New Medicare coverage model broadens prior authorization requirements

Although widely used by Medicare Advantage and commercial plans, traditional Medicare has rarely imposed prior authorization (PA) requirements for health care services. In 2023, less than 1% of traditional Medicare spending on Part B items and services were subject to PA.  

In June 2025, the Centers for Medicare and Medicaid Services (CMS) announced the Wasteful and Inappropriate Service Reduction (WISeR) model, which will expand PA requirements within traditional Medicare. This model will run from January 2026 through December 2031 in six states, including Ohio, and mandate participation by all clinicians and suppliers providing included services, such as knee arthroscopy, nerve stimulator placement and skin substitutes. CMS plans to work with companies that employ technologies like artificial intelligence (AI), machine learning or algorithmic-derived decision logic in order to partially automate the PA process.  

A newly published Viewpoint column published in JAMA considers the promise and perils of this model. The team of co-authors from The Ohio State University College of Medicine, Yale University and the University of California San Francisco, was led by Vinay Rathi, MD, MBA, who is an assistant professor in the Department of Otolaryngology - Head & Neck Surgery at the College of Medicine.

“This model does hold promise to generate meaningful savings and deter medically unnecessary care,” Dr. Rathi says. “But it is unusual that CMS did not solicit public feedback before finalizing the design. And some important questions remain unanswered.” 

The WISeR model provides direct financial incentives for technology firms to deny care, though CMS has implemented some safeguards against poor performance. Nonetheless, the authors express concern that the introduction of artificial intelligence-driven denial volume could overwhelm the administrative capacity of Medicare, intensify the already contentious denial and appeal processes, inflate associated administrative costs. The authors additional raise concerns about the potential negative effects of algorithmic bias. 

“There are already well-documented concerns that PA requirements can result in delays or denials of appropriate care and exacerbate clinician burnout,” Dr. Rathi says. “Adding artificial intelligence to the mix could introduce bias that disadvantages vulnerable populations. Private payers are already wrestling with this problem. We think CMS needs a strong monitoring plan in place to protect patients.” 

For now, provider organizations in selected states must prepare for participation or risk inability to provide recommended care and payment denial. The authors conclude in the Viewpoint: “WISeR may reduce traditional Medicare spending, but at what cost remains unclear.”