Collecting data in clinical research and drug development is important, but what matters is collecting the right data, using it well, and making connections that ultimately provide the greatest benefit to patients. Michelle Hoiseth, chief data officer with Parexel, discussed data linking with Outsourcing-Pharma and how it can elevate results.
OSP: Could you please talk about the evolution of data technologies, and how that’s impacted the clinical research industry?
MH: The use of data technology in the clinical research industry has been rapidly evolving. For example, decentralized clinical trials (DCTs) have transformed the way we think about conducting clinical trials. As an industry, we’re in the throes of evaluating our experience with DCT components over the last 18 months and determining what we keep, what we discard, what we monitor, or further mature.
Industry groups such as the Association of Clinical Research Organizations (ACRO) have developed quality-by-design manuals and data flow maps to help us standardize our use of DCT advances. Regulatory bodies are collaborating with industry to try to understand the impact of rapidly increasing the use of remote methods in response to COVID.
With more than 80% of our studies now including components of DCTs, what’s next? We need to keep learning what to use where. We also need to move from thinking about DCT technologies as point solutions to thinking about the overall strategy for patient data for a given indication or product and how, on the back end, we will leverage these data so we can derive more value from the effort to create each patient data point.
DCTs are just one example of the evolution of data technologies in our industry, and I’m excited to see what new technologies are on the horizon.
OSP: Industry professionals that OSP has spoken with have hinted that the research industry has become increasingly efficient and effective at gathering data, but maybe less so at putting those mountains of data to use. Could you talk about that?
MH: It has become fairly easy for companies to collect all sorts of data, assemble them and then extract specific learnings from that data. What’s more difficult, however, is contextualizing that data and figuring out how to use data in a way that addresses a range of clinical questions more productively with greater benefit to patients.
This is where data linking becomes important. When we seek to understand a patient’s journey, we need data from multiple times and sources in patients’ lives, stitched together, so that we can have a more complete view, whether that be their experience over time, or a cross-sectional view of their interactions with healthcare providers, wearables, clinical studies, pharmacies, and payers.
OSP: Please share why a gap between data collected and data utilized might be a problem or at least represent missed opportunities.
MH: Gaps between data collected and data utilized can result in missed opportunities for us to better understand the patient experience. Gaps can include incompleteness, errors or other quality issues, or a lack of understanding of the processes in play that generated the data. If we jump to utilizing data before fully understanding all the data we’ve collected, what the data is and what types of uses it is best suited for, we can easily miss key learnings, or, worse, draw erroneous conclusions.
OSP: Could you please explain what data linking is, and how it works in the context of clinical research and drug development efforts?
MH: One of the most significant limitations to the use of any patient data set — whether it be from a randomized clinical trial or from the real world — is that it exists in isolation. A complete data set to support a broader array of life science use cases requires linking of data from different sources on the patient level.
Keeping our health data private is also paramount, which generally requires that we remove personal identifiers when we work with healthcare data. When we remove the personal identifiers, we lose the ability to match the data on a patient from one source to data on that patient from another source.
Tokenization is a process through which sensitive or personally identifiable data are replaced with a non-sensitive code that does not have meaning in and of itself. Tokenizing data allows for accurate linking and is a core component in creating data interoperability for us to use a broad range of patient data sources to support the kinds of analyses needed in the development of therapies.
OSP: How can data linking be harnessed to improve the usefulness of research to patients?
MH: Data linking allows us to meet patients where they are in their treatment to seamlessly engage them in clinical research, A massive opportunity that we have is being able to understand patients’ journeys through their disease up until the point in enrolling in research, in the research program itself and then after as they return to standard of care. Data linking allows us to link the data from the healthcare systems to the study data and back again to give us a more complete view of the patient.
This approach to developing a patient data asset allows us to ask multiple clinical questions of the data and avoid conducting additional discrete studies. From a patient-centricity perspective, it potentially addresses two very important issues that patients have shared with us:
- Reduces the burden of bringing medical history and records to an investigator – A huge burden in some indications such as oncology
- Allows patients to bring some of their data and experience back as they return to care, which they often describe as a ‘research cliff’ or limiting the value of their participation.
OSP: Do you have any examples of projects where data linking served to make a difference in a project?
MH: Two examples come quickly to mind. In the first, patients are opting in to a patient community and providing information about their current disease status, their treatment regimen, and their quality of life. These data are then linked with other data to look at the burden of disease and healthcare resource impact. Traditional sources of aggregated, de-identified data were not enough to address the questions. Details from the patient were needed to tell the full story. In the second, EMR and EDC data are being combined on a standard of care control arm to reduce the burden on sites to enroll and collect these data from the patient prospectively.
OSP: Please share ways in which Parexel (and companies serving clinical trial teams) might help clients put data linking to work.
MH: Parexel and other CROs can partner with customers to consider protocol designs where appropriate that leverage RWD vs. defaulting to a traditional EDC-only patient data acquisition approach.
OSP: Do you have anything to add?
MH: Going forward, I see our industry being able to move faster with better insights through development of clinical research roles and processes working from fewer sources of patient data. Doing so will allow us to potentially reduce steps and technologies, apply advances such as AI to model or detect patterns or signals sooner, among other benefits.