Amy Abernethy, MD, PhD, President of Clinical Studies Platforms at Verily, joined our Chief Medical Officer, Matthew Roe, MD, MHS for an invigorating Fireside Chat earlier this month. It reinforced how critical the work that is being done at companies like Verana Health and Verily is to harnessing the full value of real-world evidence (RWE). This conversation came on the heels of the recent release of the draft guidance from the FDA surrounding real-world data (RWD). Before Verily, Dr. Abernethy was the Principal Deputy Commissioner and Acting Chief Information Officer of the FDA, where she initiated several agency-side efforts including FDA’s technology and data modernization action plans and FDA’s efforts to leverage RWD and RWE to address urgent questions during the COVID-19 pandemic. In this Fireside Chat, she shared her perspective on how to deliver the greatest value with RWE, approaches for prospective evidence generation, the evolution of the FDA, and the focus of her work at Verily.
See below for some highlights from their conversation.
Dr. Matthew Roe: Based on your experience, can you please share the challenges you’ve come across in your career with processing and analyzing real-world data to create real-world evidence?
Dr. Amy Abernethy: Back in 2008-2009 at Duke University, we conducted chart reviews on hundreds of patients for colorectal and breast cancer and it was clear to me that we needed to figure out how to do chart reviews at scale. To do this, there were several important things to keep in mind — the first being clinical expertise. We needed, for example, oncology fellows, and others doing this work who could make sense of the medical case notes. A second thing was the importance of cross-checking. One person would conduct the chart review and then a certain proportion of charts were re-reviewed by another fellow for a quality check and for quality improvement to learn ways to improve future reviews.
Later in my career, we tried to figure out how to recapitulate exactly what those fellows were doing, but with adding a combination of software and people. We learned that we could leverage structured data, start to harmonize and normalize it, and clean it up so it was rapidly analyzed, all in one common ontology. But, practically speaking, we still needed really experienced humans to review the clinical notes and unstructured documents, then abstract key data points from those notes, and then develop variables that would fit in the right place in the data sets.
Fast forward to 2020 at FDA, thinking about writing RWE guidance, we needed to consider the variability of the variables’ importance, make sure that we are documenting and focusing upon data quality, and that it’s done so in a way that’s consistent and so that others can understand the importance of linking and filling in data gaps when leveraging data from multiple sources.
Dr. Roe: You’ve shared publicly before how curating real world data into high quality data sets is one of the hardest things that we do in the field, even before generating evidence to inform decision making. What have you learned over your career about how to do this well and what structure or collaborations do we need to do it at scale?
Dr. Abernethy: I think the thing that has been most important is the old adage “it takes a village” which really is true. It takes the software, data, and product collaborators, as well as the clinical, statistical, and quantitative sciences collaborators to come together to solve this problem.
One of the challenges that gets in the way a lot of times is that each of those disciplines comes with its own approach to frankly, who’s in charge. Whose voice is the winner or loudest? In my experience, the answers are almost always in the middle and the biggest thing to solve is actually how do you get all those different collaborators to listen to each other as equals at the table. The group needs to figure out how to listen, how to collaborate, and also how to build a glossary or a lingua franca so that the same words mean the same things to the whole group.
Dr. Roe: Absolutely, I think that’s what the latest FDA guidance is focused on – creating a playbook and shared understanding of data sources and processes being used. We’ve discussed data documented during routine care delivery, most of which is looked at retrospectively. In contrast, I’m interested in your thoughts about how the field is shifting toward prospectively collected data and evidence generation, with data collected from both patients and clinicians.
Dr. Abernethy: A huge piece of prospective data collection is including a way for a patient to provide informed consent but also to enable this to be done while clinicians are engaged in their regular interactions with their patients.
This may include permission to pull in data from passive sources like electronic health records (EHRs) or the ability to interact with them and collect new data points. In order to accomplish this and meet participants where they are, state-of-the-art software is going to be really important. In the pandemic, we talked a lot about decentralized clinical trials, telehealth virtual visits and the ability to leverage software to not only interact with people remotely, but also to get permission for data collection today and permission for interaction in the future. I think we’re going to see registries or communities of people who’ve already said you can talk to me again in the future and are predisposed to participate in research and participate in routine data collection of the different types. So the question is how do we make it more commonplace to ask for permission to participate in prospective research in routine practice conditions? I think we’re moving from a landscape where you signed a form to join a clinical trial in a separate research office during a separate research-only visit, to a true understanding and partnership with people across time as they contribute their data legacy to the future of healthcare. There has already been huge technological progress towards this goal to collect patient data at scale.
Dr. Roe: Speaking of technology, you’re well known for launching technology modernization plans during your time at the FDA. Can you share a bit more about why those plans were launched and how those plans bridge into where the FDA is going with real-world evidence?
Dr. Abernethy: When I arrived at the FDA in February of 2019, I said I was going to focus on three things: precision medicine, patient centricity, and RWD/RWE. Once I started talking to everyone about those three things, it became apparent that they all have the same features, which is an increasing number of data points and data sets, increasing complexity of analyses and the expectation that the FDA was going to be able to do the work. It was clear we needed a way to bring data in and rapidly analyze it and to do so, we needed to assemble strategies, teams and budgets to push forward technology modernization to achieve any of these things.
In September 2019, we announced the technology modernization action plan which said, as the FDA, we were going to take a cloud-forward stance, and we are going to start to leverage APIs and other mechanisms to bring data into the agency. Quickly thereafter, the pandemic hit and we were able to turn into a virtual agency overnight with 18,000 employees because of that infrastructure. Once we had that virtual infrastructure, we started to think about what it meant to leverage data and technology in new ways for regulatory applications. The data modernization action plan was announced just before I left, which included how the FDA would manage or handle data, especially structured data, in new ways. Having the technical skill sets internally to do that will help the FDA to even think through how to manage the ever enlarging real world data sets that will be available in the future.
Dr. Roe: Several months ago you left the FDA and transitioned to Verily. Can you please describe your vision for the work you’re focused on at Verily.
Dr. Abernethy: Verily’s heritage is in the Baseline Health Study, a longitudinal cohort of individuals committed to providing a 360 degree view of their health with deeply characterized in-person visits and data collected through digital tools. This serves as a starting point to plan for our future and allows us to think about how we can build software that reduces the burden of participating in research to make consenting, virtual visits and patient-reported data collection easier. The second thing is to pair that with reducing the burden of being a study coordinator at a study site, which is why acquiring SignalPath was so important. We’re also working on building longitudinal datasets. By linking data from many different data partners, including possibly Verana Health in the future, we’re exploring how we develop solutions that bring increased value to different data sets. Lastly, we’re thinking about what clinical study designs will look like in the future that we want to help enable.