On September 30, 2026, CMS finalized the Global Benchmark for Efficient Drug Pricing (GLOBE) Model — a mandatory, five-year test that links Medicare Part B payment for certain high-spend physician-administered drugs to net prices in 19 developed countries, including Canada, France, Germany, Japan, and the United Kingdom. Earlier September issues of this newsletter tracked GLOBE as a pending proposal. It is now a rule, and it starts in 91 days.
GLOBE applies to single-source drugs and sole-source biologicals in oncology, rheumatology, immunology, ophthalmology, and endocrinology with more than $100 million in annual Original Medicare spending. Excluded: biosimilars, reference biologicals once a biosimilar enters the market, orphan-only drugs, plasma-derived products, certain cell and gene therapies, and drugs already under negotiated prices.
| GLOBE milestone | Date |
|---|---|
| Final rule | September 30, 2026 |
| Model start (delayed from October 2026 in the proposal) | January 1, 2027 |
| Lower beneficiary coinsurance begins | April 1, 2027 |
| Model performance period ends | March 31, 2032 |
| Rebate reconciliation ends | March 31, 2034 |
The mechanism is a manufacturer rebate: drugmakers pay rebates into the Medicare Supplementary Medical Insurance Trust Fund based on the gap between U.S. and international net-price benchmarks. On the provider side, CMS adjusts what Medicare pays — per the final rule, the allowed amount minus the GLOBE coinsurance amount, subject to other adjustments. In plain terms: in a GLOBE area, beginning April 1, 2027, the patient’s coinsurance on a GLOBE drug falls, and the claim must be adjudicated on that lower coinsurance figure.
The savings estimate collapsed between proposal and final. CMS now projects about $80 million a year in Part B savings and $440 million through 2034, against roughly $2.3 billion a year in the proposal. Total beneficiary savings are estimated at $111 million ($50 million in coinsurance, $61 million in premiums).
The smaller savings number does not shrink the billing work. A practice in a GLOBE area that infuses an in-scope oncology or rheumatology drug will see a new coinsurance calculation on its Medicare remits starting April 1, 2027, and its secondary-payer and patient statements must match. Practices outside the selected areas bill exactly as before — which is why the first task is simply to find out whether your ZIP codes are in.
MGMA’s September 28 federal outlook laid out what practices should plan against:
Two numbers hit on January 1: a proposed 1.68% conversion-factor cut and, in GLOBE areas, a new drug payment model. Neither is in your 2027 budget yet if you built it before September. Rebuild the Medicare line of the budget before the final fee schedule lands in November.
On September 28, CMS released the 2027 Medicare Advantage and Part D landscape and called it stable: weighted average MA premiums are projected to fall 16.5%, from $14.37 to $12.00, and 99% of beneficiaries will have at least one MA plan. Underneath the headline, the payer mix your front desk verifies in January is moving.
| Insurer | Counties exited for 2027 | Plans removed (approx.) |
|---|---|---|
| Centene | 344 | ~3,000 |
| CVS / Aetna | 103 | — |
| UnitedHealthcare | 63 | ~690 |
| Elevance | 56 | ~150 |
| Humana | — | ~2,400 |
Source: Healthcare Dive analysis of the CMS 2027 landscape, September 29. CMS projects about 34 million MA enrollees in 2027 — roughly 2 million fewer — with maximum out-of-pocket limits rising about 10% on average and Part D deductibles about 30%. That last pair matters to collections: higher out-of-pocket ceilings mean more patient responsibility on Medicare-age accounts next year.
At least 33 health systems have ended or are ending in-network MA contracts in 2026, several effective October 1 — two weeks before open enrollment opens October 15 (it runs through December 7). Examples include Ohio State Wexner and The James dropping Humana individual MA plans, NewYork-Presbyterian exiting most UnitedHealthcare MA plans absent a deal by September 30, and NKC Health, Valley Health, and Norman Regional leaving UnitedHealthcare. Systems cite prior authorization denials and slow payment.
A Health Affairs study published September 29 puts the brinkmanship in context: across nearly 15,000 hospital-insurer relationships from August 2021 through July 2025, only 8% of contract disputes went public, 28% of those ended in an actual network exit, and 2% of all relationships broke up.
For an independent practice, a hospital’s MA exit is a referral problem before it is a billing problem. If your surgeons, oncologists, or imaging partners are leaving a plan your patients are in, the downstream prior authorization and out-of-network exposure lands on your patients’ January claims.
MGMA’s September 1 poll asked whether prior authorization turnaround improved in 2026 versus 2025: 44% said slower, 40% about the same, 7% faster, 9% unsure (178 responses). This is despite federal requirements, effective January 1, 2026, that impacted payers respond within 72 hours for expedited and seven calendar days for standard non-drug requests. 61% of practices access seven or more payer portals every week. Medicare Advantage ranked as the most burdensome payer type. Practices that did get faster cited reduced authorization lists, electronic submission tools, and internal workflow redesign.
Before October 15, build a one-page list of the MA plans in your market that (a) are exiting your county, (b) have lost a major health system from network, or (c) are new for 2027. Give it to schedulers and patient-financial staff so patients asking “will you still take my plan?” get an accurate answer during open enrollment.
Two data points from the last ten days frame the AI question for practices this fall. First, payers are shifting where they find money — away from front-end denials and toward post-payment recovery, the part of the cycle where automated record review scales best. Second, practices are not waiting for vendors: a third have already built their own AI tools.
Revenue cycle leaders increasingly describe an AI “arms race”: payers ingest claims at scale, flag outlier patterns, request records, and then use AI to read those records against coverage rules and contract language. Kodiak Solutions’ first-half 2026 data (Section 05) shows the result — fewer initial denials, more clawbacks. A modifier error on a high-volume service can now trigger automated recoupment across an entire patient population, not a single denial on a single claim.
MGMA’s September 22 poll (228 responses) asked whether practices had used AI to build an app or workflow tool in-house: 34% yes, 8% tried but didn’t launch, 54% no. The tools respondents built were mostly operational: scheduling, prior authorization, dashboards, inventory, coding support, and patient intake. MGMA’s guidance before building: assess whether the tool touches patient data, map every data flow (including the AI provider), prototype on synthetic data, put ownership in writing, plan for record retention and auditability, and confirm your insurance covers homegrown tools.
A homegrown prior authorization tracker or coding helper is a business associate question, an audit-trail question, and a key-person question all at once. If the staff member who built it leaves, who maintains it? Write the owner, the data it touches, and the fallback process down before it goes into daily use.
At Becker’s IT + Revenue Cycle Conference, ISG’s healthcare lead reported health systems seeing 10–20% denial-rate reductions, 30–50% fewer manual touches, prior authorization turnaround cut from days to hours, and 20–40% coding productivity gains. Bain and KLAS research finds buyers generally want 3x or better ROI with a short time to value. R1’s new Phare UM (September 17) applies AI to flag medical-necessity denial risk while the patient is still in the hospital — part of the industry’s shift from back-end remediation to front- and mid-cycle prevention.
The October 1 changeover happened yesterday. This week’s coding job is no longer readiness — it is watching the first remits and catching problems before they compound.
| File | What changed | What to watch on remits |
|---|---|---|
| FY 2027 ICD-10-CM | 190 new, 30 deleted, 4 revised codes; largest gains in Chapters 19 (60), 15 (44) and 13 (31) | Rejections for deleted codes still in favorites lists or problem lists |
| NCCI PTP v32.3 (practitioner and hospital outpatient) | Quarterly additions, deletions, and modifier-indicator revisions | New bundling denials on code pairs that paid in September |
| NCCI MUE quarterly | Practitioner, outpatient hospital, and DME unit limits revised | Unit-limit denials on injectables and therapy codes |
| HCPCS Level II quarterly | October drug and supply code changes | Unlisted-drug denials where a new specific code now exists |
New ICD-10-CM specificity includes dedicated codes for hereditary cancer syndromes — Lynch syndrome, familial BRCA1/BRCA2, and Li-Fraumeni — which previously shared a single “genetic susceptibility” code. Oncology, GI, and genetic counseling practices should confirm these codes are mapped in their EHR problem lists, not just the billing system.
The October 1 MLN Connects newsletter reminds practitioners that MACs deny duplicate lab tests that are not medically necessary — its example is a second vitamin D test — and that modifier 91 must be used when a repeat clinical diagnostic lab test on the same day is medically necessary. Modifier 91 may not be used when a test is repeated because of equipment or specimen problems, or to confirm an initial result. CMS also advises against ordering individual tests that duplicate components of a panel already ordered. Separately, MLN flagged a January 2027 update to the lab NCD edit software adding and deleting ICD-10-CM codes — a reason to re-check diagnosis pairing on your highest-volume NCD-governed labs (lipid panels, A1c, TSH, PSA) in December.
Pull every lab claim with modifier 91 from the last 90 days and every same-day duplicate CPT without it. Duplicates without 91 are denials waiting to happen; 91s attached to a re-run for a bad specimen are an audit finding.
CPT 2027 — including the new AI service codes and the global maternity restructure covered in September — takes effect January 1, 2027, 91 days from today. The proposed fee schedule values the G2211 complexity add-on at roughly $16 per use. Any practice that saw a spike in deleted-code or NCCI rejections this week should use that list as the test script for its January cutover.
Kodiak Solutions’ first-half 2026 benchmark, drawn from revenue cycle data on 2,400 hospitals and 375,000 physicians and published September 23, contains the most counterintuitive number of the quarter: denials got better, and cash got worse.
| Metric (H1 2026 vs. H1 2025) | 2025 | 2026 | Direction |
|---|---|---|---|
| Initial denial rate | 12.00% | 10.63% | Better |
| Final denial write-offs (% of net revenue) | 3.31% | 3.07% | Better |
| Payer clawbacks (% of revenue) | 1.38% | 1.57% | Worse |
| 60-day cash performance | 100.28% | 96.74% | Worse |
| Private payer initial denial rate | 21.44% | 19.98% | Better |
| MA initial inpatient denial rate | 10.75% | 9.15% | Better |
| Final MA denial rate vs. Traditional Medicare | — | 3.76% vs. 2.22% | MA still 1.7x |
The takeaway: a denial dashboard alone no longer measures payer friction. Money that used to be withheld at the front door is now being paid and then recouped — which shows up as negative adjustments, offsets on later remits, and A/R that looks clean but nets less.
Add a “net collected after recoupment” line to your monthly KPI pack. Track recoupments by payer, by CPT, and by reason code, and age them like denials. If your initial denial rate is falling and your collection rate is flat or falling, clawbacks are the first place to look.
The Commonwealth Fund’s survey released September 17–18 found one in three privately insured working-age adults carrying medical debt. About half of them owe $2,000 or more; 43% attribute the debt to doctor’s office visits; 37% drained savings; 30% delayed needed care. Pair that with MA out-of-pocket maximums rising about 10% for 2027 (Section 02), and patient balances are getting larger while patients’ capacity to pay shrinks.
| KPI | Warning zone | Target |
|---|---|---|
| Initial denial rate | > 10% | < 5% |
| Recoupments as % of payments | Not tracked | Tracked monthly, < 1% |
| Net collection rate | < 95% | ≥ 96% |
| Days in A/R | > 45 | < 35 |
| Point-of-service collection on known balances | < 50% | ≥ 70% |
| Payment plans offered before statement 2 | Never | Every balance > $500 |
Bad provider data is a revenue cycle cost hiding in plain sight: wrong addresses on payer directories send patients to the wrong place, mismatched NPIs and taxonomy codes trigger credentialing and enrollment denials, and outdated network status produces out-of-network surprises. CMS is now building national infrastructure to fix it.
On September 24, H1 Insights won a five-year, $77,635,070 contract to provide IT development for the CMS National Provider Directory (NPD), beating 12 other bidders. On September 29, CMS also selected H1 for ground-truth data sourcing and licensing, plus data aggregation, quality, and assurance services — sourcing authoritative data, normalizing and validating it, and enriching and correcting provider records.
In parallel, CMS is moving Medicare Advantage provider directories onto Medicare Plan Finder. Plans had an executive attestation deadline of September 1, 2026 for contract year 2027, production deployment begins in October 2026, and the program transitions to FHIR-based submissions in 2027. Plans must update directories within 30 days of a change, each record carries a “lastUpdatedOn” timestamp CMS uses to verify compliance, and a CEO, CFO, or COO must certify accuracy. CMS may suppress a plan’s directory from Plan Finder if validation thresholds are not met. CMS validates format and refresh — not whether a provider actually practices at a listed location.
Plans are on the hook, but they get their data from you. Every location, NPI, accepting-new-patients flag, and departed provider you fail to report becomes a wrong listing on a federal website patients use during open enrollment — which starts October 15.
On September 25, the American College of Emergency Physicians, American College of Radiology, and American Society of Anesthesiologists sent Congress an analysis of the No Surprises Act independent dispute resolution (IDR) process arguing that payer behavior — not provider abuse — drives most provider wins.
| Finding | Figure |
|---|---|
| Line items insurers lost by default (2025) | ~25% |
| Line items with payer offers at or below the QPA (excluding defaults) | Nearly 40% |
| Line items with payer offers of $1 or less | 8.2% |
| Provider win rate in finalized determinations | 85% |
| Typical award vs. QPA | 2.5x–4x |
Critics point to a Georgetown estimate of $22 billion in additional costs from the IDR system over four years. The societies’ reply is that payers who skip participation or submit nominal offers are choosing to lose.
If your practice bills out-of-network emergency, anesthesia, radiology, or other NSA-protected services, register on the IDR Gateway now, track every award’s 30-day payment deadline, and log unpaid awards by payer. The documentation is your leverage in any complaint or future enforcement action.
Medicare Plan Finder’s MA directory publication (Section 06) makes directory accuracy a compliance item for practices, too: an inaccurate listing that sends a patient to an out-of-network location can become a surprise-billing dispute.
Black Book Research surveyed attendees at the MGMA Annual Conference in San Antonio (September 27–30) — a quarter of them CFOs, finance directors, and revenue cycle leaders, and a fifth physician owners and medical directors. The results, released September 29, describe practices heading into 2027 under pressure from every direction at once.
MGMA marked ten years of its weekly Stat poll on September 29 — more than 380,000 responses. The trend lines independent practices will recognize: the share saying prior authorization requirements increased ran 82% in 2016 and 90% by 2019 and has stayed high; 84% in 2026 say costs are outpacing revenue (after 92% in 2024); most groups have gone years without a commercial payer rate increase; and MGMA’s federal outlook found 90% of practices report patients shifting to MA and 79% report negative effects from it.
| Metric | Where most independents are | Where you should be |
|---|---|---|
| Initial denial rate | 10–14% | < 5% |
| Recoupments tracked by payer and reason | Not tracked | Monthly |
| Days in A/R | 42–50 | < 35 |
| Payer portals worked weekly | 7+ (61% of practices) | Consolidated through a clearinghouse or RPA |
| Commercial contracts reviewed for rate escalators | Years since last increase | Every contract reviewed annually |
| 2027 budget reflects the proposed CF cut | Not modeled | Modeled at −1.68% |
| In-house AI tools with a named owner and fallback | Undocumented | Written owner, data map, fallback |
| MA plan list for open enrollment | None | Distributed to front desk by October 15 |
(1) Model the 1.68% conversion-factor cut against 2027 Medicare revenue; (2) add a recoupment line to the monthly KPI pack; (3) send a rate-increase request to every commercial payer whose contract has not moved in two years; (4) document every in-house AI tool’s owner and data flow; (5) publish an MA plan cheat sheet for open enrollment.
This week’s specialty picture is driven by the GLOBE final rule, the October 1 code sets, and the proposed 2027 fee schedule — landing unevenly across service lines.
| Specialty | This week’s update | Data point | Priority action |
|---|---|---|---|
| Primary Care | Proposed 2027 CF cut; G2211 add-on valued around $16; new-patient waits longer at 28% of groups; Medicaid work requirements start by January 1 | −1.68% proposed CF | Model 2027 Medicare revenue; prepare front desk for Medicaid eligibility churn in January |
| Cardiology | Journal of Cardiovascular CT study (September 30): inflation-adjusted Medicare pay for cardiac CT fell 27–68% from 2010 to 2026 while CCTA volume rose 408% (2013–2024) | CCTA −68.3% real | Re-check CCTA cost per study against payment before 2027 volume planning |
| Orthopedics | Year two of three of CMS’s inpatient-only list phase-out; EBRI site-neutral analysis renews ASC vs. HOPD debate | 637 procedures leaving IPO list | Confirm site-of-service and prior auth rules per payer for newly eligible outpatient cases |
| Oncology | GLOBE finalized: in-scope Part B oncology drugs in selected areas get international benchmark rebates; coinsurance falls April 1, 2027 | 25% of Original Medicare in GLOBE areas | Confirm whether your ZIP codes are in GLOBE; flag in-scope drugs in the charge master |
| Rheumatology / Immunology | High-spend biologics without biosimilar competition fall within GLOBE scope | >$100M Medicare spend threshold | List infused biologics that cross the threshold; plan for new coinsurance math |
| Ophthalmology | Single-source intravitreal drugs above the spend threshold fall within GLOBE scope | 5 therapeutic classes in GLOBE | Update patient estimates for April 1, 2027 coinsurance changes |
| Radiology | ACR co-signed the IDR analysis; 85% provider win rate in finalized determinations | ~25% of line items lost by payer default | Register on the IDR Gateway; track 30-day award payment deadlines |
| Mental / Behavioral Health | Medicare telehealth flexibilities run through December 31, 2027; psychotherapy codes are shielded from the 2.5% efficiency adjustment | 0% efficiency cut on psychotherapy | Keep POS 10 and audio-only modifiers clean; plan 2028 in-person capacity |
| Gastroenterology | New ICD-10-CM codes for Lynch syndrome now in effect; endoscopy codes remain under the 2.5% efficiency adjustment | Dedicated Lynch syndrome code live October 1 | Update surveillance colonoscopy diagnosis mapping for hereditary risk |
| Pathology / Lab | MLN reminder: duplicate tests denied unless medically necessary; modifier 91 for valid same-day repeats | Lab NCD edits update January 2027 | Audit modifier 91 use; check panel/component duplication |
GLOBE is the first mandatory payment model to touch infusion revenue directly, and it lands on the same practices — community oncology, rheumatology, retina — that already carry the largest buy-and-bill exposure. The billing change is a coinsurance change, but the cash risk is patient collections: a lower coinsurance figure on the Medicare remit that does not match the estimate, the secondary claim, or the statement creates rework on every infusion visit.
Ten moves for the week of October 2, 2026 — each tied directly to something in this issue.