Practical Applications of FDA’s Guidelines on Use of Real-World Data and Real-World Evidence to Support Regulatory Decision-Making for Drug and Biological Products

Sep 5, 2023

On August 31, 2023, the U.S. Food and Drug Administration (FDA) released “Considerations for the Use of Real-World Data and Real-World Evidence to Support Regulatory Decision-Making for Drug and Biological Products: Guidance for Industry.”[1] FDA issued this guidance as part of its Real-World Evidence (RWE) program stemming from the 21st Century Cures Act (Cures Act).[2],[3] The Cures Act was signed into law in 2016 to accelerate medical product development. One of the key provisions was for FDA to create the RWE program to evaluate the use of RWE in regulatory decision making.  

The newly released Guidance clarifies FDA’s expectations for clinical studies using real-world data (RWD) but does not establish any legally binding requirements. This Guidance discusses the applicability of 21 CFR Part 312 (investigational new drug application)[4] for studies using RWD and summarizes regulatory considerations for non-interventional studies using RWD. We summarize below the key facets of these considerations and discuss their practical application to the current landscape of available RWD.  

Applicability of 21 CFR Part 312 

21 CFR Part 312 outlines the requirements for Investigational New Drug Applications (IND) for review by the FDA. While they specifically concern interventional studies, these guidelines recognize the utility of RWD in these studies. The example applications of RWD are identifying potential participants for trials, ascertaining outcomes (e.g., diagnoses, hospitalizations), and creating external control arms. The guidelines also note that non-interventional studies are not clinical investigations subject to 21 CFR Part 312. 

Practical Application: The potential use of RWD in clinical research has been long recognized but remains an emerging practice. One challenge to overcome is finding sources of data with enough details to ascertain study inclusion/exclusion criteria, covariates and outcomes. Often, multiple sources must be used. It is also important to understand the underlying characteristics of the RWD source population, as most RWD are samples of convenience. Critical subgroups could be missing from the source, regardless of the level of detail that the RWD source offers. Finally, RWD sources that are readily available are generally de-identified. This presents challenges to finding patients to participate in trials. Careful evaluation of available data sources must be conducted before committing to use RWD in INDs. The FDA’s recommendations noted in these guidelines offer a path to evaluate the usability of RWD sources for this purpose. 

Regulatory Considerations for Non-Interventional Studies 

The majority of the FDA’s new Guidance discusses use of RWD for non-interventional studies. It describes the common sources of RWD for non-interventional studies, including registries, electronic health records (EHRs), and administrative claims. It recognized that while many non-interventional studies involve already collected RWD, some involve collection of additional data (e.g., questionnaires, laboratory tests, and imaging studies), which are subject to regulations protecting human subjects. Data privacy is also noted as a key consideration in using RWD.  

The Guidance discusses requirements to establish transparency in the design of the study. Steps to be taking include sharing drafts of protocols and statistical analysis plans (SAPs), publicly posting protocols, evaluating sources of RWD for fit and assurance that the sources as well as study cohort selection and analysis do not favor a certain conclusion, justification of selected data sources, and maintaining audit trails and documentation of the data extraction and analyses. 

The Guidance recommends that for studies intended for use in a marketing application, sponsors should ensure that they can submit patient-level data and can make the source data available if further verification is necessary. The source data could include “original records of clinical findings, observations, or other activities in a clinical investigation used for reconstructing and evaluating the investigation”, though the Guidance does note that appropriate justification may exist for why the sponsor cannot submit patient-level data. Programming code and algorithms submitted to the FDA must be well documented and complete enough to allow for study replication. 

The Guidance also offers recommendations for study monitoring (from RWD extraction to application of the protocol/SAP), safety reporting, study oversight, and researcher and third-party roles and responsibilities. 

Practical application: Study sponsors typically rely on outside parties for access to RWD. Most of these guidelines cannot be successfully implemented without close partnership with the RWD source. Input from the RWD source can be critical to protocol and SAP design, particularly in determining whether the source meets study requirements. RWD sources vary in their ability (contractual and technical) to track the linage of the patient record, share records with additional parties, or provide detailed documentation. It is important to communicate study requirements and FDA expectations to the RWD source and ascertain their ability to meet these requirements. RWD sources have varying strengths and limitations.  

It is also important to discuss how sources can or cannot meet these guidelines and engage with the FDA in the selection of the optimal source given specific study requirements. To start, the source of RWD is not always the generator of the data. For example, RWD could be sourced from a vendor who is aggregating claims from multiple payers or an aggregator who is procuring EHRs from hospital software vendors. Accessing RWD closer to the source offers greater transparency and traceability, as well as opportunities to collect additional retrospective or prospective data. Also, most sources of RWD include only de-identified patient data. De-identification involves the removal of direct identifiers and the removal or obfuscation of indirect identifiers. Additionally, the RWD source may be required to anonymize providers, facilities, or payers. It is important to understand what data elements have been redacted and consider any impact to the study.  


Stakeholders in medical product and drug development have long recognized the practical application of RWD to accelerate innovation and bring them to patients who need them faster. The FDA’s Guidelines summarize many of the methods already in practice and stress the importance of planning and communication between the sponsor and regulator. It is equally important to find and partner with the right RWD sources who can offer not just the patient population and data points necessary for a study, but offer enough transparency about the data and their origin to meet the recommendations in this Guidance while respecting patient privacy and ensuring data security.  





By continuing to use the site, you agree to the use of cookies. More information.

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below, then you are consenting to this.