FDA Finalised Its Real-World Evidence Guidance for Medical Devices in December 2025

The US Food and Drug Administration (FDA) has published a final guidance on the use of real-world evidence (RWE) to support regulatory decision-making for medical devices. The Federal Register notice announcing its availability is dated 18 December 2025 and identifies the docket as FDA-2023-D-4395. According to the notice, the final guidance supersedes the final guidance issued on 31 August 2017 and provides expanded and updated recommendations.

The document responds in part to section 3629 of the Food and Drug Omnibus Reform Act (FDORA), which directs the FDA to issue or revise guidance on the use of real-world data (RWD) and RWE to support regulatory decision-making. The guidance sets out the factors that sponsors should assess to show whether RWD are relevant and reliable for a particular regulatory decision. Commentators describe it as a document of around 44 pages that explains how the agency evaluates whether RWD are fit for purpose.

As with all FDA guidance, the document is not legally binding. Secondary sources reviewing the text note that it does not lower the evidentiary bar: RWE is evaluated in context to determine whether it meets the quality standards for the decision at issue, and the strength of submitted RWE is considered on an application-by-application basis.

The most widely reported change concerns access to patient-level data. Reporting by the Regulatory Affairs Professionals Society and others indicates that the FDA will accept RWE in eligible submissions without always requiring identifiable patient-level data. The earlier draft acknowledged that patient-level data could not always be obtained but stopped short of endorsing the use of large de-identified databases; the final text goes further in accepting such sources where they are shown to be fit for purpose.

Law-firm analyses identify several further changes. The final guidance adds examples showing when the FDA will and will not accept investigational device exemptions that draw on RWD. It places greater emphasis on protocol development and cautions that sponsors should not access outcome results while the protocol is still being written. It also expands on how sponsors should account for potential differences between treated and control groups, and Appendix B now includes more detailed examples of RWE used for new premarket approvals and for expanded indications.

Not all observers see a large shift. One law-firm commentary noted that, given the limited changes from the draft, it remains to be seen whether review practice will change materially. Sponsors should therefore read the final text alongside their own regulatory history rather than assume a broad relaxation of expectations.

The Federal Register notice recognises that the agency and industry may need time to put the recommendations into practice. For submissions pending at publication, and for those received within 60 days after publication, the FDA generally does not expect sponsors to have included the newly recommended information. An FDA webinar timeline lists 17 February 2026 as the point after which the new recommendations are generally expected, and the FDA held a town hall on the final guidance on 18 February 2026.

The guidance sits within a wider FDA programme on real-world evidence. For drugs and biological products, the agency has separate guidance documents, including an August 2023 guidance on considerations for the use of RWD and RWE and a December 2023 guidance on assessing registries, and it issued a final guidance implementing ICH M14 on non-interventional safety studies, announced in the Federal Register on 4 March 2026. Device sponsors should not assume that drug-side guidance applies directly to device submissions.

For sponsors, the practical effect is a clearer statement of what the FDA expects to see when RWD form part of the evidence package. Submissions should explain why the chosen data source is relevant to the device, population and outcome in question, and how its reliability has been assessed, including data accrual, completeness and quality assurance. Where de-identified databases are used, the submission should show how the absence of identifiable patient-level data has been addressed.

Protocol discipline matters. Because the final text warns against looking at outcome data while a protocol is being developed, sponsors should document the timing of protocol finalisation and data access. Early engagement with the FDA through the pre-submission process remains the most direct way to test whether a proposed RWD source and design are likely to be acceptable for a given regulatory decision.

Medical writers preparing device submissions should ensure that study reports describe data provenance, linkage and handling of missing data in enough detail for reviewers to judge fitness for purpose. Clear documentation of how control groups were constructed, and of any sensitivity analyses, will help reviewers apply the guidance consistently.