The pharmaceutical industry is eager for more information about Real-World Evidence. How can it be used? Will the FDA accept it? How can it best be collected, and from what sources?
Yesterday, the FDA offered some clarity about one aspect of RWE that you might not have thought about: How can RWE be formatted in submissions to the FDA?
It’s actually a really complex question, and it’s not clear that even the FDA has a great answer at the moment. AgencyIQ researchers Laura DiAngelo and Kedest Tadesse have more information for us below.
Also, do you like free webinars? We’re going to be hosting one on Thursday, November 4th at 1 PM ET looking at the FDA’s upcoming user fee reauthorizations. We’ll be talking about PDUFA, MDUFA, GDUFA, BsUFA and what we know about what’s coming next. We’ll have a deep dive on all the known details and what they mean for the life sciences industry. You can sign up for free here.
Interested in getting my daily newsletter, FDA TODAY, in your inbox instead of on LinkedIn??You can sign up for free here.?
FDA issues draft guidance on how to submit real-world data, previewing future activities
The top line: Drug sponsors will need to convert real-world data (RWD) sets into established standards when submitting this information in their regulatory applications, according to a new draft guidance by the FDA. While the agency acknowledges that its current catalogue of standards doesn’t necessarily reflect data derived from real-world sources, it indicated that it’s looking into updates – and recommended mapping, for now.
First, let’s get you up to speed on FDA submission requirements.
- Under section 745(A) of the Food and Drug Administration Safety and Innovation Act (FDASIA) of 2012, many types of regulatory applications are now required to be submitted electronically. In general, the change was intended to help move away from paper-based applications and toward digital applications, which are easier for FDA to receive, review, store and analyze – especially as data sets get larger.
- The FDA requires study data to follow certain standards in electronic submission. Study data contained in “certain” Investigational New Drug Applications (IND), New Drug Applications (NDA), Abbreviated Drug Applications (ANDA) and certain Biologics License Applications (BLA) must be submitted electronically using a “format that FDA can process, review, and archive.” This requirement also applies to submission subsequent to the initial application (e.g., amendments, supplements, reports).
- This involves formatting the data into a specific data standard. According to the FDA’s guidance, “study data standards describe a standard way of exchanging study data between computer systems.” These standardization elements include data organization, standard names for variables, templates, among other factors. Standardization of data allows for a uniform data presentation, minimizing inconsistency and increasing clarity for regulatory bodies. It also accelerates the speed at which the FDA can ingest, exchange and analyze data.
- The agency maintains a Data Standards Catalogue (“Catalog”) that outlines the currently supported and/or required standards for particular uses of data, as well as their date of recognition and when they were required. In general, clinical and non-clinical data are submitted to the FDA in standards from the Clinical Data Interchange Standards Consortium (CDISC). These include the Study Data Tabulation Model (SDTM), used for clinical data submissions, and the Standard for Exchange of Nonclinical Data (SEND), which is used for nonclinical data submissions.
- Submission in a “format that FDA can process, review, and archive” (i.e., in the particular standard outlined in the Catalog for the use) is critical. The FDA will issue a ‘refuse to file’ or ‘refuse to receive’ notice for any submissions that are not filed electronically or follow the required format – unless a waiver was granted.
Why are we telling you this? Well, if companies want to be able to use Real-World Evidence, they need to be able to submit that evidence.
- Data sources for clinical research and regulatory have expanded from data collected only from clinical trials to data collected from the real-world settings – including electronic health records (EHRs), medical claims data, patient generated health data (PGHD), clinical registries and mobile digital tools, among other sources. However, these data tend to follow their own rules of capture and standardization and are typically not formatted or presented in the ways that the FDA accepts and reviews data.
- Under the 21st Century Cures Act, the FDA was directed to provide guidance on the use of real-world evidence (RWE) in regulatory decision making – a goal on which progress has been slow. Overall, the agency is concerned about the relevancy and reliability of real-world data (RWD), which is captured for non-regulatory purposes and may not be of high enough quality for regulatory use or capturing enough data points to make causal inferences about specific products and their impacts on patients.
- However, the agency has committed to working through these issues. Under the current iteration of the Prescription Drug User Fee Act (PDUFA), the agency held workshops and issued draft guidance documents to outline preliminary thinking on the use of real-world evidence (RWE) to support drug submissions. Under the upcoming reauthorization, the agency will go further, establishing a pilot program to review volunteers’ RWE studies for regulatory use.
- FDA first the agency needs to establish an operational baseline for the way that the data itself is handled, described, and transmitted. Some sources of RWD have their own data standards – for example, data in most EHRs are captured in accordance with standards recognized by the Office of the National Coordinator for Health Information Technology (ONC), diagnosis codes typically follow the ICD system, and claims and billing information will be uniform by payer. However, none of these directly align with the way that the FDA prefers to receive data, and other sources of RWD (e.g., PGHD, mobile app data) do not necessarily conform to any recognized standard.
- In September 2021, the agency issued a draft guidance outlining baseline expectations for assessing whether electronic health data from EHRs and claims would be appropriate for analyses. While the guidance did not provide recommendations on how sponsors should consider the analysis of RWD to build RWE, it delineates recommendations – and areas of concern – for transitioning source data from EHRs and claims into RWD for potential regulatory use. [Read AgencyIQ’s analysis of that draft guidance here.]
Now the FDA is out with a new draft guidance document on the submission of RWE in accordance with submission standards.
- On October 21, the FDA published draft guidance on applying the section 745(A) requirements on data standard submission in regulatory applications. The guidance outlines FDA data standard considerations for RWD that is submitted to the agency in support of an application. The newly published document discusses the application of currently supported data standards for RWDs in addition to providing an extensive list of definition of terms.
- As outlined above, there are several challenges associated with standardizing data collected in the real world. In addition to the various RWD sources and the formats used (for example, data captured in an EHR is different from data captured in a dedicated disease registry), the terminologies, database structure, coding systems, data aggregators and exchange formats used are not uniform across these sources. The guidance acknowledges these issues in the guidance, stating that “FDA recognizes the challenges involved in standardizing study data derived from RWD sources for inclusion in applicable drug submissions.”
- When considering the submission of RWD, the agency still expects the sponsor to align with the standards in the Catalog and the formatting in the Study Data Guidance, “absent a waiver,” states the guidance. However, those standards do not technically align with the way that real-world data will likely be captured in their original sources (potentially barring LOINC, the data capture system for laboratory products, or SNOMED metrics of indication and usage).
- Going forward, “FDA plans to issue further guidance and/or to update the Catalog with standards for study data that are derived from RWD sources,” the guidance indicates. Until that time, however, sponsors will need to convert their data into one of the recognized FDA standards – a process called mapping (“the process of creating data element linkages between two distinct data models”).
- When mapping the data from its source data model into the FDA-recognized version, the agency expects sponsors to show their work, explains the guidance. This means that sponsors should document their mapping methodology, as data domains may not correlate directly between the two models. For example, the guidance cites disparities in the definition of sex and gender – “sex as a variable may be codified in CDISC’s terminology as a concept based on physical characteristics, whereas EHRs may use gender identity,” the guidance notes.
- Sponsors will need to document their reasoning for choosing “particular CDISC data elements,” the differences in the two standards’ definitions of key variables, and the impacts of mapping in their submissions “as part of or in an appendix to the Study Data Reviewer’s Guide,” and should include a data dictionary outlining the data elements, various definitions, and its “relationships to other data, origin, usage and format.”
- Beyond the difficulties of mapping one data standard to another, the process of transforming source data into RWD that can be aligned with the FDA’s standards is likely to be challenging , the agency acknowledges in the draft guidance. For example, inconsistent coding or miscoding, duplicative records for single patients (or duplicative records of certain procedures) or missing information is likely to be an issue with all source data. For now, the agency expects sponsors to “document data challenges” and “justif[y] their approach to enable the application of an FDA-supported data standard.” Both mapping and transformation “can be handled” using the Define-XML, which “transmits metadata that describes any tabular dataset structure” – in this case, the CDISC SDTM and/or SEND, as well as the Analysis Data Model (ADaM) for analysis data sets – with a narrative of the process and justifications outlined in the Study Data Reviewer’s Guide, states the guidance.
What should companies be thinking about? Here’s AgencyIQ’s take:
- This guidance sets another step forward for the regulatory use of RWD by outlining the agency’s recommendations for the operational submission of these data. Like the draft guidance on the electronic health data, however, the agency does not recommend any particular methodology. Instead, the approach taken in both guidance documents is to highlight the challenges with working with source data, expected issues with converting it into RWD, and the ways in which sponsors will need to be mindful of potential issues, document their processes and justify (and assess the impact of) their choices.
- Notably, the guidance comes after both the agency and relevant standard organizations have been looking to standardize and align source data with regulatory expectations. In September, CDISC announced that it had partnered with HL7, the organization that designs EHR data standards, to develop a mapping tool. The FHIR to CDISC Joint Mapping Implementation Guide is intended to streamline the processes outlined it the guidance – in effect, translate EHR data into regulatory submission-ready taxonomy. When following mapping took, a sponsor can theoretically convert EHR data into CDISC – exactly the issue that the agency is seeking to address with the draft guidance.
- However, both organizations at the time acknowledged that the first iteration of the mapping tool is very much a first step – the guide is “purely a ‘descriptive’ guide,” and a manual process. Further, as HL7 and CDISC identified during the guide development process, “there are several areas where different approaches between FHIR And CDISC create challenges for mapping between standards,” such as different definitions and placement of adverse events/clinical events and medical history variables between the standards.
- This means that the organizations that develop the standards themselves have faced challenges mapping their taxonomies together – an issue that sponsors are nearly certain to face as well. Taking a longer-term view, however, the partnership could present a reason for optimism: with CDISC and HL7 working together, the organizations will have insight into each other’s processes and “understand each other’s terms and better align both sets of specifications as they continue to evolve.”
- Notably, the question of evolving standards will also likely present a key issue for sponsors going forward, especially when trying to map RWD to FDA-recognized standards . Data standards for non-research sources are always evolving, with new elements, codes, and data fields added regularly to capture new and evolving clinical information, even as coding practices change due to clinical guidelines or financial incentives. These issues are acknowledged in the FDA’s recent baseline RWD guidance documents, with the agency stating that sponsors are responsible for identifying and adjusting for how coding and data capture evolve in real-world settings over time. However, the agency does not have any answers for how to account for these issues – yet.
- Overall, it’s likely that the FDA needs a more in-depth understanding of the data itself before it can make specific recommendations on these issues – or recognize new data standards in its Catalog that may adjust the mapping expectations, as the agency noted it intended to do in the guidance. However, the FDA’s work on contextualizing data quality from certain real-world sources has already begun. Under the Sentinel Initiative, which is FDA’s real-world surveillance and pharmacovigilance system, the agency recently announced a new research project to standardize the way that data from EHRs can be considered of high enough “quality” to be used for regulatory decisions. According to the project announcement , Sentinel’s Data Partners will “create a publicly-available common protocol for Data Quality Metrics for electronic health record (EHR) data.” Researchers will work to define specific metrics of data quality for both inpatient and ambulatory care setting data. This work, when it’s completed, will likely help inform both the sponsors who are seeking to use this data (and describe its attributes in their submissions) and the review teams tasked with understanding its relevance and reliability to answer key regulatory questions.
What we're watching and reading
- …this announcement that the FDA plans to work with Aetion to “advance regulatory science and innovation” through the use of the company’s Evidence Platform . As the announcement explains: “The aim of this project is to further data familiarity and protocol standards to support real-world data (RWD) analyses among the broader research community. FDA will work with Aetion to define and prioritize key research questions; identify fit-for-purpose data sources; develop appropriate, validated, and applicable measurement algorithms to capture key exposures, subgroups, confounding variables, and outcomes; design template epidemiological studies applicable to a range of treatments; implement studies and generate transparent reporting using AEP; and build and disseminate knowledge via peer-reviewed publications and other avenues.”
- …this article indicating that the Council for Responsible Nutrition supports mandatory product listing for dietary supplements.
- …this announcement from the Digital Medicine Society (DiMe) that it is working on a collaboration – including with the FDA – on integrating sensor-generated data into data platforms to help power clinical research and patient care.
- …these new standards for digital therapeutics developed by the Consumer Technology Association, which it says will help define key terminology, eliminate industry confusion, and establish a “common vocabulary.”
- …this report looking at industry reaction to the FDA’s hearing aid proposals.
- …this question raised by an announcement today: Will the FDA’s Oncologic Drugs Advisory Committee actually meet next week? The committee plans to review Pepaxto, a drug being developed by Oncopeptides AB. However, the company today announced that it has withdrawn its drug from the market.
- …this report that three Senate Democrats are backing a plan that would have the FDA require mandatory prescriber education for opioids.
- …this FDA rejection of a petition by an anti-vaccine group that had sought the withdrawal of Emergency Use Authorization for Moderna and Janssen’s vaccine. The shorter version of the FDA’s response; “we conclude that the CP does not contain facts demonstrating any reasonable grounds for the requested action.”
- …this newly-announced workshop on bioequivalence for long-acting injectable and implantable drug products.
- …this upcoming free event discussing ways to accelerate access to non-pandemic diseases. As the prompt asks: “Where’s our Operation Warp Speed?”
- …this report that Agenus has pulled a cervical cancer application for approval at the request of the FDA.
- …this petition from BioDelivery Sciences International seeking to complicate the FDA’s review of generic versions of Belbuca, a buccal buprenorphine product. In our experience, the FDA almost always denies these petitions, though it’s possible that they raise issues that are then incorporated into the review of the product in other ways.
- …this Federal Register notice from the FDA about the withdrawal of approval of 216 generic drugs due to the failure of companies to submit required annual reports. That’s usually a good sign that the drugs weren’t being marketed in the first place.
- …this FDA notice clearing the way for generic versions of previously-marketed versions of Robaxin, a drug used to treat the symptoms of muscle spasms.
- …this recall announcement for Merck’s CUBICIN due to the presence of glass particulate matter.
Analysis available exclusively to AgencyIQ subscribers:
The FDA this week denied a petition from Stryker Medical to exempt powered patient transport devices from the 510(k) premarket notification process, citing a lack of relevance in much of the data presented in the company’s petition.
Manufacturers using the transitional provisions of the new E.U. Medical Device Regulation (MDR) and operating under Medical Device Directive certificates through 2024 still need to meet some of the new MDR requirements. Even devices that were placed on the European market before May 26, 2021 have a few MDR duties. A new document from the Medical Device Coordination Group offers “legally defendable and pragmatic” advice on how to apply the new MDR requirements to “legacy” devices.
Earlier this month the International Council for Harmonisation (ICH) announced the release of its revised E8 guideline, “General Considerations for Clinical Studies.” The document is now entering the final phase of development in which it is adopted by regulators, including the FDA and EMA.
Late last week, the FDA approved the first interchangeable biosimilar monoclonal antibody. Cyltezo, a biosimilar to AbbVie’s adalimumab (Humira). However, the biologic won’t enter the U.S. market until 2023 due to agreements highlighting the issues around the monopoly created by AbbVie’s Humira patent thicket, and there are several pending developments that may further complicate the regulatory landscape.
The House of Representatives this week advanced a bipartisan bill to allow the FDA to establish National Centers of Excellence in advanced and continuous manufacturing process and fund collaborative research.
