Medicare's AI-Assisted Prior Authorization Pilot Cut Epidural Injections 15 Percent. The Study Cannot Say Whether They Were Needed.
The first study of the WISeR model, in JAMA Oct. 26, found targeted services fell 15 to 20 percent in six states. The injection effect faded by spring.
Epidural steroid injections fell 15 percent and other targeted services about 20 percent in the six states testing Medicare's WISeR prior authorization pilot, relative to the rest of the country, the first study of the model found in JAMA on Oct. 26.
WISeR, the Wasteful and Inappropriate Service Reduction model run by the Center for Medicare and Medicaid Innovation, began Jan. 1. It brings prior authorization to traditional Medicare for selected outpatient services in Arizona, New Jersey, Ohio, Oklahoma, Texas and Washington, through technology companies that CMS encourages to use artificial intelligence. The research letter, by Hannah T. Neprash, PhD, of the University of Minnesota School of Public Health and Michael L. Barnett, MD, of the Brown University School of Public Health, notes that until now there was "no evidence to date on its initial impact."
What did the study measure?
Monthly service use, not outcomes. The researchers used Epic Cosmos, an electronic health record dataset covering more than 300 million patient records, and compared the six states with all other states and Washington, D.C., over the two years before launch and the first six months after, the University of Minnesota said. Services that had been planned for WISeR but were never added showed no significant decline, which the authors read as evidence that the prior authorization requirement itself drove the drop.
Did the effect last?
It faded, especially for injections. The reduction in epidural steroid injections was about 21 percent in the first quarter of 2026 and about 9 percent in the second. For the other targeted services, it went from about 21 percent to about 18 percent. By The Script Pad's arithmetic, the injection effect shrank by more than half within three months, while the effect on the other services held at roughly four-fifths of its starting size. The authors suggest the fading may reflect early workflow problems being fixed, clinicians learning the requirements, or both.
"The key question we can't yet answer is whether the care not received was wasteful or necessary," Neprash said in the university's announcement. Barnett said, "The early results suggest that doctors may be adapting rapidly to the new system."
A 15 percent drop is a measure of friction, not of value. The study cannot say whether the injections that did not happen were the ones that should not have.
What are the limits?
The design is observational, and it covers six months of a six-year model. It did not measure pain, function, complications, denials, spending or whether patients moved to other treatments, Medical Daily reported. CMS has not published its own evaluation. The scale is not small: KFF estimated that 6.4 million traditional Medicare beneficiaries lived in the six states in 2024, and about 207,500 of them received a service now on the WISeR list that year.
CMS says the model does not change Medicare coverage or payment policy, excludes emergency and inpatient-only services, and requires licensed clinicians to make any recommendation against payment. Participants are paid a share of the spending tied to averted care, adjusted for performance measures that include provider experience. The model page also says providers with records of compliance "may be exempt from the WISeR review process in the future."
What it means in clinic
Practices in the six states that perform injections, nerve stimulator implants or skin substitute applications should track their own submission dates, turnaround times and denials now, because that record is the evidence for any future exemption and for appeals. Document conservative treatment tried and functional limits in the note before the request goes out. Our guide to AI in prior authorization and claim review and our look at the quiet math of prior authorization cover how to build that log.
Reported, fact-checked and edited by The Script Pad's editorial staff. See How we use AI.
This article is for professional education and does not replace clinical judgment. Treatment decisions should be based on the individual patient and current guidelines.
