Medicare AI Prior Authorization Cut Targeted Services 20%: What WISeR Means for Your Practice

Medicare AI prior authorization is now real in traditional Medicare, and the first numbers show it cuts care fast. A JAMA research letter published October 26, 2026, found use of targeted services fell about 20% in the 6 states where CMS runs its WISeR pilot, compared with the rest of the country. Epidural steroid injections fell 15%.

Then the records landed. The Electronic Frontier Foundation published 800-plus pages of CMS documents on October 27, 2026, and they show the rollout was rough. CMS sent warnings to 4 of the 6 vendors in the first 3 months. One vendor denied 53% of the requests it decided in Washington. One request sat unanswered for 83 days.

If your practice bills traditional Medicare in Arizona, New Jersey, Ohio, Oklahoma, Texas, or Washington, this pilot already touches your claims. If you practice anywhere else, it still matters, because CMS can expand it to more states and more services.

Here are the JAMA numbers, the FOIA records, and the math on what a denied request costs, plus the exact documentation that gets requests affirmed the first time.

Table of contents

• What the JAMA study found
• The FOIA records: 4 of 6 vendors warned in 3 months
• Vendors get paid when care does not happen
• What WISeR covers and who it hits
• What a non-affirmation costs your practice
• WISeR vs Medicare Advantage vs standard Medicare
• How to get a WISeR request affirmed the first time
• What happens next: expansion, Congress, and January 1, 2027
• Your WISeR checklist
• What this means for your revenue
• Frequently asked questions

Key takeaways

• Medicare AI prior authorization under WISeR cut targeted services about 20% in 6 states in the first 6 months, per a JAMA research letter published October 26, 2026. Epidural steroid injections fell 15%.
• CMS sent warnings to 4 of the 6 WISeR vendors in the first 3 months, per 800-plus pages of records the Electronic Frontier Foundation published October 27, 2026.
• In Washington, vendor Virtix Health did not affirm 53% of the 6,096 requests it decided in the first quarter. In Arizona, Zyter did not affirm 19% of 14,301 requests.
• WISeR vendors earn up to 25% of the spending their reviews avert, per records reported by KUOW. A federal memo in those records admits the setup can reward denials.
• A practice that files 20 WISeR requests a month loses about $18,398 a year to rework and procedures that never get approved, on our worked example using Washington's 53% first-pass rate.
• The fix is documentation that proves conservative care failed, with dates, imaging, and the exact codes attached before you submit.

What the JAMA study found

WISeR stands for Wasteful and Inappropriate Service Reduction. CMS launched it January 1, 2026, in 6 states: Arizona, New Jersey, Ohio, Oklahoma, Texas, and Washington. It runs through December 31, 2031. For the first time, traditional Medicare requires prior authorization for a list of services, and private vendors use AI to screen the requests.

Researchers Hannah Neprash of the University of Minnesota and Michael Barnett of Brown University published the first independent study of the pilot on October 26, 2026, in JAMA. The University of Minnesota announced the findings the same day. They used Epic Cosmos, an electronic health record database covering more than 300 million patient records. They compared monthly use of WISeR services in the 6 pilot states with use everywhere else, over the 2 years before launch and the first 6 months after. Becker's Hospital Review covered the study on October 30, 2026, and Medical Daily ran its own summary on October 28, 2026.

Epidural steroid injections, the highest-volume service in the pilot, fell 15% in the WISeR states relative to the rest of the country. Other targeted services fell about 20%. The drop was sharpest right after launch. Epidural injections fell about 21% in the first quarter of 2026, then eased to about 9% in the second quarter. Other services eased from about 21% to about 18%.

Services that CMS had planned for WISeR but never added showed no drop at all. The researchers say that points to prior authorization itself as the cause, not some wider trend.

The study leaves the biggest question open. It measures service use, not diagnoses, so it cannot say whether the care that did not happen was wasteful or needed. Neprash said the key unanswered question is whether the care patients did not receive was wasteful or necessary. The paper also does not measure pain, function, complications, or patient satisfaction. It does not measure total spending, so it cannot say whether the pilot saved money after substitute care is counted.

That gap is the whole argument in one number. A 20% drop looks like savings if the care was wasteful. It looks like denied care if it was needed. Nobody has published the data that settles it yet.

Denials were already the top pressure on practices before WISeR. Our claim denials rising in 2026 post covers the Experian survey that found 75% of providers say denials are increasing. WISeR adds a new denial front inside traditional Medicare, where prior auth barely existed before.

The FOIA records: 4 of 6 vendors warned in 3 months

The JAMA study shows what happened to service use. The FOIA records show what the launch felt like inside.

The Electronic Frontier Foundation sued CMS in March 2026 for WISeR records. It published a first batch in September 2026, then 800-plus more pages on October 27, 2026. The new batch holds requests for information and corrective action plans CMS sent to at least 4 of the 6 vendors, plus vendor responses, applications from 66 companies, and program contracts. MedCity News covered the release on November 2, 2026, and HealthExec covered it on October 29, 2026.

CMS flagged Virtix Health, the Washington vendor, in a March 17 letter. The letter cited slow turnaround, denials that looked inaccurate or inconsistent with coverage criteria, and weak customer service. Virtix rejected 53% of the more than 6,000 prior authorization requests it received over 3 months. The company blamed an influx of claims and provider confusion, including faxes sent to the wrong agency.

Zyter, the Arizona vendor, got a February 27 letter. CMS said determination letters went out late and the company may lack adequate processes for provider concerns. Zyter denied 19% of its 14,301 requests in the first quarter.

Put the two vendors together and the first-quarter picture is stark. The two vendors issued 5,944 non-affirmations on 20,397 decisions, or 29%. Washington's rate ran 2.8 times Arizona's. Same program, same rules, same quarter. The vendor deciding your request changed your odds by 34 points.

Innovaccer, another vendor, got 2 corrective action plan requests and a Performance Notice of Non-Compliance tied to its failure to run a working provider portal. CMS kept the January 1 launch date even after Innovaccer warned its portal might not be ready.

One request sent to the vendor Genzeon sat unanswered for 83 days, per the records. CMS expects answers in 72 hours. An 83-day wait is 27 times that limit.

CMS did act on the worst case. It sent Virtix a corrective action plan on June 23, 2026, for missing turnaround times. Virtix says the plan closed August 14, 2026, and reports average turnaround of 1.18 days for prior authorization after the fix. Those are Virtix's own reported figures, and they are averages, so slow cases still hide inside them.

The Washington State Hospital Association surveyed hospitals in April 2026 and found procedures that used to take about 2 weeks to approve were taking 4 to 8 weeks. A request that waits 4 to 8 weeks pushes the procedure, the claim, and the payment out by a month or more. Your days in AR stretch with every week a request sits in a vendor queue.

Vendors get paid when care does not happen

The pay structure is the part of WISeR that draws the most heat, and the records explain why.

WISeR vendors earn a share of the spending their reviews avert, meaning the value of care Medicare did not pay for because the vendor did not affirm it. Records reported by KUOW put that share at 25%, reducible by up to 10% for poor performance on timeliness and accuracy. daisyBill covered the pay setup on November 2, 2026, and TechTarget covered the backlash to the pilot on November 3, 2026. A federal memo in those records admits the structure can create an incentive to deny.

CMS says the pay is adjusted for accurate performance, and the model requires a licensed clinician to review every denial recommendation before it goes out. AI can screen a request and approve it, but only a human clinician can turn one down. The EFF report says the records give little proof that human reviewers have real power to overrule the AI recommendation, or enough time to do it.

Here is the tension in plain terms. A vendor that affirms every request earns nothing from the pilot. A vendor that denies requests earns up to 25% of what those denials save. CMS built in a performance adjustment and a clinician review rule to check that pull. The first-quarter numbers show how hard that check has to work: 53% non-affirmed in Washington against 19% in Arizona, under the same coverage rules.

Congress has noticed. The House Appropriations Committee voted unanimously in June 2026 to defund the model. A Senate vote to end it failed 46 to 50 in July 2026. Senators Patty Murray and Ron Wyden introduced a bill in December 2025 to prohibit WISeR outright, and renewed their push at an October 9, 2026, roundtable in Vancouver, Washington, where doctors described months-long appeals and patients waiting in pain.

What WISeR covers and who it hits

WISeR applies to traditional Medicare only. Medicare Advantage plans already run their own prior auth programs, and WISeR does not touch them.

About 6.4 million traditional Medicare beneficiaries live in the 6 pilot states, per KFF estimates reported by AARP in February 2026. About 207,500 of them received a WISeR service in 2024. Oklahoma and Texas hold a combined 2.7 million of those beneficiaries, Arizona and Washington hold 1.5 million combined, and Ohio and New Jersey hold 1.1 million each.

The pilot covers 13 services in its first year. The list includes skin substitutes for chronic wounds, epidural steroid injections for pain, cervical fusion, electrical nerve stimulator implants, knee arthroscopy for osteoarthritis, deep brain stimulation, and incontinence-control devices. Two services CMS first planned were dropped before launch and will be rechecked later. Emergency services, inpatient-only services, and services that would pose a big risk to patients if delayed are excluded.

Skin substitutes drive the money. KFF found skin substitutes accounted for $10.3 billion, or 83.4%, of 2024 spending on the targeted services, a 2,000% jump from 2019, when they were 21% of nearly $510 million. Spending on all the other targeted services combined rose just 6.6% over the same 5 years.

That is why CMS picked this list. One product line grew 20-fold in 5 years and now eats more than 4 of every 5 dollars in the pilot's scope. A practice that bills skin substitutes in a WISeR state now faces prior auth on its highest-dollar service line.

The 6 vendors split the states. Cohere Health covers Texas, Zyter covers Arizona, Virtix Health covers Washington, and Genzeon, Humata Health, and Innovaccer cover the rest. OmniMD's WISeR guide lists the vendor for each state and the codes the pilot covers. Your vendor depends on your state, and the first-quarter numbers show the vendor matters as much as the rule.

For scale, compare WISeR with Medicare Advantage. KFF found Medicare Advantage plans denied 7.7%, or 4.1 million, of 53 million prior authorization requests in 2024. WISeR's Washington vendor denied 53% in its first quarter. That is about 7 times the Medicare Advantage denial rate, on a program that was supposed to bring the same discipline to traditional Medicare.

Prior auth pressure already runs high across the board. The AMA's 2025 survey found 32% of physicians say prior auth requests are often or always denied, and 74% say denials rose over the past 5 years. Physicians complete an average of 40 prior auths per physician per week, taking 13 hours of physician and staff time. KFF found nearly 70% of insured adults call prior auth a burden, and 34% call it their single biggest burden, beyond costs.

What a non-affirmation costs your practice

Percentages hide the dollars, so here is the math on a round practice, with every input tied to a published number or labeled as an illustration.

Take a pain practice in Texas that files 20 WISeR requests a month. Use Washington's 53% first-pass non-affirmation rate as the worked rate, since it is the only vendor rate published in full. Use the MGMA rework cost of $25.20 per denied claim, the same figure from our denials post. Assume $400 average allowed per WISeR procedure, labeled here as an illustration number, not a published rate. Your allowed amount will differ.

20 requests a month times 53% is 10.6 requests that come back non-affirmed on the first pass. Round it to 11. Each one needs a resubmission: records pulled, note rewritten, request refiled, status tracked. 11 resubmissions times $25.20 is $277.20 a month in rework, or $3,326 a year.

Then count the procedures that never get approved. Virtix told CMS that 63.7% of requests were affirmed on first review and 84.3% after resubmission. That leaves 15.7% never affirmed. 20 requests times 15.7% is 3.14 procedures a month that never happen. At the $400 illustration rate, that is $1,256 a month in lost revenue, or $15,072 a year.

Add the two lines. $3,326 in rework plus $15,072 in lost revenue is $18,398 a year. Each WISeR request costs this practice $76.66 on average: $13.86 in rework plus $62.80 in lost procedures.

These are illustration numbers built on Washington's published rates, not your numbers. Your vendor, your service mix, and your allowed amounts will differ. Arizona's 19% rate would cut the cost to about a third of the Washington figure. But the shape holds: a non-affirmation costs rework plus lost revenue, and the lost revenue is the bigger line by far.

The delay costs money too, even when the request is eventually affirmed. A procedure approved after 4 to 8 weeks instead of 2 pushes the claim out by a month. That claim sits in AR the whole time, and your revenue cycle management team works the same queue with older dates on it.

Run the same math on your own numbers. Pull your WISeR requests from the last 90 days, count first-pass non-affirmations, count requests that never got affirmed, and multiply by your average allowed amount. Our medical billing services team runs this math inside the free audit, and our pricing page shows what that work costs.

WISeR vs Medicare Advantage vs standard Medicare

Three kinds of Medicare prior auth now sit side by side, and the rules change with the program. Use this table to see which one your request falls under.

The row that matters most is the decision deadline. WISeR has a 72-hour goal with no hard penalty in the records for missing it, and one request waited 83 days. Medicare Advantage has a hard 7-day standard deadline and a 72-hour expedited deadline under CMS-0057-F, starting January 1, 2026, with electronic prior auth APIs due January 1, 2027.

How to get a WISeR request affirmed the first time

Virtix's own numbers show why documentation wins. The company affirmed 63.7% of requests on first review, then 84.3% after resubmission with better records. That 20.6-point jump came from paperwork, not from a change in the patient or the procedure.

The first submission has to carry the full case. A vendor reviewer reads dozens of requests a day, and a thin file gets a non-affirmation. Attach the proof before they ask for it.

For an epidural steroid injection, write the note like this: "Patient completed 6 weeks of conservative care: physical therapy twice weekly from September 1 to October 12, 2026, and daily NSAIDs with no relief. MRI dated October 5, 2026, shows L4-L5 disc herniation with nerve root compression. Requesting epidural steroid injection, CPT 62321, for L4-L5 radiculopathy, ICD-10 M51.16."

That note names the failed treatment, the dates, the imaging, the code, and the diagnosis in 4 sentences. A reviewer can check every line against the coverage rule without opening another file.

For a skin substitute request, write: "Wound present since [date], measured [size] on [date]. Standard wound care applied for 4 weeks from [start date] to [end date] with no closure. Vascular study dated [date] shows adequate perfusion. Requesting [product name] application, CPT [code], for [wound diagnosis, ICD-10 code]."

For a resubmission, open with a cover note that does the reviewer's work for them: "Resubmission for request [number], date of service [date]. Enclosed: office notes dated [date], MRI report dated [date], physical therapy records from [start date] to [end date]. The patient completed conservative care with no relief. Please affirm and issue a tracking number."

Three habits cut first-pass non-affirmations. First, submit the imaging and the prior treatment records with the first request, not after the denial. Second, match the diagnosis code to the procedure code before you submit, since diagnosis code reconciliation problems showed up in the FOIA records. Third, track every request by its tracking number and its 72-hour clock, and chase any request that passes 72 hours without an answer the same day.

Your clean claim rate benchmark is the number to watch alongside this. A clean WISeR submission is the same habit as a clean claim: right codes, right records, right the first time.

What happens next: expansion, Congress, and January 1, 2027

WISeR runs through December 31, 2031, and CMS has room to grow it. Records reported by KUOW show the administration hopes to expand WISeR to services including oncology treatment, cardiac catheterization, and advanced imaging. The CMS Innovation Center has the authority to add states and services without new legislation.

Congress may still stop it. The House voted unanimously in June 2026 to defund the model, and the Senate vote to end it failed 46 to 50 in July. The Murray and Wyden bill to prohibit WISeR is still live. Until one of those moves passes, the pilot keeps running.

Two dates matter for your calendar. First, January 1, 2027, when payers under CMS-0057-F must have electronic prior auth systems live, including the FHIR-based APIs that let your EHR send a request and get an answer without a fax or a portal. That rule covers Medicare Advantage, Medicaid managed care, and marketplace plans, not WISeR itself, but it sets the plumbing WISeR will eventually use. Our 2027 Medicare fee schedule post covers the other January 1, 2027, changes, including the conversion factor cut to $32.63, per the CMS fact sheet. Medical Economics covered the final rule on November 5, 2026.

Second, CMS has promised a 6-month report on WISeR's early results. It has not published appeal or overturn rates by vendor or service yet. That report is the one that answers the open question from the JAMA study: whether the 20% drop was waste or needed care.

Your WISeR checklist

What this means for your revenue

Medicare AI prior authorization is no longer a proposal. It runs in 6 states, covers about 6.4 million beneficiaries, and cut targeted services about 20% in its first 6 months. One vendor denied 53% of requests in its first quarter. Another denied 19%. Same program, same rules, a 34-point gap between vendors.

The pay structure rewards denials with up to 25% of the spending averted. CMS says clinician review and performance adjustments check that pull. The first-quarter numbers show the check has real work to do.

Your defense is the file you submit. Attach the imaging, name the failed conservative care with dates, match the diagnosis code to the procedure code, and track the 72-hour clock. Virtix's own numbers prove the point: resubmission with better records lifted affirmations from 63.7% to 84.3%.

Run the math from this article on your own requests. If you file 20 WISeR requests a month at Washington's rates, you are losing about $18,398 a year. Cutting your first-pass non-affirmation rate in half saves about $9,199 of that.

Get a free audit of your last 90 days of WISeR requests. We price your non-affirmation rate, your rework cost, and your lost procedures in 48 hours, and we show you the 3 fixes worth the most money.

Frequently asked questions

Sources

• Neprash HT, Barnett ML. Changes in Service Use Following the Introduction of AI-Enabled Prior Authorization in Medicare. JAMA. Published October 26, 2026.
• University of Minnesota, Medicare's prior-authorization pilot test led to an immediate drop in use of services, October 26, 2026: https://twin-cities.umn.edu/news-events/medicares-prior-authorization-pilot-test-led-immediate-drop-use-services
• Becker's Hospital Review, Medicare's AI-enabled prior authorization pilot linked to nearly 20% drop in some services, October 30, 2026: https://www.beckershospitalreview.com/finance/medicares-ai-enabled-prior-authorization-pilot-linked-to-nearly-20-drop-in-some-services/
• MedCity News, 7 Potential Problems with Medicare's AI Prior Authorization Experiment, November 2, 2026: https://medcitynews.com/2026/11/7-potential-problems-with-medicares-ai-prior-authorization-experiment/
• HealthExec, WISeR AI rollout marred by technical failures and limited communication, CMS documents reveal, October 29, 2026: https://healthexec.com/topics/healthcare-management/healthcare-policy/wiser-ai-rollout-marred-technical-failures-and-limited-communication-cms-documents-reveal
• daisyBill, AI-Driven Prior Authorization Infects Medicare, November 2, 2026: https://blog.daisybill.com/medicare-ai-prior-authorization
• TechTarget, CMS prior auth pilot shrinks service use amid public backlash, November 3, 2026: https://www.techtarget.com/revcyclemanagement/news/366653513/CMS-prior-auth-pilot-shrinks-service-use-amid-public-backlash
• Medical Daily, Medicare's AI-Assisted Prior Authorization Pilot Linked to 15% Drop in Spinal Steroid Injections Across Six States, October 28, 2026: https://www.medicaldaily.com/medicare-prior-authorization-pilot-cuts-epidural-injections-480513
• AARP, AI Prior Authorization Pilot Hits Original Medicare, February 19, 2026: https://www.aarp.org/medicare/original-medicare-ai-prior-authorization-pilot/
• OmniMD, WISeR Model: Medicare Prior Authorization Rules and Codes, November 2026: https://omnimd.com/blog/cms-wiser-model/
• American Medical Association, Only 1 in 3 doctors trusts insurers' prior authorization promises, May 18, 2026: https://www.ama-assn.org/practice-management/prior-authorization/only-1-3-doctors-trusts-insurers-prior-authorization
• Medical Economics, Physician advocates urge Congress to counter cuts in final 2027 Medicare Physician Fee Schedule, November 5, 2026: https://www.medicaleconomics.com/view/physician-advocates-urge-congress-counter-cuts-final-2027-medicare-physician-fee-schedule
• Centers for Medicare and Medicaid Services, Calendar Year 2027 Medicare Physician Fee Schedule Final Rule fact sheet, November 4, 2026: https://www.cms.gov/newsroom/fact-sheets/calendar-year-cy-2027-medicare-physician-fee-schedule-final-rule
• KFF, Claims Denials and Appeals in ACA Marketplace Plans in 2024, March 24, 2026: https://www.kff.org/patient-consumer-protections/claims-denials-and-appeals-in-aca-marketplace-plans-in-2024/
• Experian Health, State of Claims 2026: Why denials remain an uphill climb, October 28, 2026: https://www.experian.com/blogs/healthcare/state-of-claims-2026-why-denials-remain-an-uphill-climb/