FDA shares insights on growing use of real-world evidence in medical device approvals

A new FDA report describes how real-world data (RWD) were used in 73 medical device marketing authorizations between fiscal years 2020 and 2025, highlighting the expanding range of data sources, study designs, and regulatory uses associated with real-world evidence (RWE).

The report, ‘Examples of Real-World Evidence Used in Medical Device Regulatory Decisions (Fiscal Years 2020–2025)’ from the FDA’s Center for Devices and Radiological Health (CDRH) builds on its earlier 2021 review of 90 examples from fiscal years 2012 to 2019. Taken together, these analyses reflect the increasingly routine integration of RWE into medical device regulatory decision-making, supporting activities such as premarket submissions, indication expansions, software validation, and selected post-market requirements.

According to the report, the examples demonstrate “the continued evolution and expansion of real-world evidence (RWE) use in medical device regulatory decision-making across the Center for Devices and Radiological Health (CDRH).” The 73 examples span all eight Offices of Health Technology and cover a range of regulatory pathways, including 44 premarket notification submissions, seven de novo classification requests, one humanitarian device exemption application, nine premarket approval applications, and 12 PMA supplements. While most relate to premarket activity, three examples also demonstrate how RWE can be used across both premarket and post-market stages.

A consistent theme throughout the report is the diversification of data sources used to support regulatory decisions. Registries remain widely used, but the examples also reflect increasing reliance on electronic health records, administrative claims, device-generated data, public health surveillance data, billing records, and linked datasets. Newly added sections on device-generated data and public health surveillance further highlight what the report describes as “the expanded list of RWD source types described in the Food and Drug Administration (FDA) 2025 guidance… and the evolving landscape of data sources used to generate RWE for medical devices.”

The examples also illustrate that RWE is used in multiple ways depending on the regulatory objective. In some cases, it functions as the primary source of clinical evidence supporting a submission, while in others it complements traditional clinical trials, supports comparative analyses, or contributes to post-approval requirements. These differences are reflected in the range of case examples described throughout the report.

Several case examples demonstrate how these approaches are applied in practice. For example, TheraSphere (P200029) received approval in 2021 using retrospective medical record data as a primary source of clinical evidence to support conversion from a humanitarian device exemption to a premarket approval pathway. The submission also incorporated surveillance data on adverse events reported during commercial use, demonstrating how multiple data streams can contribute to a single evidentiary package.

A later example involved Intuitive Surgical’s da Vinci X and Xi Surgical Systems (K240852), which were cleared in 2025 for a labeling expansion to include tracheobronchoplasty (TBP). In this case, medical records and chargemaster billing data from the Premier Health Database (PHD) were used to assess safety and effectiveness outcomes in patients who underwent TBP procedures at 6 US hospitals from 2013 to 2023, with a systematic literature review providing additional comparative context. This approach shows how routinely collected healthcare data can support procedural expansions without requiring entirely new prospective trials in all cases.

Registry-based strategies also feature prominently across the examples. Inspire Medical Systems used retrospective data from its ADHERE Registry to support an indication expansion for its upper airway stimulation system (P130008/S090), including updated patient eligibility parameters. Similarly, Motiva USA’s breast implants (P230005) incorporated supplemental registry evidence from both the Australian Breast Device Registry and the Dutch Breast Implant Registry, demonstrating how international registry data can strengthen evidence packages alongside prospective clinical studies.

Beyond traditional devices, the report highlights increasing use of RWD to support software validation, particularly for AI and machine learning-enabled technologies. It identifies “multiple examples of artificial intelligence and machine learning-based medical devices that demonstrate the validation of device software functions using RWD,” including systems assessed using intensive care unit records, electrocardiogram data compared with echocardiography results, and archived sleep recordings used to evaluate automated detection software.

The examples reflect continued methodological development, including greater use of linked datasets, hybrid designs that combine prospective and RWD, and advanced statistical approaches such as “propensity score methods and other advanced statistical techniques to address confounding and selection bias,” highlighting increasing attention to rigor in observational studies. At the same time, the report shows how RWE is becoming more systematically integrated into medical device regulatory pathways, offering practical insight into how evidence-generation strategies are evolving alongside advances in data availability and analytical methods.