Real-world evidence: advances in medical devices

Authors: Daniel Caños, Ph.D., MPH, Director, Center for Devices and Radiological Health (CDRH) Clinical Evidence Analysis Office, and Michelle Tarver MD, Ph.D., Director, CDRH

Today’s healthcare ecosystem is generating richer and more diverse data streams, providing new opportunities to modernize how real-world data (RWD) produces relevant and reliable real-world evidence (RWE). The U.S. Food and Drug Administration is working to more clearly endorse and more consistently accept RWE as valid scientific evidence. RWE is clinical evidence about a medical product’s use and potential benefits or risks derived from the analysis of RWD. Properly validated RWE can support a wide range of regulatory activities, including its role as key clinical evidence in premarket applications.

Based on the basic research presented in the 2021 publication “Examples of Real-World Evidence (RWE) Used in Medical Device Regulatory Decisions,” FDA announced 73 new examples of marketing authorizations using RWE from FY 2020 to FY 2025. FDA’s first guidance on RWE was issued in 2017 by FDA’s Center for Device and Radiological Health (CDRH). This guidance clarified how CDRH evaluates the relevance and reliability of RWD to support medical device regulatory decision-making. In 2019, CDRH published 90 examples of various types of premarket applications supported by RWE. Then, in December 2025, CDRH published an updated version of its RWE guidance titled “Using Real-World Evidence to Support Regulatory Decision-Making for Medical Devices.” This guidance explains how the least burdensome approach can be applied when assessing whether RWD is relevant and reliable for marketing decisions. For example, expanding the indications for use, licensing, and approval of new devices. It also describes recommended methods for using RWE and provides FDA review considerations and best practices.

The 73 new examples represent a wide range of clinical and device areas across CDRH across the product lifecycle and illustrate the evolving landscape of data sources used to generate RWE. However, these do not constitute a complete list of all applications that relied at least in part on RWE to inform regulatory decisions. While registries have long played a central role, other data sources such as electronic health record aggregators, administrative claims and claims data, device-generated data, digital health platforms, and increasingly sophisticated data linkages are also contributing to a broader and more dynamic evidence environment.

These new examples also highlight the increasing use of RWD to validate device software functionality, such as artificial intelligence (AI) and machine learning (ML)-enabled technologies. for example:

• FDA relied on retrospective medical record data generated during a patient’s hospitalization in an intensive care unit to support approval of an ML-based software feature to predict hemodynamic instability (K200717). The ICU-generated data came from an evaluation of device performance using physician-reviewed records as a reference standard.
• FDA evaluated the validation of two separate AI-enabled electrocardiogram-based software features supporting the approval of two separate premarket notifications (K250119 and K250649) using medical record data across multiple geographically dispersed U.S. locations. Medical record data helped FDA compare the device’s output to echocardiogram-derived left ventricular ejection fraction measurements obtained in routine clinical practice.
• FDA evaluated automated event detection software for polysomnography using archived sleep records from routine clinical care. As part of its evaluation, the FDA compared the device’s output to manual scoring by clinicians in the same study. These data support the clearance of premarket notifications K241960 and K221179.

These examples demonstrate how RWD can support rigorous validation of new device software features across multiple disease areas.

FDA is redoubling its efforts to leverage RWE across a broader range of medical devices to support timely access to safe and effective technologies, foster innovation, and strengthen the foundations of regulatory decision-making. We also continue to work with the National Health Technology Evaluation System (NEST) to evaluate and identify high-quality RWE sources that can be used for regulatory purposes. This comprehensive approach allows RWE to be used more robustly and efficiently to characterize the performance of medical devices in real-world environments.

We encourage interested parties to consider these examples to better understand how RWE has been and will be widely applicable.