Medicare is testing the use of AI to conduct prior authorizations to target wasteful medical spending
The AI pilot program started in January in six states.
A Medicare pilot program that used AI to pre-approve patient care led to a large drop in medical services immediately after the venture started, a new study found.
Under the new program that started in January, about 6.4 million Americans enrolled in traditional Medicare in New Jersey, Ohio, Oklahoma, Texas, Arizona and Washington state face prior authorizations conducted by AI.
Prior authorization means that a patient must get approval before an insurer will pay for a medical service or drug. Traditional Medicare currently has fewer prior-authorization requirements than Medicare Advantage, which is the private-plan alternative to Medicare.
The program - known as the Wasteful and Inappropriate Service Reduction (WISeR) model - requires additional approvals for more than a dozen medical procedures, such as epidural steroid injections for pain management, minimally invasive treatments for spinal stenosis in the back and osteoarthritis in the knees, and nerve-stimulation therapies for conditions such as sleep apnea and incontinence.
The drop in medical services was biggest right after the launch of the program and leveled off slightly over time, according to the study from the University of Minnesota School of Public Health and Brown University School of Public Health. The study, published in JAMA, provided the first statistical evidence about the pilot program.
"Prior authorizations seems to be an effective tool to reduce the use of those healthcare services," Hannah Neprash, lead author of the study and associate professor at the University of Minnesota, told MarketWatch. "There was a large drop in utilization in those services that suddenly needed prior authorization."
The WISeR program is part of the Trump administration's efforts to cut what it calls waste, fraud and abuse in government. The companies contracted to do the review process receive payments when they reduce costs.
The study comes as older adults currently face healthcare-coverage decisions. The annual Medicare open-enrollment period runs through Dec. 7. During that time, people aged 65 and older can switch from traditional Medicare to a private Medicare Advantage plan, change from one Medicare Advantage plan to another, and join or change a Medicare Part D prescription-drug plan. The changes go into effect on Jan. 1.
More than 11,000 people in the U.S. turn 65 every day, and more than 70.5 million are enrolled in Medicare. Medicare Advantage currently controls more than half of the overall Medicare market. Medicare Advantage plans offer extra benefits such as vision and dental coverage and cap annual out-of-pocket costs but have a limited choice of doctors and extra requirements or prior authorizations for specialist referrals.
In the study, the medical services that needed prior authorization declined sharply. Compared with trends in states outside the pilot, epidural steroid injections for back pain - the highest-volume service in the model - dropped 15% over the six-month study, while other targeted services declined about 20%, the study found.
By quarter, the epidural steroid injections saw the biggest decline immediately of 21% in the first quarter of 2026, but eased to a drop of 9% in the second quarter. The other services saw the drop narrow from 21% in the first quarter to a decline of 18% in the second quarter, according to the study.
However, services that were used as a control group - and did not need prior authorization - showed no significant change in usage patterns, according to the study.
It was unclear why the usage leveled off somewhat over time, Neprash said. It could be, for example, that the vendors worked out initial "kinks" or problems with the system, or that physicians got better at knowing what was necessary to get a case approved, she said.
The study's researchers looked at monthly use of WISeR-targeted services among traditional Medicare patients for the two years before the program's launch and the first six months after. Future work should examine whether the usage patterns persist over the total duration of the six-year pilot program, the study said. Also, researchers should study patient outcomes, the physicians' burden to meet the prior authorization threshold and Medicare spending over the longer term, the study said.
-Jessica Hall
Comments