Biopharma and healthcare organizations were tapping real-world data to inform their decision-making long before COVID-19 roiled global health. Its value was never in doubt: Using information collected outside controlled clinical environments had always been viewed as a crucial step toward better outcomes, even if such information was often considered secondary to more robust randomized data.
However, the pandemic removed choice from the equation. In the absence of extensive clinical trial data, healthcare stakeholders were forced to consider real-world data from the uptake of vaccines and other therapies. And so it was that real-world data was used to inform full approvals, safety guidance and labeling and booster recommendations.
Since then, industry observers have expressed their hope that real-world data might be leveraged more as a lead component that drives healthcare decisions, versus its traditional and secondary confirmatory role. But two-and-a-half years after the first vaccines were deployed, it’s questionable whether this has come to pass.
The only thing that seems certain is the industry’s real-world data disciples are playing the long game. Indeed, there’s little debate among experts that real-world data has been elevated to a level of importance never seen before.
“Before COVID-19, real-world data was historically viewed as a limited research method, but the pandemic has broadly widened the way we think about it,” says Amy Abernethy, M.D., president of product development and chief medical officer at Verily and a former deputy commissioner at the Food and Drug Administration.
One of the most transformative outcomes of the pandemic was a “revolution in pharmacovigilance” with real-world data at its core, says Amit Dang, M.D., founder and CEO of MarksMan Healthcare, a consulting firm that works with healthcare organizations attempting to optimize their use of real-world data. Abernethy agrees, adding, “We saw through COVID that we can accelerate the time it takes to bring products to market and rely on post-marketing data to evaluate safety.”
Jaime Smith, senior director, global head of scientific data at Parexel, a clinical research organization, notes real-world data gave us the ability to understand the virus’ impact in several ways, notably tracking cases, testing and spread before and after vaccine intervention. “This really galvanized the community around real-world data and brought forward this concept as an important factor in addressing population health outcomes for some sub-populations,” she says.
Furthermore, Dang notes, data-sharing collaborations between healthcare stakeholders have resulted in a shift toward more open data ecosystems, which hold the promise of advancing research, innovation and patient care.
So where are we now? An FDA spokesperson acknowledges the obvious: While the use of real-world data was increasing before the pandemic, COVID-19 accelerated awareness and adoption. The information is now being used to expedite the enrollment of patients into clinical studies and informing study design from endpoint selection to trial duration, the official notes. Another “additional important benefit”: the potential to uncover previously unknown patient/drug interactions.
“We’ve made amazing progress over the last couple years, but in terms of understanding real-world data, the industry is still a work in progress,” says Smith.
As the use of real-world data evolves, companies face a host of associated challenges, such as proper data aggregation and better alignment of regulatory frameworks. “Many companies don’t want to be the next — or first — product to take the path that diverges significantly from traditional clinical trial settings, especially when they haven’t fully harnessed how to aggregate real-world data in the right way,” explains Francesco Lucarelli, chief commercial officer at Boundless Life Sciences Group.
These challenges aren’t easy ones to surmount. Among myriad other concerns, the pandemic revealed the complexity of assembling health data from different sources.
“During my time at the FDA, while we had many data sets coming our way, they were often not high-quality enough to use for evaluation purposes,” Abernethy notes, adding that traditional claims and EHR data are often not collected for research purposes and these sources are often incomplete.
Inaccurate or incomplete data can obviously lead to erroneous conclusions and misguided interventions, Dang says. That’s why he stresses that rigorous data validation processes, standardized collection methods and continuous quality assessments are essential to ensure the integrity of real-world data.
Abernethy, for her part, believes that one way to address this is by using techniques such as data linkage, which combines multiple data sets and then fills in the gaps that may exist between data sets. Verily is working on just such an approach with several of its life sciences partners.
Dang notes that robust statistical techniques, such as propensity score matching, can help adjust for selection bias and enhance the generalizability of results, while sensitivity analyses can disentangle confounding factors from true associations. “Patient selection bias and confounding variables are other concerns common in observational studies that can distort findings,” he adds.
As for the FDA, when asked via email about source-related challenges, the spokesperson responded: “In regards to real-world data sources, we find issues related to data reliability and clinical relevance, a need for linkage to other data sources, missing or ‘mistimed’ data and insufficient capture of endpoints. For non-randomized study designs, issues include the threat of residual confounding, problems with index date (‘zero time’) and use of an inappropriate comparator.”
Then there’s the ongoing (and nearly complete) shift to digital. Adoption of health IT, Smith stresses, is at the crux of how we use real-world data. Dang adds that telemedicine, digital therapeutics and remote patient monitoring have become an integral facet of healthcare delivery.
While the pandemic spurred the rapid adoption of many innovative digital healthcare solutions, not all health systems and providers have been able to keep pace. From an ex-U.S. perspective, real-world data are often unable to move beyond local lines. The infrastructure to support EHRs in some countries is less robust than what is seen in the U.S.
For example, some rural areas in the APAC region may not have the same connectivity; as a result, they still keep paper trails. This becomes a headache when attempting to collect uniform data on a global scale.
To further convolute the concept of data digitization, ethical considerations “loom large,” Dang says, when handling patient data. He believes that building trust among patients and stakeholders hinges on ethically sound data practices that include anonymization protocols, adherence to data protection regulations and transparent governance frameworks.
The regulatory-alignment piece of the puzzle is similarly vexing. Smith says that while real-world data had already made its way into many global regulatory guidelines prior to the pandemic, there is a “movement” toward better understanding real-world data in a clinical trials context.
Dang concurs, noting the use of real-world data is common in pharmacovigilance but less established in earlier stages of drug development.
Starting in 2021, various FDA guidelines published on the use of real-world data pointed to its future use for drug approvals. In these documents, the agency highlighted what it would be looking for, including high data quality, methods for linking between disparate data sets and statistically robust methods for analyzing real world data.
While different study designs hold varying strengths and limitations, the agency is “open,” its spokesperson says, to all study designs when reviewing submissions. This may include randomized trials, externally controlled trials and noninterventional studies, though the spokesperson notes the possibility of methodological bias in the latter option.
There have also been payer initiatives that emphasize the importance of real-world data as a component of the evidence used to inform coverage and payment. The Centers for Medicare and Medicaid Services is in the process of updating its “Coverage with Evidence Development” program to take greater advantage of real-world data in the context of “fit for purpose” study designs, Abernethy notes.
Change is likely to arrive sooner than later. To that end, Smith says regulators are in open discussions about the use, safety and security of artificial intelligence and machine learning, and especially how the industry should be thinking about the inevitably complicated evolution of future data sources.
As for what comes next, the pandemic revealed the opportunity to use real-world data to fill in the gaps and achieve a “totality of the evidence” approach, Abernethy says. This means a reliance on both clinical trial data and real-world data to inform regulatory decisions. The FDA recently launched its Advancing RWE Program, intended for discussion of very early drug development proposals with a focus on novel and creative uses of real-world evidence that are likely to meet regulatory requirements.
Both Smith and Lucarelli note proactive steps by biopharma to increase the use of AI in the application of real-world data. This would help determine optimal clinical trial designs, inclusion criteria and treatment settings. Smith notes that it will likely take an interdisciplinary effort — bringing together health informaticists, AI/machine learning experts and individuals from traditional clinical trial backgrounds — to truly realize the benefits of this approach.
Meanwhile, Abernethy says Verily is partnering with health systems and sponsors to use its tools and infrastructure to build longitudinal registries of patients who have consented for their data to be used in research. Data from EHR records, sensor-based data collection and patient reported outcomes can be brought together to conduct different types of studies, including randomized clinical trials, but also novel study designs using real-world data.
“We are focused on leveraging real-world data sources while also making sure that the data captured meets the needs of regulators and other stakeholders,” she adds.