Flatiron Exec Describes Broadening Scope of Research Using Real-World Data
Flatiron Health, which has built up an oncology-focused EHR network and a de-identified database with more than 4 million patient records, recently announced a milestone of over 1,000 research publications using its EHR-derived real-world data. Neal Meropol, M.D., the company’s vice president of research oncology, recently spoke with Healthcare Innovation about the expansion of the types of research being done with its data.
Founded in 2012, Flatiron is an independent affiliate of the Roche Group. It collaborates with partners in academic institutions, policy groups, regulators, and biopharmaceutical companies.
A medical oncologist, clinical investigator, and outcomes researcher, Meropol oversees Flatiron’s clinical teams supporting retrospective and prospective evidence generation.
Healthcare Innovation: At Healthcare Innovation, we focus on writing about the healthcare provider side and not so much about life science companies or drug development. But obviously the two worlds are linked together, and Flatiron’s work is one of the places where that happens. You've been at Flatiron for seven years. What are some of the significant changes you've seen in terms of the company’s capabilities?
Meropol: In a lot of ways, it's been remarkable, but all within the context of a consistent mission, which has been to improve and extend lives by learning from the experience of every person with cancer. The way we do this has largely remained unchanged over this time, and that's through providing software and services to practicing oncologists, where we're an EHR vendor, but also by curating data largely derived from electronic health records that's collected during routine care to gain insights that help address care quality and policy, and that help accelerate drug development and contribute to a learning healthcare system.
HCI: So what has changed over the past five years or so?
Meropol: When we first started, our data were most commonly used for market tracking — how are new treatments being adopted? So the customers, in a way, were commercial arms in biopharma. But we've evolved to have a much broader scope of research areas that our data are now applied to. That includes academia, government, NCI and regulators, nonprofits like the American Cancer Society and Friends of Cancer Research. Having a broader array of consumers of Flatiron real-world data opens up avenues to all different types of of applications. Also, we're now no longer just dependent on the data in the electronic health record; we now have linkages to other external sources of data, like mortality databases, claims data, digital images, genomic data. That really opened the door to lots of different types of insights that we’re able to glean.
HCI: Does broadening out the types of research include things like looking at health equity, at how people are receiving care in lots of different types of settings?
Meropol: Exactly. The applications of our data today not only include understanding adoption of new therapies, but also understanding the real-world outcomes associated with different patterns of care, being able to perform comparative effectiveness, to understand differences in approaches that might not be amenable to doing a prospective clinical trial.
We've done work with what are called natural experiments — looking at the impact of drug shortages on healthcare delivery and the outcomes of patients, and also looking at national policies — for example, expansion of Medicaid and how that impacts cancer care. We measure quality with Flatiron data, looking at how people are using biomarker testing, and whether patients are receiving care directed by biomarkers, and how patients are being treated at the end of life.
We've looked at many different areas of health equity, including how quickly patients initiate care, including whether adoption of new innovations is equal across different groups, whether where you receive care, whether it's an academic setting or community setting, impacts the care you receive. The data are just very rich in terms of what is possible.
HCI: Has the number of oncology practices or academic medical centers using Flatiron’s OncoEMR or its other tools continued to grow? And does the OncoEMR continue to evolve with new features?
Meropol: Our footprint in community oncology and in academia continues to grow. In terms of new features in OncoEMR, something that's really notable is we have over the past several years developed decision support tools. We have an application called Flatiron Assist, which helps guide the appropriate selection of treatment based on patient characteristics, and this has been rapidly adopted by our OncoEMR practices in an effort for the users to improve their quality of care and streamline their decision making.
HCI: Usually, we hear from academic medical centers that creating and updating clinical decision support rules is a complex governance task. So is that something that you have to get input from the network on?
Meropol: Two things: One aspect is that out of the box, the Flatiron Assist solution is built on National Comprehensive Cancer Network clinical practice guidelines. The second feature is that it's highly customizable, so practices can choose to customize the recommendations of Flatiron Assist based on their own practice preferences.
HCI: What about adding these AI scribe tools to the EHR as well? Is that something Flatiron is working on?
Meropol: Yes, we've facilitated the testing of ambient AI for practices that want to go in that direction, absolutely.
Also, one of the new areas which is really critical is that in order to leverage data on all of the 4 million cancer patients we follow, we have to find more efficient ways to curate unstructured information, and we're now routinely applying machine learning and artificial intelligence to assist with curation where the performance is high-quality, so that we can expand the number of patients that we're able to have meaningful information about.
HCI: Could the ambient AI tool become smart enough to figure out what to put in the structured data and what to put in the narrative?
Meropol: There's a lot of discussion happening about what the electronic health record of the future will look like. Historically electronic health records were not built for research, but if there are tools that enable better capture of information at the point of care in structured or usable format or applications that include clinical care billing, but also research applications, that would be good for everyone and very enabling in terms of advancing the quality of cancer care. So I think everybody is looking towards a future where that becomes more possible, where that becomes a reality, but it's not going to happen overnight, because EHR vendors like Flatiron are really focused on workflows and ensuring that we are helping clinicians provide the most efficient and high-quality care possible. And there's always a risk that implementation of new technologies causes increased clicks at the computer, and results in more burnout and has unintended consequences.
HCI: Are the customers who are working in community oncology practices different from the people working in the academic medical centers as far as the toolsets they want or need?
Meropol: The common drivers in community oncology settings and academic settings are that they are trying to provide the highest quality care to patients, but a major driver for clinicians in an academic setting is professional advancement. They might be more interested in taking part in research and publishing results, but I should note that Flatiron’s community oncology network has many practices in it that conduct clinical trials research and are heavily motivated to do so. So the incentives and the desires to be able to provide cutting edge care in the form of clinical trials exists both in the community and in the academic center. And one of the things that that we've really built out over the past several years is the provision of software and services to community oncologists to better enable them to take part in clinical trials. That includes helping with clinical trial recruitment patient matching, but also with the transfer of data from the electronic health record to study databases. So we have a tool called Clinical Pipe that is a data transfer tool for clinical trials, which reduces site burden and enables more sites to participate in more studies than they might otherwise have been able to do.
HCI: I remember that there was an announcement of a Flatiron partnership with the Association of Cancer Care Centers to work on broadening access to that tool.
Meropol: Community oncologists, in fact, are often very interested in building out their clinical research capabilities, but the investments in clinical research in an academic center and in the community practice may be very different. What we're trying to do is help support the infrastructure of conducting research in routine care settings. So we're trying to align clinical studies with clinical care, and in doing so, make it more of a routine approach to how patients are cared for.
HCI: Do you want to talk a bit about your clinical trial network?
Meropol: At the highest level, we realize that data collected in routine care — what we consider retrospective real-world data — can’t answer all questions in oncology. There are certain research questions, certain evidence needs that require the collection of what we call intentionally collected information that goes beyond what is routinely collected, whether that’s measuring tumor dimensions, or whether it is making sure that the patient's functional status is accurately documented before care is initiated. So Flatiron has built out a business where we will support clinical trials’ prospective data collection, ideally for studies that are pragmatically designed and highly aligned with routine care.
So we're building out the capabilities for conducting clinical studies within the community setting. That includes working with sponsors on the design of studies, using Flatiron data to help inform study design, supporting patient identification for studies, and then supporting the data collection that we've talked about. We believe that we have an opportunity to decrease the complexity of clinical trials that exist today through pragmatic study design, but also pragmatic operational design that will broaden accessibility to clinical trials for cancer patients.