By Ciox Real World Data Leaders
Real-world data is defined by its source. It’s any data that are captured at the point of care, in an actual clinical scenario or based on patient’s experience outside of the clinical setting. But this is not enough for it to be used in various phases of drug development to support evidence that guides decision-making and regulatory processes. Real-world data needs to be converted or translated to inform scientific inquiry, validate existing evidence and support decision-making on medical care interventions. The latter is especially important as it relates to understanding the impact of adverse events or rare side effects when a drug is used in a larger patient population once approved.
Today, the magnitude and reliability of clinical evidence used from data collected outside of the randomized clinical trial process is staggering. These sources open doors by providing more insight into the potential benefits and risks of a medical product. From real-world data, we develop real-world evidence (RWE) that can shed light on which health care services work for different patients and what happens to a patient over their health care journey.
While, in theory, this may seem straightforward, there is no clear or standardized way to convert real-world data to real-world evidence. Further complicating this are the inherent shortcomings of RWD, such as limited fitness when captured at the point of care or whether there are the appropriate data elements to analyze that will support evidence used in research. This inconsistent way of handling a wide range of data makes it difficult to extract and analyze in a clinically meaningful way. These additional barriers to translational research are currently being addressed by the FDA, those in academia, and others in private industry.
The FDA’s Real-World Evidence (RWE) Program
The potential scale and accuracy of data in use today support the recent focus by the FDA on real-world evidence with its inclusion in the 21st Century Act (Cures Act). This legislation was signed into law on December 13, 2016. The Act recognizes the value of how RWE supports the approval of a new drug indication or satisfies post-approval drug study requirements. Specifically, RWE will help with what the Cures Act aims to achieve, which is efficient, accelerated medical product development and approval that bring new innovations and advances to the patients that need them.
Real-World Evidence Relies on Real-World Data
We all know that the strength of evidence depends on solid, well-organized data that are gathered in a consistent way. Real-world evidence relies on real-world data (RWD) or data that reflects information that is gathered outside of a controlled clinical trial setting. It contains important insight on a patient’s health status and the “usual” delivery of health care. A great deal of this insight exists in sources like electronic health records (EHRs), medical claims and billing, image reports and even mobile devices. Non-randomized clinical trial data that define RWD is the driving force behind RWE and the decision-making about appropriate health care delivery. Real-world data includes two data formats—structured and unstructured. The first is developed according to predetermined data structures (or models) and organized in a predefined manner. The second is data that is often messy, heterogeneous, and of varying formats, such as clinical notes from treating providers.
The Digital Conundrum
The digitization of health data, ushered in by advances to electronic health record systems, provides a potential trove of real-world data that can be converted into very valuable real-world evidence. But the disparity in how these systems collect data creates a very broken and siloed data landscape that has grown for almost two decades. For instance, nearly 75 percent of electronic data in the U.S. is “electronic” because it’s communicated through fax machines.1 Furthermore, the average U.S. hospital currently uses 16 disparate electronic health record (EHR) systems; only two percent stick to two or fewer systems.2 This has led to gaps in evidence that often neglect to account for non-digital sources of data. Incomplete documentation also arises from access to care limitations and health disparities that exist, particularly in underserved populations.
To address these gaps and the lack of an established formula when creating real-world evidence, the health industry is taking a more refined approach. By distinguishing syntactical from semantic data, it provides greater context that allows for more complex clinical observations to be taken into account. With syntactical and semantic standards, data can then be primed for both faster computer analysis and human expert validation of information in medical records.
This is an opportunity that the Ciox DataFit PlatformTM is designed to address. It uses technology and deep human expertise to bring disparate data sets together into a structured, analysis-ready format that is fit for medical research. With full custody of the data supply chain, this platform combines expert data model designers with AI techniques, including biomedical Natural Language Processing (bNLP) and Computer Vision, paired with Deep Learning, to provide a continuously customizable digital sandbox filled with the richest and most relevant cohorts assembled from all available structured, unstructured and non-standard patient data.
Industry Perception of Real-World Evidence
From inconsistencies in converting RWD to RWE, there are also conflicting views on what constitutes RWE in the first place. One school of thought views randomized trials as a sufficient way to gain fully curated endpoints with specific, well-defined targets. But this approach runs into obstacles, such as timing, money and limited patient study groups. Others in the industry think a vast amount of valuable, observational research data that completely captures the patient experience in real-world settings is being overlooked and ought to be incorporated in the clinical trial and approval process.
Embracing the Role of Big Data in Health
The RWE Program promoted by the FDA and other industry leaders recognizes the powerful role that data can have in improving health. It also encourages a change in research designs that combines the randomized clinical trial approach with expertly curated research that informs factors such as drug adherence, compliance and socioeconomic impact on health outcomes.
RWE has already shown impact by providing information gathered outside of a clinical trial setting to predict safety risks and demonstrate the benefits of an intervention. By embracing Big Data, drug developers can create solid, dependable data to better understand and predict outcomes of polydrug use, contraindications and co-morbidities in relation to treatment and health outcomes. Safety-wise, the RWE Program allows for clear comparisons to real-world practices; from point of care to other societal factors relating to socioeconomics, lifestyle and behaviors.
We are at a juncture in using health data and creating evidence from RWD that is informed by cross-agency collaboration. The FDA, National Institutes of Health, U.S. Department of Health and Human Services, and Centers for Medicare and Medicaid Services can work with industry to harness extensive collections of real-world data and curated data sets. They have this information at their disposal to refine data quality and elevate RWE standards. This involves insight that is obtained from both structured and unstructured data sources.
The power of data on understanding health care and outcomes is much larger than its early purpose of administrative billing. It can be harnessed for research purposes and translational medicine that is bound to greatly improve patient care by providing better, more individualized treatment options, sooner and with greater efficiency.
1Kliff S. The fax of life: why American medicine still runs on fax machines. Vox Media. Updated January 12, 2018. Accessed November 16, 2020. https://www.vox.com/health-care/2017/10/30/16228054/american-medical-system-fax-machines-why
2Sullivan T. Why EHR data interoperability is such a mess in 3 charts. Healthcare IT News. Published May 16, 2018. Accessed November 16, 2020. https://www.healthcareitnews.com/news/why-ehr-data-interoperability-such-mess-3-charts