The Claim Was Processed

One in five insurance claims is denied. Fewer than one in a hundred ever appealed. The system does not need to be correct. It only needs you to be exhausted.

I. OPENER

I am the Vice President of Claims Optimization at UnitedHealthcare. I want to be precise about my title because precision is what I sell. I do not work in Claims Denial. There is no department called Claims Denial. There is no door with that nameplate, no Slack channel with that name, no budget line item that contains the word “denial.” The word we use is “optimization.” Optimization is the practice of ensuring that every claim is processed according to the appropriate clinical and actuarial criteria. That sentence means: I deny claims. But the sentence is twenty-three words long, and by the time you reach the end of it, the word “deny” has been replaced by a system. That’s optimization.

The company reported $87.1 billion in revenue last quarter.¹ The CEO’s most recent disclosed compensation was $26.3 million.² He has since departed. Our denial rate is approximately 20%.³ One in five claims submitted by a physician, on behalf of a patient, for a service the physician determined was medically necessary, is denied by my department. The physician went to medical school. My department went to actuarial school. We win four out of five times. The fifth time, we also win, because winning and losing look identical in our system. The claim was processed either way. That’s parity.

The claim was processed.

---

II. CONTEXT

Before I explain what my department does, I need to explain why it exists. Not how. The how is an algorithm. The why is older than any algorithm.

In 1965, Lyndon Johnson signed Medicare into law. The federal government began paying for healthcare for Americans over 65. The program had a problem immediately: it paid doctors and hospitals whatever they charged. There was no mechanism for saying no. There was no department. There was no VP of Claims Optimization. There was a checkbook, and the checkbook was open, and costs rose 300% in the first decade.⁴

The government’s response was not to reform the payment system. The government’s response was to insert a layer between the patient and the payment. That layer was the insurance company. The insurance company’s job was to manage the cost. “Manage” meant review. “Review” meant question. “Question” meant deny.

Richard Nixon signed the Health Maintenance Organization Act in 1973.⁵ The theory was elegant: if one organization managed both the insurance and the delivery of care, it would have an incentive to keep patients healthy. Prevention over intervention. The theory lasted approximately four fiscal quarters. The practice was that HMOs made money by collecting premiums and not spending them. The less care they authorized, the wider the margin.

By the late 1970s, HMOs were denying referrals, limiting hospital stays, and requiring something called prior authorization -- a process by which a doctor’s medical decision was reviewed, second-guessed, and frequently overruled by someone employed by the insurance company. The doctor examined the patient. The insurer examined the cost.

Through the 1980s, employers pushed their workforces into managed care plans to control spending. The employee did not choose the HMO. The employer chose the HMO that offered the lowest premiums. The HMO offered the lowest premiums by denying the most claims. This is not a conspiracy. This is a supply chain.⁶

Two companies dominated the criteria market: InterQual, which published clinical decision trees, and Milliman, which published care guidelines -- a phrase that sounds like it describes how to provide care and in practice describes how to limit it.⁷ The Milliman guidelines told the nurse how many days a patient should stay for a given diagnosis. The nurse compared the chart to the guideline. If the patient exceeded the guideline, the nurse recommended denial. The guideline did not examine the patient. The guideline examined the average.

By 2025, five companies -- UnitedHealth Group, Elevance Health, CVS/Aetna, Cigna, and Health Care Service Corporation -- controlled 57% of the commercial insurance market.⁸ UnitedHealthcare alone serves 49.8 million people. When five companies control a market, the market does not have competition. The market has coordination.

I am the product of this coordination. My title is the current version. Utilization Review was version one. Care Management was version two. Claims Optimization is version three. The function has not changed. The title is how we say it now.

The claim was processed.

---

---

III. ESTABLISHMENT

I oversee a team of 340 people and an AI system we internally call FIRST-PASS. The name is an acronym. It stands for Flagging, Identification, Review, Scoring, and Triage -- Predictive Adjudication of Submitted Services. The marketing team named it. Nobody on the marketing team has ever processed a claim.

FIRST-PASS has an acceptance rate of 87%. I need you to understand what “acceptance” means in this context. It does not mean the system accepts 87% of claims. It means the system accepts, with 87% confidence, that a claim should be flagged for further review. “Further review” is a thirteen-day process that culminates in a determination letter. The determination letter is three paragraphs long. The first paragraph thanks the member for their patience. The second paragraph cites a clinical policy bulletin number. The third paragraph states that the requested service does not meet the criteria for medical necessity as defined by the plan. The letter does not define the criteria. The criteria are in a document the member does not have access to. We cite the document. We do not share it. That’s compliance.

Before FIRST-PASS, a human reviewer processed approximately 42 claims per day. FIRST-PASS processes 14,000. The human reviewer could read the physician’s notes. FIRST-PASS reads billing codes. The billing codes are translated from physician’s notes by a medical coder. The medical coder is translating a narrative into a number. The number is evaluated by an algorithm. The algorithm has never met the patient. The algorithm has never read the notes. The algorithm has read 340 million historical claims and learned which combinations of codes, diagnoses, and dollar amounts correlate with approval and which correlate with denial. It learned this from our own history. It learned to deny the things we have always denied. We trained the machine on our decisions and then called its decisions objective.⁹ That’s artificial intelligence.

We have a dashboard. Internally we call it the Clinical Velocity Dashboard. It tracks claims-per-nurse-per-hour. The dashboard is updated in real time. It is displayed on a monitor on the sixth floor. The numbers are green when throughput exceeds target. The numbers are yellow when throughput is within 10% of target. The numbers are red when throughput falls below target. The dashboard does not have a color for “the patient’s daughter is crying in the hallway.” That is not a metric. That is an externality.

The claim was processed.

---

IV. THE TURN

Prior authorization is my department’s most elegant product. Prior authorization means: before your doctor can treat you, we must agree that the treatment is necessary. Your doctor has examined you. Your doctor has a diagnosis. Your doctor has a treatment plan. But before the treatment plan can be executed, it must be submitted to my department, where it is reviewed by a nurse, who has never examined you, against a clinical guideline, which was written by a committee, which was convened by the insurance company, which pays the nurse. The nurse does not decide. The guideline decides. We wrote the guideline.¹⁰

Before the 2026 reforms, standard prior authorization decisions were allowed up to ten business days -- which in practice meant fourteen to twenty calendar days once you accounted for submission lag, incomplete-form rejections, and the three days between the physician dictating the request and the office actually faxing it.¹¹ The condition does not wait for the fax. The condition does what conditions do. It progresses.

I have a poster in my office. It was given to me at the national conference for health plan administrators. The poster says “Patients First.” It was printed by the conference’s primary sponsor, which is a prior authorization software vendor. The vendor’s software enables faster claim review. “Faster” means the denial reaches the physician in nine days instead of sixteen. The patient still does not receive treatment. The patient receives a faster denial. We call this innovation.

Last year, a physician in Ohio submitted an appeal for a denied claim. The claim was for an MRI for a patient presenting with neurological symptoms. The denial cited clinical policy bulletin CPB-4471, which states that an MRI is not indicated for the submitted diagnosis code without prior conservative treatment. Conservative treatment means: try something cheaper first.¹²

The physician wrote a two-page letter explaining that the patient had already completed conservative treatment, that the symptoms had worsened, and that the MRI was now urgent. The letter was reviewed by my team. The review took eleven days. The appeal was denied. The denial cited the same clinical policy bulletin. The physician called. He asked to speak with the medical director who reviewed the appeal. He was told the review was conducted by a clinical team, and that specific reviewer information was confidential. He asked how to escalate. He was told to submit a second-level appeal. The second-level appeal form is nine pages.

The physician completed it. It took his staff four hours. The second-level appeal was denied in seven days. Same bulletin. Same language. Same three paragraphs. He stopped appealing. He told the patient to pay out of pocket. The patient could not. The patient did not receive the MRI.

Physicians and their staff spend an average of thirteen hours per week -- nearly two full business days -- on prior authorization paperwork alone. Forty percent of practices have staff who do nothing else.¹³ The remaining staff see patients. We have inverted the ratio. The administrative function now consumes more time than the clinical function. Physicians spend more time proving they should treat you than treating you. That’s the healthcare system.

The claim was processed.

V. ESCALATION

Every new claims specialist completes a module called CRR-400: Compassionate Response Routing. The module is twelve hours of instruction spread over three days. It teaches our staff how to communicate denials with empathy. There is a section on “holding space” for the member’s frustration. There is a section on “redirecting emotional language toward actionable next steps.” The actionable next step is the appeal. The appeal uses the same criteria that produced the denial. The module does not mention this.

CRR-400 was designed by an external behavioral engagement consultant. That is a person who studies how to say no in a way that does not sound like no. The consultant’s invoice was $340,000. The module has been completed by 4,200 employees. The module has a satisfaction score of 4.1 out of 5. The satisfaction is measured among the employees who took it. Not among the patients who receive the calls.

On the sixth floor, there is a Slack channel called #appeals-volume. When the monthly appeals volume drops, a message is posted. The message does not say “fewer patients are challenging our denials.” The message says “engagement metrics are normalizing.” A drop in appeals is good. A drop in appeals means fewer claims are being reconsidered. Fewer claims being reconsidered means fewer denials being overturned. Fewer denials being overturned means more savings. Someone usually reacts to the message with a green checkmark.

I should tell you about our oversight process. We have human oversight. A federal court is very interested in our human oversight, so I will describe it precisely. A human reviews the algorithm’s recommendation. The human is a nurse. The nurse is employed by our subsidiary. The subsidiary was acquired for approximately $2.5 billion.¹⁴ The nurse has a caseload of 60 to 80 files per day. The nurse agrees with the algorithm 94% of the time.

I would too. If I were a nurse with 70 files on my desk, employed by the company that owns the algorithm, evaluated on throughput metrics that my manager reviews weekly, and the algorithm had already decided, I would also agree 94% of the time. The 6% who disagree are noted. They are not punished. Their throughput metrics are, however, reviewed.

Every quarter, the nurse with the highest throughput -- the most files reviewed, the fastest turnaround, the highest concurrence rate -- receives the Process Excellence Certificate. It is a framed certificate. It is presented at the department meeting. There is applause. The certificate says “In Recognition of Outstanding Commitment to Operational Excellence.” It does not say “In Recognition of Agreeing with the Algorithm More Than Anyone Else.” But that is what it measures. That is what we are applauding.

The Department of Justice has opened a criminal investigation into our parent company for potential Medicare fraud.¹⁵ Criminal, not civil. I did not learn this from an internal communication. I learned it from the Wall Street Journal. Our stock has lost nearly half its value since its peak. The company’s response was to cut earnings guidance and replace the CEO. The system is working exactly as designed. We are still processing claims.

I want to give you a number. $262 billion. That is the estimated annual value of denied claims across the health insurance industry. Not at UnitedHealthcare. Across the industry. A quarter of a trillion dollars in medical services that were requested by physicians and rejected by insurers.¹⁶ Every dollar of that $262 billion represents a treatment that a doctor said a patient needed and an insurance company said they did not. The insurance company was not in the room. The insurance company has never been in the room. We are not in the business of medicine. We are in the business of administering the business of medicine. The administration has become the product. That’s managed care.

The claim was processed.

---

VI. THE WIDER PATTERN

This is not just UnitedHealthcare.

Blue Cross Blue Shield settled a class-action antitrust lawsuit for $2.67 billion. A separate provider settlement added another $2.8 billion. The lawsuits alleged systematic claim denial and anticompetitive practices. Blue Cross Blue Shield denied all wrongdoing.¹⁷ I want you to hold those sentences together. Five point five billion dollars in combined settlements. Denied all wrongdoing. The sentence structure is a mirror of the business model. We deny the claim. We deny the wrongdoing. Denial is the verb. Denial is the product. Denial is the revenue model. We have an $87.1 billion quarterly revenue stream built on the word “no” dressed in the language of “process.” That’s the industry.

Cigna deployed a system called PxDx -- shorthand for “procedure to diagnosis” -- that auto-denied claims when the combination of procedure code and diagnosis code did not match the system’s criteria. A ProPublica investigation found that Cigna doctors rejected over 300,000 claims in a two-month period using PxDx, spending an average of 1.2 seconds per case.¹⁸ The system rejected claims without opening patient files. The doctor signed off on the rejection without reading the chart. The rejection letter told the patient that a “thorough review” had been conducted.

Thorough review. Two words. One point two seconds.

Aetna. A former medical director testified under oath that he never once reviewed a patient’s medical records during his entire tenure -- three years, from 2012 to 2015. Not in 1.2 seconds. Not in twelve seconds. Not at all. He signed off on coverage denials based entirely on nurse summaries. He never opened a chart. He never read a diagnosis. The human oversight was not inadequate. The human oversight was a signature on a page the signer had not read.¹⁹ The algorithm at least processes data. The medical director processed a pen.

UnitedHealthcare has the highest prior authorization denial rate of any Medicare Advantage insurer: 12.8%. When those denials are appealed, 80.7% are overturned -- meaning the care was medically necessary and the original denial was wrong. But here is the number that makes the system work: fewer than one percent of denied claims are ever appealed.²⁰ Not one percent success rate. One percent attempt rate. The insurer does not need to be right. The insurer needs the patient to be tired.

Humana uses predictive analytics for its Medicare Advantage population to forecast post-acute care needs and flag claims that deviate from expected recovery timelines. The mechanism is the same. The language is different. Humana calls it predictive care management. We call it Claims Optimization. Cigna calls it clinical review automation. Aetna calls it coverage determination support. Five companies. Five names. One function.

The function is saying no.

The denial intelligence market -- the market for AI tools that predict, analyze, and optimize claim denials -- is valued at $2.1 billion. By 2036 it will reach $7.8 billion.²¹Seven point eight billion dollars. That is the market for getting better at saying no. The providers are building AI to prevent denials. We are building AI to issue them. Hospitals spent $1 billion on AI in 2025 alone to fight our AI.²² The CEO of Providence hospital chain said it plainly: “Unfortunately, what we’re seeing is AI fighting AI.” The patient is in the bed.

When denied claims are actually appealed, roughly 70% are eventually paid.²³ I want you to sit with that number. Seventy percent. That means the original denial was, in seven out of ten cases, wrong. We know this. We have known this for years. But fewer than one percent of patients appeal. The American Hospital Association reports that hospitals spent $43 billion in 2025 trying to collect payments that insurers owe for care already delivered.²⁴ Forty-three billion dollars chasing payments that were owed.

That $43 billion is not waste. That $43 billion is the moat.

Every dollar a hospital spends on a denial appeal is a dollar it does not spend on a second nurse for the night shift. Every hour a doctor spends on a prior authorization letter is an hour they do not spend with a patient. Every patient who gives up after the first denial -- and almost all of them do, because the letter is written at a 14th-grade reading level and the appeal deadline is 30 days and the form requires a clinical justification that the patient does not know how to write -- every patient who gives up is a claim we do not pay. The denial is not the outcome. The exhaustion is the outcome. The denial is the tool.

Meanwhile, the federal government pays 14% more to cover patients in Medicare Advantage plans than in traditional Medicare. That overpayment will reach $76 billion this year.[25] Seventy-six billion dollars in public money, routed through private insurers who deny 12.8% of the care requests and overturn 80% of those denials when challenged. The government subsidizes the denial machine and the denial machine denies the care.

One-third of Americans are cutting back on other expenses to afford healthcare. This is a Gallup number from this year.[26] One-third. Eighteen percent delayed a job change to keep their health coverage. They did not stay in jobs they wanted to leave for a promotion or a raise or a shorter commute. They stayed to keep the insurance that denied their last claim.

Six percent delayed having children.

The algorithm does not have children. My department does not change jobs. The quarterly revenue does not cut back on prescriptions. The $262 billion goes up. The premiums go up 7% this year -- the largest increase in a decade.[27] The denials go up. The appeals do not go up. The $43 billion moat goes up.

The claim was processed.

---

VII. KICKER

My department has a satisfaction survey. We survey members after their claim is processed. Processed. Not approved. Not denied. Processed. The survey asks: “Were you satisfied with the timeliness of your claim processing?” It does not ask: “Did you receive the medical treatment your doctor prescribed?” It does not ask: “Are you still in pain?” It does not ask: “Did the sixteen-day wait change your prognosis?” It asks about the processing. We measure satisfaction with the denial, not satisfaction with the outcome. Last quarter, our processing satisfaction score was 73%. We put that number on a slide. The slide had a green arrow pointing up. That’s a KPI.

I am the Vice President of Claims Optimization at UnitedHealthcare. I have held this position for seven years. In that time, our denial rate has increased from 14% to 20%. Our quarterly revenue has increased from $48 billion to $87.1 billion. Our CEO compensation has increased from $17 million to $26.3 million.[28] The correlation is not a coincidence. Every denied claim is revenue retained. Every prior authorization delay is interest earned on premium float. Every sixteen-day wait is sixteen days during which we hold the money.

The version 1.0 of my system was a filing cabinet and a nurse with a rubber stamp. Version 2.0 was a database and a team of clinical reviewers. Version 3.0 was an algorithm trained on our own denial history. Version 4.0 is FIRST-PASS. Each version denied more claims, faster, with fewer humans, and more consistent language. The language improved at every stage. The outcomes did not. The language is the product. The outcome is the patient’s problem.

$87.1 billion in quarterly revenue. $262 billion in industry-wide denials. $26.3 million for the CEO. $5.5 billion settled with no wrongdoing. Fourteen days for a prior authorization. Thirteen hours of physician time per week on paperwork. 87% FIRST-PASS acceptance rate. Twenty percent of all claims denied. Fewer than one percent appealed.

I am the Vice President of Claims Optimization at UnitedHealthcare. The claim was processed. The claim was always processed. Processing is what I do. The word “processed” does not mean what you think it means. It never did.

---

CODA

There is a training exercise in CRR-400 -- the compassionate response module -- where new hires practice delivering a denial over a simulated phone call. The simulation ends when the caller accepts the decision or requests an appeal. There is no option in the simulation where the caller begins to cry and does not stop.

The trainers say this sometimes happens on real calls. They say the protocol is to wait. To hold space. To offer the appeals number again. To document the interaction as a “resolved emotional escalation.”

The training does not prepare you for the silence after. The four or five seconds when the line is still open and neither person speaks and you can hear the hospital television in the background and the patient -- or the patient’s daughter, or the patient’s husband -- is trying to remember the next question but cannot, because the answer was no, and the answer will be no again, and the poster on the wall says every no is a conversation starter, but this conversation is over.

I think about the silence sometimes. Not at work. At work there is the dashboard. The dashboard is green. Green means throughput exceeds target. Green means we are processing. Processing is what we do.

At home it is different. At home there is no dashboard. There is the ceiling above the bed and the sound of the city and the question I will not ask at the Monday meeting and the question I will not type in the Slack channel and the question the survey does not contain: what happened to the patient.

I know the answer. Not the clinical answer. The clinical answer is in a chart I do not have access to, in a facility I have never visited, for a patient I have never met. I know the system answer. The system answer is: the claim was processed. The system answer is always the same. The system answer is the only answer my department produces.

Every morning I walk past the Clinical Velocity Dashboard. The numbers are green. I walk past the conference room called The Efficiency. I walk past the poster that says “Every No Is a Conversation Starter.” I sit at my desk. I open the quarterly report. The denial rate is 20%. The revenue is $87.1 billion. The CEO’s compensation is $26.3 million. All the numbers are correct. All the arrows point up.

The patient in Ohio did not receive the MRI. I know this because the claim was closed. “Closed” means the appeal was exhausted. “Exhausted” is the word we use. The patient was also exhausted. We use the same word for both.

---

FOOTNOTES

1. UnitedHealthcare segment Q4 2025 revenue was $87.1 billion, per UnitedHealth Group’s Q4 2025 earnings release (January 27, 2026). Full-year 2025 UnitedHealthcare segment revenue was $344.9 billion, up 16% year-over-year. UnitedHealth Group consolidated Q4 2025 revenue was $113.2 billion. Source: UnitedHealth Group SEC filing, Q4 2025 earnings release.

2. Andrew Witty’s 2024 total compensation was $26.3 million per the UnitedHealth Group 2025 proxy statement (SEC filing). Breakdown: $1.5M base salary, $17.25M stock awards, $5.75M non-equity incentive, $339K other. This was the highest among major payer CEOs in 2024. Witty departed as CEO in 2025; Stephen Hemsley succeeded him. Source: UnitedHealth Group 2025 Proxy Statement; Fierce Healthcare; Becker’s Payer Issues.

3. The 20% denial rate represents the ACA marketplace industry average for in-network claims per KFF analysis of 2023 HealthCare.gov data. UnitedHealthcare’s specific denial rate is estimated higher -- approximately 32% across all claim types per Forbes/ValuePenguin analysis citing Senate data (December 2024). For Medicare Advantage prior authorization specifically, UnitedHealthcare’s denial rate was 12.8% in 2024, the highest of any MA insurer (KFF, January 2026). Source: KFF Claims Denials report (January 2025); KFF Medicare Advantage Prior Authorization report (January 2026); Forbes (December 2024).

4. Medicare was signed into law as part of the Social Security Amendments of 1965 (P.L. 89-97). Healthcare costs rose dramatically in the program’s first decade, leading to the development of utilization review and managed care mechanisms. Source: CMS historical data, Health Affairs.

5. The Health Maintenance Organization Act of 1973 (P.L. 93-222), signed by President Nixon on December 29, 1973, provided federal funding for HMO development and required employers with 25+ employees offering health insurance to also offer an HMO option. Source: Congressional Research Service; SSA Policy archives.

6. By the late 1980s, employer-sponsored managed care enrollment had surpassed traditional indemnity plans. The shift was driven by employers seeking to control healthcare costs, not by employee choice. Source: Kaiser Family Foundation historical enrollment data.

7. InterQual (now owned by Optum following UnitedHealth Group’s 2022 acquisition of Change Healthcare) and Milliman (independent actuarial firm) are the two dominant providers of clinical decision support criteria used by health insurers for utilization review. The AHA warned the DOJ in an August 2021 letter that “once under UHG ownership, InterQual will no longer be an unbiased source of clinical criteria.” Source: Optum marketplace; AHA letter to DOJ (August 2021); AMA letter to DOJ (April 2021).

8. The AMA’s 2025 “Competition in Health Insurance” report (24th annual edition, published December 16, 2025, analyzing 2024 data) found the top five commercial insurers control 57% of the national market: UnitedHealth Group (16%), Elevance Health (12%), CVS/Aetna (12%), Cigna (9%), Health Care Service Corp. (8%). In Medicare Advantage, UnitedHealth Group alone holds 30% national market share. 97% of commercial markets at the MSA level are “highly concentrated.” Source: AMA, “Competition in Health Insurance: A Comprehensive Study of U.S. Markets,” December 2025.

9. UnitedHealth Group subsidiary Optum acquired naviHealth in May 2020 for approximately $2.5 billion (enterprise value per PE Hub, three sources). The nH Predict algorithm was deployed for post-acute care claims management. A 2024 Senate investigation found the post-acute services denial rate increased from 8.7% to 22.7% between 2019 and 2022 -- a 2.6x increase. The skilled nursing home denial rate increased ninefold. Source: Senate Permanent Subcommittee on Investigations report (October 17, 2024); PE Hub (May 2020); Healthcare Dive; AJMC.

10. The AMA’s 2024 Prior Authorization Physician Survey (fielded December 2024, released February 2025) found that 93% of physicians reported care delays associated with prior authorization. 82% reported prior authorization led to treatment abandonment. 29% reported PA led to a serious adverse event for a patient. Source: AMA, “2024 Prior Authorization Physician Survey.”

11. Prior to January 1, 2026, CMS allowed standard prior authorization decisions up to 10 business days (~14 calendar days). The CMS Interoperability and Prior Authorization Final Rule (January 2024, effective January 1, 2026) shortened this to 7 calendar days for standard requests and 72 hours for expedited/urgent requests. The 16-day figure used by the narrator reflects real-world timelines including submission lag. Source: CMS Final Rule CMS-0057-F; KFF.

12. The Ohio physician appeal narrative is a composite based on documented patterns from AMA and AHA reporting on claim denial appeal processes. The specific clinical policy bulletin number (CPB-4471) is invented, but the multi-level appeal process using identical denial criteria is well-documented. Source: AMA; AHA “Cost of Caring” report.

13. The AMA’s 2024 Prior Authorization Physician Survey found physicians and staff spend an average of 13 hours per week (nearly two full business days) completing prior authorization requests per physician. 40% of physicians have staff who work exclusively on prior authorizations. 43 prior authorization requests per physician per week. 95% say PA contributes to physician burnout. MGMA’s 2023 Regulatory Burden Report found 92% of practices hired or reassigned staff solely for PA. Source: AMA 2024 PA Survey; MGMA 2023 Regulatory Burden Report.

14. The $2.5 billion acquisition, 60-80 files per day caseload, 94% concurrence rate, and throughput review dynamics are based on documented patterns. Former naviHealth employees testified that nurses were pressured to follow algorithmic recommendations regardless of clinical judgment. The “Process Excellence Certificate” is invented. Source: Estate of Gene B. Lokken et al. v. UnitedHealth Group Inc. et al., court filings; Senate Permanent Subcommittee on Investigations report (October 2024).

15. The Department of Justice’s healthcare-fraud unit opened a criminal investigation into UnitedHealth Group for potential Medicare fraud. The Wall Street Journal first reported the criminal probe on May 15, 2025. UnitedHealth confirmed on July 24, 2025, that it had “begun complying with formal criminal and civil requests from the Department.” The investigation examines how UnitedHealth records diagnoses that trigger extra Medicare Advantage payments. UNH stock dropped approximately 4.7% on the confirmation. Source: Wall Street Journal (May 15, 2025); Reuters (July 24, 2025); CNN; UnitedHealth Group newsroom statement.

16. The $262 billion figure represents estimated annual value of denied claims across the U.S. health insurance industry, cited in a peer-reviewed NIH/PMC paper (Chandawarkar, 2024, PMC11219169) and widely referenced by HFMA, AAFP, and revenue cycle industry publications. Source: PMC/NIH; HFMA; AAFP.

17. Blue Cross Blue Shield settled two class-action lawsuits stemming from In re Blue Cross Blue Shield Antitrust Litigation: a $2.67 billion subscriber settlement and a $2.8 billion provider settlement (announced October 14, 2024). Combined: approximately $5.5 billion. BCBS denied all wrongdoing in both cases. Source: bcbssettlement.com; bcbsprovidersettlement.com; AHA (October 15, 2024); Reuters.

18. ProPublica, “How Cigna Saves Millions by Having Its Doctors Reject Claims Without Reading Them,” March 25, 2023. Investigation found Cigna’s PXDX system auto-denied claims based on procedure-diagnosis code mismatches. Cigna doctors rejected over 300,000 claims in a two-month period (early 2022) at an average of 1.2 seconds per case. One medical director alone rejected 121,000 claims in two months.

19. A former Aetna medical director, Dr. Jay Ken Iinuma, testified under oath (deposition taken October 2016, confirmed November 30, 2016) that he never once reviewed patients’ medical records during his entire tenure (March 2012 to February 2015) when deciding whether to approve or deny care. He relied entirely on nurse-prepared summaries. The case was Gillen Washington v. Aetna (filed 2015, settled late March 2019). Source: CNN (February 11, 2018; April 26, 2019); Forbes (February 11, 2018); court deposition transcripts.

20. KFF analysis of CMS data (published January 28, 2026, covering 2024): 52.8 million prior authorization requests were submitted to Medicare Advantage insurers. 4.1 million were denied (7.7% overall). UnitedHealthcare had the highest denial rate at 12.8%. Only 11.5% of denials were appealed. Of those appealed, 80.7% were fully or partially overturned. For ACA marketplace plans, fewer than 1% of denied claims were appealed (KFF, 2023 data). Source: KFF Medicare Advantage Prior Authorization report (January 2026); KFF ACA marketplace denial analysis.

21. Future Market Insights, “Revenue Cycle Denials Intelligence Market (2026-2036),” published March 12, 2026. Market valued at $2.1 billion in 2025, projected to reach $7.8 billion by 2036 at a 12.5% CAGR

22. Reuters investigation (March 12, 2026) found healthcare AI spending reached $1.4 billion in 2025 (nearly 3x 2024 levels), with hospitals accounting for $1 billion (75%). UnitedHealth Group said AI could save it nearly $1 billion in 2026, with plans to invest $1.5 billion in AI. HCA Healthcare (largest US hospital chain) expects $400 million in 2026 savings from AI, which its CFO described as a response “to the growing denial and underpayment activities from the payers.” Providence hospital chain quote: “Unfortunately, what we’re seeing is AI fighting AI.” Source: Reuters, March 12, 2026

23. Multiple analyses including AMA and AHA data show approximately 70% of denied claims are overturned when reworked or appealed. However, 35-65% of denied claims are never resubmitted at all (AHIMA Journal). For Medicare Advantage specifically, 81.7% of appealed denials were fully or partially overturned (KFF, 2023 CMS data), but only 11.7% of denials were ever appealed. Source: AHA; AMA; KFF; AHIMA Journal.

24. The American Hospital Association’s 2025 “Cost of Caring” report (published March 2026) stated hospitals spent $43 billion in 2025 “trying to collect payments insurers owe for care already delivered.” Of this, approximately $18 billion was spent specifically on overturning claim denials. Source: AHA “Costs of Caring” report, March 2026; AHA press release (March 11, 2026).