles.

“My family is overwhelmed,” she said. “And my mom is plateauing in her care instead of advancing.”

Why we took this story on

For several years, STAT reporters have examined the growing use of artificial intelligence in health care. As part of that work, they learned last year that Medicare Advantage plans were denying care based, in part, on predictions generated by a computer algorithm. Our early reporting revealed that the nation’s largest health insurance company, UnitedHealth, was using the unregulated algorithm in making life-altering coverage decisions. We wanted to know how and why.

Now, in this article, we report that this was part of a plan that boosted profits at the expense of patients: UnitedHealth pressured clinical employees to follow the algorithm’s length-of-stay determinations, even when they, and the patients’ own clinicians, disagreed. Our investigation revealed the algorithm’s use was resulting in premature payment denials for patients recovering from strokes, cancer, and other serious illnesses. We pursued this story because people deserve to know what they are signing up for, and how their Medicare Advantage coverage works when they are at their most vulnerable.

This story is part of a series examining the use of artificial intelligence in health care and practices for exchanging and analyzing patient data. It is supported with funding from the Gordon and Betty Moore Foundation.