CMS announced second-year Rural Health Transformation Program awards on November 9, 2026. Every state again gets $100 million plus a variable share, and 25 states received more than in year 1 while 25 received less.
A stayed ACA notice, a new OPO rule and a look inside Claude
A federal court stayed 8 provisions of the 2027 ACA payment notice before it took effect, and CMS's final organ procurement rule rewrites what follows a ranking. For health care AI, Anthropic's J-space work is an audit tool first, a way to see what a model holds but does not say.
CMS expects 35 of the 55 donation service areas to open to competition at the end of the first cycle under its 2020 standards. The final rule keeps the yardstick and rewrites what follows a ranking: territory-level accountability, faster appeals, and nonrenewal for tier 2 losses.
States have passed at least 20 laws on AI in health coverage and care, while federal policy has moved through agency guidance and a push for preemption. This analysis maps both and sets out Redlake's position.
CMS announced second-year Rural Health Transformation Program awards on November 9, 2026. Every state again gets $100 million plus a variable share, and 25 states received more than in year 1 while 25 received less.
CMS's final Medicare drug price negotiation rule sets no tier or prior authorization rules for selected drugs. Plans that restrict them more than competing brands must give CMS a clinical justification at bid review.
CMS's 2027 physician fee schedule final rule spells out how AI may support each element of the Medicare annual wellness visit. AI can draft and collect, and the clinician must review, verify and approve.
CMS's CY 2027 home health final rule, published November 9, 2026, adds new reasons to deny or revoke Medicare enrollment and makes more revocations retroactive. It also expands the Medicare Advantage and Part D preclusion list.
AHRQ's 2026 evidence map rates AI ready for screening studies and unreliable for extracting their results. Health plans that use these tools to update coverage criteria should keep clinicians on the extraction step.
The FDA built a demonstration manufacturing line inside its own secure network. Its June 2025 white paper says the first job is seeing what every connected machine on the floor is doing.
By Redlake Research and Redlake Health Team · 6 min read
NCQA's final 2027 Health Plan Ratings method scores plans against national percentiles, counts outcomes triple and unreported measures as zero, and adds a bonus that lifts every accredited plan by half a star.
A Maryland court stayed 8 provisions on July 16, 2026, and a July 21 correction fixed 3 drafting errors. For 2027 ACA marketplace coverage, the 2024 notice's rules on standardized plans and income checks still apply.
Anthropic's Jacobian lens finds a small set of directions inside Claude that behave like a global workspace. For health care AI, that makes it an audit tool first, and it leaves the question of machine experience open.
By Redlake Research and Redlake Health Team · 18 min read
Most patients will never see how a clinical model reaches its answer, and many would accept that for better care. What they cannot be asked to accept is a decision nobody can explain, check or challenge.
CMS expects 35 of the 55 donation service areas to open to competition at the end of the first cycle under its 2020 standards. The final rule keeps the yardstick and rewrites what follows a ranking: territory-level accountability, faster appeals, and nonrenewal for tier 2 losses.
Medicare's WISeR model pays its prior authorization vendors a share of the spending they avert. Records from its first year show the cost of that design and point to a better one built on accuracy.
Federal audits of Medicaid autism therapy in 4 states flagged every sampled month of billing. We review how states and CMS responded in 2026 and argue that each claim should be built from a verified session record.
Coverage files record who decided but rarely which model helped. Naming the model version and its known limits in each file would let appeals and audits check the decision.
ONC dropped the AI model card rules and narrowed real world testing to APIs. It kept the privacy and security criteria and gave C-CDA exchange until 2030.
Test kits check that FHIR exchanges have the right structure. A shared pack of synthetic patients and expected answers would check that they reach the right result.
A 2026 law requires Medicare Advantage plans to verify directories every 90 days and publish accuracy scores. The facts they must verify start in clinicians' offices.
Payers now lower the level of office visit claims based on claim data alone. States have started to require a reviewer, and a fair process would start with one.
By Redlake Research and Redlake Health Team · 15 min read
Medicare Advantage pays more for every documented diagnosis, treated or not. Audits chase the codes years later. Paying only for diagnoses that show up in care would change the incentive.
Most coverage criteria turn on a few facts, yet requests arrive with whole charts. Signed data and zero-knowledge proofs could show a criterion is met without sending the record.
By Redlake Research and Redlake Health Team · 5 min read
ASTP has logged 2,450 possible information blocking claims since 2021, and 79% name a health care provider. The first enforcement notices went to software developers.
A denial is usually rebuilt from several systems when someone asks how it was made. The trail should be written as the decision happens and travel with it.
By Redlake Research and Redlake Health Team · 5 min read
Since 2023, CMS has added tool after tool to keep fraudulent hospices out of Medicare. The one party it rarely asks is the patient, who knows whether they chose hospice.
Every plan verifies the same licenses from scratch. A credential signed by the board that issued it could be checked in seconds by anyone who needs it.
States have passed at least 20 laws on AI in health coverage and care, while federal policy has moved through agency guidance and a push for preemption. This analysis maps both and sets out Redlake's position.
Federal rules require denial notices to state specific reasons in readable form. Writing them in the order of a clinical note would make those reasons useful to clinicians and patients alike.
State utilization review laws enacted in 2026 set rules for who may deny care and what software may do. The growing patchwork makes a common decision record more urgent.
By Redlake Research and Redlake Health Team · 15 min read
The 2025 budget law adds data checks on Medicaid enrollees and providers and a tougher eligibility error penalty. The measure behind the penalty does not count people wrongly cut off.
Hyperledger Fabric can keep PHI out of its blocks with channels and private data. What it writes instead, a hash of each private record, needs as much care as the data.
By Redlake Research and Redlake Health Team · 15 min read
Federal and state exclusion lists overlap only in part, and the CMS Preclusion List follows its own rules. A plan that checks 1 list sees only part of its network.
The 2024 federal rule excluded drugs from its prior authorization reforms. The same drug can face different clocks and records depending on which benefit pays.
Fraud analytics flag providers who bill unlike their peers. Unusual is not improper, and when fraud is rare most flags fall on honest providers. A score should open a review, never close one.
In the nH Predict and PXDX cases, courts have let core claims proceed on a simple theory. Insurers promised that clinicians would decide, and plaintiffs say algorithms did. Plans should be able to prove that review was real.
By Redlake Research and Redlake Health Team · 15 min read
NCQA's March 30 release reached 9 sets of accreditation and certification standards. Most entries correct or explain the text, but a few reach daily work, including license checks in every state of practice and UM criteria at the point of care.
The December 2025 executive order targets state AI laws, but an order cannot preempt state law, and insurance regulation has its own federal shield. The state rules most likely to survive are the ones requiring a clinician to decide denials.
In July 2026, House committees voted 42-0 and 45-0 to tighten prior authorization in Medicare Advantage, while the Senate voted to keep the WISeR model in traditional Medicare. The same standards should apply to both.
DOJ's 2026 takedown charged 455 defendants in schemes alleging more than $6.5 billion in false claims. Its largest case billed Medicare for years before payments stopped, and that lag is the number payers can change.
By January 2027, most payers CMS regulates must run 4 FHIR APIs. The guides for each are mature. The pieces that connect them, such as attribution and endpoint discovery, are still optional or unfinished.
Since 2026, decisions Medicare Advantage plans make during a stay are organization determinations with appeal rights. Plans should use that change to review the episode as a whole, so each transition does not restart the process.
By Redlake Research and Redlake Health Team · 5 min read
NCQA now accepts licensure compacts when the home-state license was primary source verified. That makes the home state the license to watch, because the privileges built on it can change when it does.
Insurers pledged to answer 80% of electronic prior authorization approvals in real time in 2027, but only for requests with all needed clinical documentation. That condition is a data contract, and plans should publish its terms.
Since January 2026, payers have had 72 hours or 7 days to decide prior authorization requests, counted from receipt. A shared ledger of hashed receipts would let both sides prove when a request arrived, without putting PHI on the ledger.
By Redlake Research and Redlake Health Team · 5 min read
HTI-4 turned the Da Vinci prior authorization guides into certification criteria for the systems providers use. Payer systems face no matching test, and CMS's April 2026 proposal would require the guides without testing the payers that run them.
Since 2024, Medicare Advantage plans have had to apply the two-midnight benchmark, which turns on what the admitting physician expected. The evidence points to partial compliance, and the shrinking inpatient-only list raises the stakes in 2026.
By Redlake Research and Redlake Health Team · 17 min read
Health plans test every utilization reviewer for consistency at least once a year and track whose denials are overturned. The AI tools that now shape those decisions deserve the same audit, run the same way.
In April, insurers reported that their June 2025 pledge had cut prior authorizations by 11%. The figure is self-reported and modeled, and it is the only number behind the 4 commitments already due.
NCQA's 2026 standards extend license monitoring to every state a clinician practices in, and 3 states now give plans 60 or 90 days to credential. A record checked once and filed away cannot keep up with either.
Medicare Advantage plans must publish their internal coverage criteria with the evidence and rationale behind them. That makes criteria visible. Versioned criteria with change logs and test cases would make them usable, and would let every decision say which rule decided it.
By Redlake Research and Redlake Health Team · 5 min read
Federal rules and an industry pledge require a new plan to honor existing approvals for a transition period. A plan cannot honor an approval it does not know about, which makes the payer-to-payer exchange due in 2027 a continuity of care requirement.
A prepayment edit or fraud model is a screening test, and medicine already knows how to judge one. When problems are rare, even an accurate model flags mostly honest claims. Payment integrity programs should measure what their flags actually find.
At least 9 states now prohibit health insurers from denying care based solely on AI, and 5 of them enacted their laws between March and mid-May 2026. The laws agree on the principle and leave the hard questions to implementation.
A physician practice holds about 20 health plan contracts, and each plan verifies the same license and malpractice history on its own schedule. The verification is identical. Only the decision differs, and only the decision needs to stay with each plan.
Ledgers and backups are built so that nothing is quietly deleted. Health data sometimes has to be destroyed anyway. Encrypting each record with its own key, and destroying the key when the record must go, reconciles the two.
By Redlake Research and Redlake Health Team · 5 min read
A 2024 Senate investigation found that post-acute denial rates at large Medicare Advantage insurers ran several times their overall rates. The rules now require traditional Medicare criteria and individual assessment, and a lawsuit over an algorithm has moved into discovery.
By Redlake Research and Redlake Health Team · 12 min read
FDA's January 2026 guidance expands enforcement discretion for decision support software, including tools that give a single recommendation. Its main safeguard is the clinician's ability to review the basis, which makes the design of review the real policy question.
Payers covered by CMS's prior authorization rule posted their first public metrics in March 2026. Approval rates in the reports we read ran from about 92% to 99%, but plans reported time and appeals in different ways.
CMS's interoperability framework delivered a Medicare app library and digital check-in in April 2026, 9 months after the White House event that launched it. Payer exchange is still at the starting line, and the harder FHIR requirements come next.
The Justice Department recovered a record $6.8 billion under the False Claims Act in fiscal 2025, $5.7 billion of it from health care, and has named Medicare Advantage a top enforcement priority. A $556 million Kaiser settlement in January shows what that means.
CMS's contract year 2027 Medicare Advantage final rule removes the health equity requirements for utilization management committees and drops the 2 appeals measures from star ratings starting in 2029. The obligations remain. The financial signal attached to them does not.
By Redlake Research and Redlake Health Team · 13 min read
Federal reviews have found inaccurate listings for nearly half of Medicare Advantage directory locations. A February 2026 law will require 90-day verification and public accuracy scores. The fix has to start where directory data comes from.
On March 31, payers covered by CMS's prior authorization rule must post their first public prior authorization metrics, aggregated across every service. Minnesota has shown a more useful format for years, with denials sorted by service and appeal outcomes.
California and Texas now require health care providers to tell patients when AI is involved. Disclosure respects patients, but it shifts the burden of judging AI output onto the person least able to judge it. Review before release and a record of how the output was made are the real safeguards.
TEFCA passed 474 million shared documents and 70,000 sites by February 2026. Its rules require answers to individual access and required treatment queries, but responding to payment queries is optional and operations queries only recommended.
Medicare Part B spending on skin substitutes rose from $256 million in 2019 to more than $10 billion in 2024. Fraud was real, but the larger story is a payment design that made overuse profitable, and the 2026 fixes that finally changed it.
Most peer-to-peer conversations happen after a plan has already denied a request, when they can only argue it back. Texas requires the chance before an adverse determination, and every plan should build its review that way.
By Redlake Research and Redlake Health Team · 4 min read
HIPAA requires audit controls, but most access logs sit in systems run by the same people whose access they record. Hash chains and a shared ledger can make any edit to a PHI access log detectable, without putting PHI on the ledger.
By Redlake Research and Redlake Health Team · 5 min read
Health plans recredential most clinicians every 3 years and bolt monthly checks on between cycles. The information that matters arrives continuously, and credentialing should be built to receive it that way.
At least 10 states now require some form of gold card exemption from prior authorization. In Texas, the first state with an automatic program, fewer than 4% of physicians qualified, and lawmakers rewrote the rules in 2025.
CMS estimates about $95.5 billion in improper payments across its programs for fiscal 2025, and says plainly that the figure is not a measure of fraud. Most waste is honest error and complexity, and it calls for design fixes rather than suspicion.
Medicare Advantage plans posted their first health equity analyses of prior authorization in 2025. Rates for enrollees with and without social risk factors look almost identical, in part because the required metrics pool every service into one number.
By Redlake Research and Redlake Health Team · 15 min read
In 2021, mail and fax were still the most common ways hospitals sent and received care records. The test of a standard is not how many systems support it but which manual channel it lets an organization shut off.
Medicare Advantage rules already say an approval for a course of treatment should last as long as it is medically necessary. Plans should make that the default everywhere and re-review only when something about the patient or the treatment changes.
Since May 2025, health programs that receive federal funds, including most health insurers, must identify decision support tools that use protected characteristics as inputs and reduce the risk that those tools discriminate. The rule names the duty and leaves the method to each organization.
Medicare Advantage plans denied 7.7% of nearly 53 million prior authorization requests in 2024, and 80.7% of appealed denials were overturned. Reviewer time belongs where judgment changes the outcome.
By Redlake Research and Redlake Health Team · 5 min read
Since July 2025, NCQA-accredited plans have had 120 days, not 180, between verifying a clinician's credentials and deciding on them, and must check exclusions and licenses at least every 30 days. The change is reshaping committee schedules and vendor contracts.
This month, clinicians in 6 states began needing prior authorization or prepayment review for 14 groups of services in traditional Medicare. The companies doing the reviews are paid from the spending they avert.
HIPAA does not forbid an immutable record, since amendments are appended rather than erased. The hard problems are replication and retention, which is why the strongest design keeps PHI off the chain and puts consent and access records on it.
By Redlake Research and Redlake Health Team · 16 min read
CMS plans to audit all of roughly 550 Medicare Advantage contracts each payment year instead of about 60. How much that recovers depends on extrapolation rules a federal court vacated in September, and on an appeal still in its early stages.
Since 2024, every Medicare Advantage plan has needed a utilization management committee that reviews its prior authorization policies against Traditional Medicare each year. After 3 rule cycles, the committee is where coverage criteria and the algorithms that apply them meet.
By Redlake Research and Redlake Health Team · 16 min read
Certified EHRs now have to exchange USCDI version 3, which adds insurance and functional status data that payers and providers both need. Federal enforcement waits until March 1 after the shutdown, and several demographic elements have been set aside.
Since January 1, Medicare Advantage plans and Medicaid programs must decide routine prior authorization requests within 7 calendar days and urgent ones within 72 hours. The limits leave room to stretch, which is why the first public reports matter.
By Redlake Health Team · 16 min read
No stories match these filters.
Get Rounds in your inbox
One issue, twice a month. Case studies, data notes, and product craft. No announcements you did not ask for.