With the FDA’s introduction of new guidelines surrounding the use of real-world evidence (RWE) in medical device regulatory decisions, FDA Commissioner Scott Gottlieb advances the argument for the utility of RWE. In fact, the FDA is currently considering the role of RWE in evaluating pharmaceutical treatments. Despite much debate over what part RWE should play in regulatory approval and payer coverage decisions, many of us are left asking two fundamental questions: 1) Why do we need RWE?; and 2) If RWE is needed, how can we use it?
That is the opening paragraph of my article in Drug Discovery and Development last week. If you want to know the answers to these two questions, do read the whole thing.
The FDA’s recent finalization of the guidance on using real-world evidence (RWE) for medical device regulatory decisions, was a statement. It was a nod to the untapped potential of RWE and support to advance its use. While RWE will not supplant clinical trials, it is coming and it has “real” potential to transform the traditional paradigm of healthcare innovation. In fact, RWE could take us a step closer to making healthcare more personalized, and thereby more effective. This potentially transformative force for U.S. health care deserves discussion and insight into future directions, particularly how to approach using it.
What is the definition of RWE? The definition of RWE is still evolving, and at this point efforts to narrow the definition should be avoided. Most commonly it is associated with data derived from medical practice on a heterogeneous patient population in real life practice settings, e.g. insurance claims and electronic health records data. It is generated from multiple types of data, collected from multiple sources, amalgamated, and shared across a multiple of data holders in the healthcare system. Moreover, not all of it is outcomes data, as the term is stretched to include patient socioeconomic, environmental, and genomic data.
But RWE is not simply the extraction of data from claims, patient registries, or medical devices. It involves data curation, validations, and standardization. Furthermore, RWE is not merely passive-collection of observational data and retrospective analytic approaches. It allows for prospective capture of a wider range of clinical data. It should also be noted that “RWE is not just “Big Data” – its the integration of multiple sources of data” (1). Regardless, whatever consensus is eventually reached regarding the definition, the promise is more effective health care that ideally is tailored to the needs of individual patients.
Why use it? Real world data (RWD) and RWE has potential application in all phases of the innovation process beginning from clinical research through pre-regulatory and post-regulatory approval. For instance, generating a research hypothesis is a critical step in clinical research. Analysis RWD can expedite this endeavor and even sharpen the focus of clinical research. The use of RWD can even shape the design of RCTs and assist in the recruitment of patients. In the setting of pre-regulatory approval, RWE analysis can support conventional RCT data to provide better insight on the safe and effective use of innovations given the fact that RWD provides the diversity that reflects real-world practice. In the realm of post-regulatory approval, RWE generated from long-term observation can provide important findings related to safety and efficacy that are difficult to identify from the short-term, conventional RCTs with homogenous patient populations. Another potential benefit of RWD and RWE is the ability to provide insight into settings where conducting traditions RCTs are impractical or unethical to conduct, e.g. rare disease drug development or areas of high unmet need. In short, RWE has real use.
How do we use it? The truth is that utilization of RDW and RWE for regulatory purposes is not entirely novel. The FDA has occasionally used RDW and RWE to support new drug approvals in rare disease and areas of high unmet need. It has also been used on a limited number of occasions to support label expansions and revisions such as a new indication or new populations. More routinely, RDW has been used in post-market safety monitoring and surveillance through systems such as Sentinel.
The recent Duke report provided an insightful framework for the regulatory use of RWE. For manufacturers and stakeholders who are considering the use of RWE, there are several considerations for generating RWE fit for regulatory purposes. First, it is imperative to assess the regulatory context (i.e. what specific question is the FDA considering) along with the clinical context. If the clinical question may be reliably addressed by RWE, then an evaluation of the quality of RWD sources must be conducted. Is there minimal missing data? Is the data sufficiently reliable and validated? Additionally, the strength of available study methods must be considered. It is only after matching data sources with appropriate methods for a given clinical question relevant to the regulatory context that This goes to show, that RWD and RWE far from passive. It is an active process, where careful study design, including study protocol and analysis plans, is required prior to accessing data sources.
What are barriers to the use of RWE? A discussion of RWE also warrants a mentioning of the real world barriers that hamper its use and limit the realization of its benefits. First, the data is not collected or formatted for research purposes. The bulk of the data are collected to carry out the daily practice of medicine, billing purposes, or other business objectives. This requires data researchers to perform significant leaps in order acquire data. The inevitable gaps and inconsistencies require the development of novel methods to “clean” the data, which lack acceptance for statistical validity.
Secondly, relatively little of what is considered RWE is derived from more than a single source. Maximizing the potential of RWE will require the collaboration of disparate organizations in disparate industries to integrated multiple sources of data. This brings up a third potentially significant barrier, patient protection, and privacy. Policy regarding patient consent for the use of clinical data remains unclear. In the wake of frequent data breaches in the healthcare industry the will of stakeholders to support and invest in RWE development is certainly undermined.
In the end, in order to actualize the full promise of RWE, data sources must be available for integration and analysis. This requires formats, networks, and methods, which have to be supported and accepted for their validity. At this point in time, RWE lack this widely shared framework that to clarify the appropriate collection, integration, analysis, and use of data.
Fortunately, the interest of many players in the healthcare system, from manufacturers to payers, in using RWE has been growing. Additional support is drawn from a vigorously promoted movement by patient advocates, clinical leaders, the NIH, and CMS towards open data and open science.
As the U.S. health system continues to push for innovation, efforts to advance the use of RWE will expand. With it, current barriers to RWE will be significantly lifted through concerted efforts. Motivated stakeholders eager to tap into maturing RWD will explore the application of these data sources to support regulatory decision making. In this endeavor, it will behoove stakeholders to explore within a regulatory decision-making framework as at the one proposed to ensure an RWE approach is sensible.