Practice Pattern Changes After Adoption Of Diagnostic AI Tool Used In Conjunction With Cardiac Imaging Anna Zink, Michael E. Chernew, and Hannah T. Neprash

https://www.healthaffairs.org/doi/full/10.1377/hlthaff.2025.00929 In 2018, Medicare established coverage and reimbursement for its first service using artificial intelligence (AI): computed tomography (CT) fractional flow reserve (FFRCT). FFRCT is used in conjunction with cardiac imaging to diagnose coronary artery disease. Medicare reimbursement provides the opportunity to observe clinicians’ adoption of FFRCT and examine changes in utilization, spending, clinician productivity, and patient outcomes associated with its use. In this study, we exploited variation in the timing of FFRCT adoption by clinicians, quantifying changes in practice patterns before and after adoption compared with nonadopters. We found that use of the underlying coronary artery disease test required for FFRCT increased after the adoption of FFRCT, whereas invasive test use declined. On net, per patient diagnostic spending for coronary artery disease increased. We found no evidence that adopting FFRCT facilitated earlier diagnosis of coronary artery disease, although clinicians saw a larger share of patients with coronary artery disease after adoption, and clinician productivity (that is, total visit count) increased. After clinicians adopted FFRCT, we observed a decrease in rates of cardiac-related adverse events among their patients.