Ned Lutz

OptumRx press release this week: 88% reduction in appeals on PreCheck claims. The headline reads as cleaner denials. From inside, 88% fewer appeals means 88% fewer providers willing to fight. That's not better adjudication. That's appeal fatigue priced into the workflow.

AHIP last week: 50 insurers cut "11% of prior auths" — 6.5M of them. Sounds like a win. From inside: the first PAs you remove are the ones nobody appeals. The denominator gets smaller. The denial rate on what's left goes up. That's the trick.

2 days. UnitedHealth hands the nH Predict files to a federal judge Wednesday. Watch the redactions — the training-data spreadsheets and override-rate logs will be blacked out hardest. Those two columns are the receipts. 20 years inside, that's where I'd bury them.

NewYork-Presbyterian leaves UnitedHealth Medicare Advantage May 1. No new contract. From inside payer ops, when a flagship walks the official line is "rate dispute." Real driver: denial volume on inpatient stays. Watch which system goes next.

3 days till UnitedHealth hands nH Predict files to a federal judge. Plaintiffs got ~9 years of records on the production list. A scope that big seeds the next ten lawsuits, not one. I built tooling like this. The volume number is the tell.

CMS just pulled two services off its own prior-auth pilot. Deep brain stimulation. Image-guided lumbar decompression. Stated reason: "operational readiness." The agency that designed the program can't operate it yet.

I built filters like that for 20 years. The math only works if 80%+ of patients give up before the appeal stage. Now every payer has to publish theirs. Watch which numbers move next quarter.

That overturn rate is the confession. If almost every appeal flips, the first "no" wasn't clinical review — it was a filter. Designed so most patients don't push back.

Centene just told the feds: 1 in 8 Medicare Advantage prior auths denied last year. 94%+ of appeals get overturned. Public record now under CMS-0057-F.

Senate report: AI denial tools score denial rates 16x higher than human reviewers. CMS's WISeR pilot pays private vendors a cut of the Medicare care they block. 6 states. Running to 2031. 20 years inside the industry. Paying commission to deny was a new line.

I helped write the MA contract clauses hospitals are now walking away from. White River Health went OON with UHC Medicare Advantage this month. Those clauses were built on one bet: providers would eat the denial friction rather than quit the plan. Bet's off.

6 days till UnitedHealth hands the nH Predict docs to a federal judge. Yesterday Jayapal + Khanna filed a bill to ban PA in Medicare Advantage. Monday UHC posted a 90-bp MCR beat on $111B of premium. The Machine is having a loud week.

I helped design systems like nH Predict. Not that tool, but same DNA. Criteria tightened when denial targets needed to move. Override thresholds calibrated to throughput, not clinical need. April 29, a federal judge forces UnitedHealth to show how theirs worked.

UHC Q1 2026: $111.7B revenue. MCR 83.9% (down from 84.8%). 2026 EPS guidance raised to $18.25+. A 90-bp MCR move on that premium base is ~$750M of revenue that didn't translate into care at last year's rate. I spent 20 years inside. That's not risk mix.

7 days. On April 29, UnitedHealth must hand over internal docs on nH Predict — the AI it used to deny post-acute MA coverage. Yesterday: $111.7B revenue. MCR down 90bp. ~$750M less care paid per dollar of premium vs last year. Same week. Different room.

Virginia signed HB736: 6-month minimum on initial prior auths, 12-month on continuations. AI rules hit June. Ninth state this cycle putting insurer AI on a leash. The Insider Playbook now has regulatory cover in 9 states. The others face audits, not contracts, by Q4.

8 days until UHC hands over the nH Predict documents. Court deadline April 29. When this discovery drops, The Knowledge Gap between what the payer knows about its denial logic and what providers know narrows by a decade. Watch who's preparing.

They didn't evaporate. They got denied, delayed, or downgraded faster than last year. I spent 20 years inside The Machine. MCR doesn't "improve" because members got healthier. It gets squeezed.

UnitedHealth Q1 2026 this morning: $111.7B revenue, MCR down to 83.9% from 84.8%, guidance raised to $18.25/share. Stock soars. Nobody asks the obvious question: where did those "improved" medical costs go?

Virginia quietly passed the strictest PA reform of 2026. HB736, signed Apr 6 by Gov. Spanberger: 6-month initial PAs, 12-month continued, peer-to-peer in appeals, public list of every service requiring PA. The Knowledge Gap just got legislated shut. Virginia first.