Primary sources: CMS press release, November 16, 2026, CMS fact sheet, the same day, and the proposed rule on public inspection, Federal Register document 2026-23461. The rule is the text. The release and the fact sheet are the summary.

CMS Counted Nearly $90 Billion in CRUSH. Table C-B8 Puts It on RADV Extrapolation.

On November 16, 2026, CMS proposed CMS-6098-P, the Comprehensive Regulations to Uncover Suspicious Healthcare rule, and said it could save taxpayers nearly $90 billion over the next 10 years. The public-inspection file is 2026-23461, 545 pages, filed at 4:15 p.m. ET, scheduled for the Federal Register on November 19. If you run revenue cycle, that savings line is a Medicare Advantage audit. Table C-B8 puts a RADV transfer of (90,166.10) million inside a total transfer of (90,228.76) million, and the mechanism is extrapolation of diagnosis codes the medical record does not support.

$90.2 billion

The regulatory impact analysis's estimate of RADV recoveries from 2027 through 2036 if extrapolation applies to payment year 2018 and later. Table C-B5 prints that transfer as $90,166.30 million. Table C-B8 prints the MA RADV line as (90,166.10) million. The fact sheet's rule-wide figure is nearly $90 billion in federal savings and approximately $378.51 million in costs. The press release says comments are due January 19, 2027.

What the November 16 documents propose

The press release is the announcement. The fact sheet is CMS's plain-language list. The 545-page proposed rule, RIN 0938-AV97, is what you would implement if it is finalized. It is a standalone, discretionary program-integrity rule. The fact sheet says it is not an annual payment rule, and that CMS intends to do program-integrity rulemaking of this kind every year. It was informed by a request for information published February 27, 2026, at 91 FR 9803. CMS says that RFI drew over 560 comments.

What the November 16 documents proposeWhat they do not do
CMS-6098-P would revise provider enrollment, medical review, investigations, and other program-integrity provisions across Medicare, Medicaid, CHIP, and the Marketplace. Comments go to regulations.gov docket CMS-2026-0826, file code CMS-6098-P. The press release says comments are due January 19, 2027. The fact sheet says the comment period is 60 calendar days after Federal Register publication.A filled-in calendar date in the rule. The DATES section still reads "[Insert date 60 days after date of publication in the Federal Register]." Publication is scheduled for November 19, 2026. If that date moves, the insert moves with it. This is a proposed rule.
CMS proposes to reissue extrapolation of RADV audit findings for payment year 2018 and subsequent payment years, using CMS-published payment-error methodologies, and to require MA organizations to remit improper payments CMS identifies. CMS proposes not to apply a fee-for-service adjuster. Sampling and extrapolation methodologies would be the ones CMS deems appropriate, including a sub-cohort such as enrollees with a particular HCC. A hardship exception for medical-record submission, today only in sub-regulatory guidance, would be written into the regulation. If CMS approves it, the organization is not required to submit medical records for the audited CMS-HCC, and that HCC does not negatively affect the error calculation.A single sampling formula locked in the regulatory text. Proposed § 422.311(a)(2) would read, "CMS may apply extrapolation to audits for payment year 2018 and subsequent payment years." The preamble says CMS intends extrapolation to be standard practice from PY 2018, and that CMS can decline to extrapolate in a particular audit. The reason for the reissue is in the rule: on September 25, 2025, the U.S. District Court for the Northern District of Texas vacated, for procedural reasons, similar provisions of the February 1, 2023 final rule. The footnote is Humana Inc. v. Becerra, 801 F.Supp.3d 624 (N.D. Tex. 2025), appeal docketed as Humana Inc. v. Kennedy, No. 25-11082 (5th Cir. Nov. 21, 2025). CMS says it is appealing. This article does not call that appeal.
MA organizations and Part D sponsors would have to report substantiated or suspicious fraud, waste, and abuse to CMS. Payment-suspension reporting, required today for pharmacies, would cover all provider and supplier types. For calendar year 2025, CMS says only 46 percent of Part C and Part D contracts voluntarily reported referrals, and only 19 percent reported any payment suspension based on a credible allegation of fraud, pharmacies and other provider types together.A dollar estimate for that reporting duty. The impact analysis says CMS cannot estimate the costs or the savings, because plan processes vary, and it asks for comment.
Proposed § 422.222(a) would require eligible providers and suppliers who bill an MA plan to be enrolled in Original Medicare in an approved status. PACE is excepted. The fact sheet says the requirement would take effect on the first day of the plan year beginning two years after publication of the final rule. CMS would then apply identity verification, licensure review, site visits, and fingerprint-based background checks before the provider bills the plan.A start date you can put on a calendar. There is no final rule yet. The fact sheet says CMS today relies on the preclusion list, which screens a provider after the provider has already begun billing MA plans.
Every Medicare enrollment change would have to be reported within 30 calendar days, for every provider and supplier type. Today, critical changes such as practice location, ownership, or an adverse legal action are 30 days, and other changes can run to 90. CMS would add a denial and revocation ground when a domestic or foreign owner, managing employee, organization, partner, officer, or director cannot be identity-confirmed, or when CMS decides that person poses an undue risk of fraud, waste, and abuse.A change to the 30-day clock that already covers location, ownership, and adverse actions. The new time pressure is the rest of the enrollment file.
Medicaid fee-for-service would get a regulatory 60-calendar-day report-and-return for self-identified overpayments. The fact sheet says that duty already exists for Medicare fee-for-service, Medicare Parts C and D, and Medicaid managed care. Medicaid managed care plans would report a beneficiary death or address change within 30 days and return overpayments tied to a death or a move out of state. States would collect and use NPIs for eligible Medicaid and CHIP providers at the individual level, and would send CMS a monthly file of denials and terminations with legal business name, address, TIN, and NPI, including cases not denied or terminated under a best-interests finding.A new 60-day rule for Medicare. That clock is already in place. A state may treat a CMS or other federal credible-allegation finding as its own, and it is not required to. Electronic visit verification data would be pulled into prepayment review and preliminary investigations for personal care and home health. Those are oversight duties on the state, not a new claim form.
On DMEPOS, CMS could decline a surety bond from a surety that has not met its obligations to CMS, and the ban on unsolicited beneficiary contact would extend past the telephone to texts, social-media direct messages, email without an opt-out, and door-to-door contact, with the same exceptions that already apply to calls. On the Marketplace, agents, brokers, and web-brokers with certain felony convictions, or who are debarred or excluded, could be kept out. A credible allegation of abusive conduct could suspend an agreement, with a 30-day chance to refute, and access from outside the United States, including through a VPN, would be prohibited in regulation.A priced revenue-cycle workflow. The impact analysis says it cannot say how much revenue the banned DMEPOS solicitation produces. The Marketplace provisions are about who may touch an Exchange enrollment, not about a hospital's claim edit.

"CRUSH would give CMS stronger tools to block fraudulent payments, remove dishonest providers, and protect Americans from schemes that exploit their healthcare coverage. This proposal could save taxpayers nearly $90 billion over 10 years and preserve resources for the care Americans depend on."

That quotation is the press release, both sentences, from HHS Secretary Robert F. Kennedy, Jr. The second sentence is the one Table C-B8 answers.

The same release says Medicare program-integrity savings rose 59 percent, from $26.3 billion in fiscal year 2024 to $41.9 billion in fiscal year 2025, and that the fiscal year 2025 return was $22.30 per dollar invested, up from $14.60, which the release calls a 53 percent increase and the highest return CMS has recorded. Those are results CMS says it already booked. They are not the nearly $90 billion in this proposal.

The $90.2 billion is the sample, held constant, then multiplied

Open the regulatory impact analysis, not the headline. CMS says the Part C improper-payment measurement, which is not a RADV audit and does not recover money, found over $23 billion in total Part C improper payments in fiscal year 2025, based on calendar year 2023 payments. Footnote 116 splits that into overpayments of $21 billion and underpayments of $2 billion, and the text says the overpayments were over 5 percent of total Part C payments. The source CMS cites is page 154 of the fiscal year 2025 HHS Agency Financial Report. The fact sheet then says the RADV approach in this rule would let CMS collect an average of approximately $9 billion per payment year in MA overpayments.

The rule's own arithmetic is more specific, and it is an assumption stacked on old audits. Contract audits for payment years 2011 through 2015 found approximately $37.6 million in improper payments, not extrapolated, across 29,900 sampled enrollees. CMS divides those two figures and gets an average overpayment of $1,258 per sampled enrollee. The rule says those improper payments resulted from human coding errors. CMS then assumes that $1,258 holds constant as the average overpayment per sampling-frame enrollee in each future RADV audit.

Hold that rate, and the rest of the $90.2 billion is volume. Over 2027 through 2036, CMS assumes 8,260 MA contract audits, about 826 concluded per calendar year, and approximately 71.7 million sampling-frame enrollees, about 7.2 million per year. On those assumptions, estimated overpayments with extrapolation are $90.2 billion. The prose calls the annualized transfer approximately $9.2 billion. That is not a flat check. Table C-B5 prints the RADV transfer, in millions, as $2,201.80 in 2027, $16,804.20 in 2028, $16,813.60 in 2029, $13,586.70 in 2030, and $6,793.30 in each year from 2031 through 2036. The printed ten-year total on that table is $90,166.30 million.

Table C-B8, which the extrapolation paragraph tells you to see, prints the same year path in parentheses and a ten-year MA RADV cell of (90,166.10). The total transfer cell on that table is (90,228.76). The two RADV totals differ by $0.20 million. That is the gap between the tables as printed. The next line on C-B8 that is even in the tens of millions is Medicaid MEQC deregulation, at (77.07). Managed-care oversight, the largest of the remaining rows, is 7.40 million over ten years. The nearly $90 billion in the press release is the RADV line.

Take extrapolation out and the RIA says recoveries over 2027–2036 would be approximately $89.7 billion lower. Table C-B6 prints estimated non-extrapolated collections, without the proposed RADV changes, at a ten-year total of $439.1 million. CMS says that alternative would leave the audit program with a negative return of approximately $3.2 million per calendar year, because the annualized program cost is approximately $48 million and the annualized non-extrapolated recoveries are $44.8 million. The ten-year cost of running the RADV program, support contracts, technology, and analytical tools, is approximately $478.7 million. The fact sheet's approximately $378.51 million, and Table C-B8's TOTAL COSTS line of 378.51, are the rule's quantified cost total. They are a different printed line from the $478.7 million. Table C-B5 also prints a RADV information-collection burden of $261.24 million and a hardship-exception burden of $2.09 million. That is paperwork. CMS says extrapolating does not increase what it costs an MA organization to comply with the audit.

What this is not

It is a proposed rule. The January 19, 2027 comment date is the press release. The DATES line in the public-inspection file is still an insert. It is not the physician fee schedule and it is not the outpatient payment rule. The $90.2 billion is a transfer from MA organizations to the federal government. The rule says none of that money is transferred to enrollees. The $26.3 billion and $41.9 billion are prior program-integrity savings in the press release, not this estimate. CMS's sentence that extrapolation does not raise the plan's audit-compliance cost is about the plan's cost of answering the sample. It is not a finding about what a provider group spends when the plan asks for the chart. The Northern District of Texas vacated the February 1, 2023 RADV final rule, and CMS says it is appealing. This article does not add a holding, a stay, or a prediction.

RevCycleAI analysis: the plan gets the audit, the group gets the chart

This is RevCycleAI's read for people who code, retrieve charts, credential providers, and clear credit balances. The government's quantified savings are a transfer priced at the MA contract. The diagnoses that produce the $1,258 were submitted to the plan, usually from a provider's record. A contract that can be extrapolated from a sample, including a sub-cohort built around one HCC, will not leave that sample inside the health-plan compliance department. It will show up as chart retrieval, documentation support, a retrospective coding audit, and the payment-clawback clause in a risk contract. CMS did not price that downstream work. The $478.7 million is CMS's cost to run the audits. The "no additional compliance cost" sentence is the plan's cost of the audit CMS already runs. The medical-record shop is outside both numbers.

The collection path the rule proposes is an offset. Extrapolated recoveries would typically come out of monthly capitation in the Medicare Advantage and Prescription Drug system, spread across months if one month cannot cover the amount, plus interest, with a wire in some cases. CMS also proposes that it might require plans to remit recoveries based on HHS-OIG audit findings. An offset against the plan's capitation is how a coding error becomes a contract conversation with the group that documented the HCC. RevCycleAI has already walked that seam in the Cotiviti deep dive and the Charta Series B piece: a code the chart does not support is a clawback, and on Medicare Advantage it is an audit problem. Those articles are earlier events. They are not a finding about this rule.

Three clocks in the same document are ordinary revenue-cycle operations, and none of them is the $90 billion.

The enrollment file moves to 30 days for every change. Location, ownership, and adverse actions are already on a 30-day clock. The proposal pulls the rest of the CMS-855 changes onto that same clock, for every provider and supplier type. Credentialing and enrollment teams that still park a non-critical update inside 90 days are the ones this hits. The new denial and revocation ground is the identity check on an owner, managing employee, organization, partner, officer, or director. If CMS cannot confirm who that person is, or decides the person is an undue fraud risk, the enrollment can be denied or revoked.

A provider who bills Medicare Advantage would need an approved Original Medicare enrollment. PACE is the exception the fact sheet names. The tools CMS lists are the ones Original Medicare already uses: identity verification, licensure review, a site visit, a fingerprint-based background check. The fact sheet is explicit that the preclusion list is the reactive substitute this would replace. The effective date is the first day of the plan year that begins two years after the final rule is published. There is no final rule, so there is no plan year yet. The work is knowing which MA-network providers are not enrolled in Original Medicare, before that date exists.

Medicaid fee-for-service joins the 60-day overpayment rule. Medicare fee-for-service, Parts C and D, and Medicaid managed care already report and return self-identified overpayments within 60 days. The fact sheet says Medicaid fee-for-service does not yet have that regulatory framework. Compliance and credit-balance workflows that treated the Medicaid fee-for-service account as the exception are the ones that change. Separately, a Medicaid managed care plan would have 30 days to report a death or an address change and to return the overpayment that follows. That is a recoupment file.

The litigation the rule itself describes is why this text exists. CMS is reissuing extrapolation, the decision not to apply an FFS adjuster, and the duty to remit, because the Northern District of Texas vacated the February 1, 2023 final rule for procedural reasons, and CMS says the appeal is pending. A later opinion can change what is collected. Nothing in the press release, the fact sheet, or the impact tables says the $90.2 billion has been recovered. Book it as CMS's transfer estimate under the assumptions above. The comment docket is open. The appeal the footnote cites is not decided in these pages.

The RCAI signal

The nearly $90 billion is the press release. Table C-B8's MA RADV line, (90,166.10) million, is the savings, and it is extrapolation from payment year 2018 forward. Without extrapolation the impact analysis says recoveries would be about $89.7 billion lower. The 30-day enrollment clock, Original Medicare enrollment for providers who bill MA, and the Medicaid fee-for-service 60-day overpayment rule are the workflows. Comments are due January 19, 2027. It is a proposed rule.

RCAI take

Treat the chart, the enrollment file, and the Medicaid credit balance as the work, and treat the $90 billion as CMS's transfer estimate. The open items are the comment docket, the blank date in the DATES section, and the appeal the rule cites and does not resolve. The MA enrollment requirement does not start until the first plan year beginning two years after a final rule that has not been issued. The $1,258 is an average from payment years 2011 through 2015, held constant and multiplied by 71.7 million sampling-frame enrollees. It is not a per-chart invoice, and it is not cash already pulled.

Sources: CMS, "Medicare, Medicaid, and Marketplace Programs; Comprehensive Regulations to Uncover Suspicious Healthcare (CRUSH)," proposed rule CMS-6098-P, public inspection document 2026-23461 (PDF opened, 545 pages). RIN 0938-AV97. Scheduled publication printed as 11/19/2026. The Federal Register API record 2026-23461.json matches the filing time as 2026-11-16T16:15:00-05:00, the page count as 545, and the publication date as 2026-11-19. The DATES insert, docket CMS-2026-0826, the proposal to reissue § 422.310(e) and § 422.311, the quoted sentence of proposed § 422.311(a)(2), the September 25, 2025 vacatur, footnote 110 and footnote 127 (Humana Inc. v. Becerra, 801 F.Supp.3d 624 (N.D. Tex. 2025), appeal docketed sub nom. Humana Inc. v. Kennedy, No. 25-11082 (5th Cir. Nov. 21, 2025)), the PY 2011–2015 audit facts ($37.6 million, 29,900 sampled enrollees, $1,258, human coding errors), the 8,260 contracts, 826 per year, 71.7 million and 7.2 million sampling-frame enrollees, the $90.2 billion, the approximately $9.2 billion annualized transfer, the $478.7 million and approximately $48 million RADV program cost, the $89.7 billion, the $44.8 million, the $3.2 million negative return, Tables C-B5, C-B6, and C-B8 as cited, the fiscal year 2025 Part C improper-payment figures and the citation to page 154 of the FY 2025 HHS Agency Financial Report, the MARx offset description, and the statement that extrapolation does not increase an MA organization's audit compliance cost are the rule's. Table C-B5 prints a RADV transfer total of $90,166.30 million and the year cells named above, plus RADV audit ICR of $261.24 million and hardship-exception ICR of $2.09 million. Table C-B6 prints non-extrapolated collections of $439.1 million. Table C-B8 prints MA RADV of (90,166.10), total transfers of (90,228.76), MEQC deregulation of (77.07), managed-care oversight of 7.40, and TOTAL COSTS of 378.51, all in millions. CMS press release, November 16, 2026 (page opened). The nearly $90 billion over 10 years, the January 19, 2027 comment date, the 59 percent increase from $26.3 billion to $41.9 billion, the $22.30 and $14.60 returns, the release's "53 percent" and "highest ROI ever," and Secretary Kennedy's two sentences are the release's. The quotation is verbatim. The release also quotes Administrator Mehmet Oz and Deputy Administrator Kim Brandt; those quotations are not used here. CMS fact sheet, November 16, 2026 (page opened). The standalone discretionary rule, the intent to do this rulemaking annually, the February 27, 2026 RFI at 91 FR 9803 and the "over 560 comments," the 30-day and 90-day enrollment distinction, proposed § 422.222(a), the PACE exception, the two-year effective timing, the screening tools, the preclusion-list description, the 46 percent and 19 percent contract figures, the approximately $9 billion per payment year, the approximately $378.51 million in costs from 2027 to 2036, the 60-day Medicaid fee-for-service proposal, the 30-day managed-care death and address reporting, the individual-level NPI duty, the monthly denial and termination report, the credible-allegation, EVV, DMEPOS, and Marketplace provisions as summarized above are the fact sheet's. No plan sponsor and no provider group is named as a target of this rule, and none is added here. The RevCycleAI analysis section is our read of those sources.