First study of Medicare's WISeR prior auth pilot finds 15-20% drop in targeted services
University of Minnesota and Brown University researchers (Hannah Neprash and Michael Barnett) compared traditional Medicare use of WISeR-targeted services in the six pilot states with use in all other states. The data came from Epic Cosmos, and the study covers the two years before the January 2026 launch and the first six months after. Epidural steroid injections, the highest-volume service, fell about 15%, and other targeted services fell about 20%. Services that were slated for the pilot but never implemented showed no significant change.
Why it's notable: This is the first outside evidence that the AI-assisted prior authorization pilot cuts utilization in Arizona, New Jersey, Ohio, Oklahoma, Texas and Washington. The effect shrank over time: epidural injections went from about -21% in Q1 to about -9% in Q2. The authors say they cannot tell whether the avoided care was unnecessary or needed, which is the question clinics and patients most care about.