JULY 2021, PERSPECTIVE PAPER
Vice President and General Manager
Real-World Data Solutions
Evidera, a PPD business
Use of real-world evidence (RWE), which is the clinical evidence derived from analysis of real-world data (RWD), has been increasing in recent years as regulators and healthcare payers have grappled with understanding how therapeutics and drugs perform in real-world settings. Sourced from electronic medical records (EMRs) and other resources, RWD can be enriched by leveraging the prescribing physician’s overall experience, thus providing greater insights, flexibility, efficiencies, and cost savings.
Real-World Data and Real-World Evidence
The use of both RWE and RWD has been increasing in recent years as the industry’s awareness of the potential benefits offered by leveraging this information expands. There is growing realization that locating and analyzing RWD will lead to insights about drug performance and provide a competitive advantage.
To understand RWE, it is first important to understand RWD. RWD is accumulated in many forms by healthcare institutions via EMRs, medical claims, products and disease registries, laboratory results, and, more recently, sensors and health and fitness apps. RWE is the clinical evidence derived from the analysis of this RWD collected outside of clinical trials.
Growing Regulatory Recognition of Real-World Evidence
It is an exciting time in the field of RWE, as regulatory agencies, along with biopharmaceutical companies, are exploring greater adoption of RWE to improve a wide range of outcomes. The U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) have issued support and frameworks for advancing the use of RWD and RWE to improve regulatory decisions. They believe that RWD can complement and augment investigations into how to best use medical products, with significant utility in post-market monitoring.
The FDA’s Project Renewal is one such example. This program was established by the agency’s Oncology Center of Excellence (OCE) to update the labeling information for oncology products by evaluating relevant scientific evidence from published literature. A set of repeatable processes and procedures has been established to assist regulatory decisions for oncology product labeling updates, including potential new indications for use. The intent, according to an article published in 2020 in The Oncologist, is to “leverage the breadth of knowledge the cancer community has accumulated with the use of these drugs over time, often spanning decades of experience.”1
While RWE is not yet accepted by regulators to nearly the same degree as clinical trial data, the field is evolving at light speed. What will be critical for greater acceptance and use of RWE is assurance that the results are obtained using high-quality data and collected following protocols demonstrated to be highly rigorous.
Data Explosion Requires Proper Data Management
Data environments are growing exponentially; not only are there more data, but there are more data sources, and data regulation is constantly evolving. At the same time, the value of unlocking that data and using it to make business decisions is also increasing. Understanding these complex environments is key to generating deep, actionable insights.
The first step is to acknowledge that RWD can be vast and fragmented. Because RWD is collected outside of clinical trials, it can comprise very large quantities of information pulled from numerous sources, sometimes without context. As a type of Big Data, RWD requires special preparation owing to its size and this high degree of complexity.
Specifically, analysis of RWD requires investment in systems and resources and establishment of the expertise needed to effectively identify the right questions to ask, to perform data mapping in collaboration with the IT team, and to ensure quality control before any analytics are conducted or output is generated.
As a result, there are three key steps to leveraging RWD for generation of RWE: data preparation and modeling, assurance of data diversification, and data presentation.
Assuring Data Quality and Timeliness
The main challenges to working with the massive data sets from disparate sources that comprise RWD are data quality, missing data, and data timeliness.
High-quality data is appropriate for its intended use and correctly represents the real-world construct that the data describe. To ensure that data are of high quality, their representativeness, robustness, and perspective must all be considered. While a sample of 1,000 women can be very robust, it would not be representative of the general population, as it would lack the representation of male subjects. And while any data set itself has a single set of features, those features can be interpreted differently depending on the perspective of the user. For instance, the numbers 6 and 9 can be seen as the same if they are viewed from the top versus the bottom — contextual information is critical.
Missing data is a prevailing issue in any type of data analysis, and as data sources become more numerous and varied, which is true for RWD, it becomes even more problematic. In addition, RWD was initially collected for specific purposes — think about EMRs — and not in the context of establishing RWE. Fortunately, missing data is not a problem if the analytical methodology is well designed. In fact, the absence of information is itself an important piece of information. Forcing the inclusion of missing data is not recommended either, because it will typically introduce bias into the data set. The key is to understand whether the missing data results from a quality issue or if it is a true reflection of what is happening in the real world.
Speed of data is another important issue when generating RWE based on RWD. Traditional site-based studies offer rich clinical data, but often require manual processes that lengthen timelines and are prone to data entry errors. Technology-driven data collection offers many advantages, including direct data extraction from the source to minimize data entry errors and decreased data query volumes. Because the data are extracted directly into standard formats conducible to analyses, data collection and cleaning are also streamlined. In addition, automation allows for efficient repeat data extractions.
Key User Needs and Wants for RWD
To generate reliable and robust RWE that affords valuable insights not accessible through the analysis of traditional clinical trial data, RWD must be not only high-quality and timely, but relevant and appropriate to the question(s) being asked.
Flexibility is also essential to enable users to find the data that meet their specific needs for their specific applications, particularly with respect to geographies, patient populations, and data parameters.
In addition, as the use of RWE continues to evolve, there are several factors that could help realize its full potential, including increased availability of high-quality and reliable RWD in more geographies and more diseases, greater data normalization, and the introduction of transparent standards for RWD curation.
Data Abstraction, Not Extraction
The key to meeting user needs is to not just extract data, but to abstract data. Extraction of data can be very useful, because it brings all of the data available in the patient medical record into a data set for analysis. It has the disadvantage, however, of being associated with heavy processes of data curation, data mapping, and data prioritization.
Healthcare professional (HCP) EMR data abstraction offers a more tailored approach. The HCP will transfer specific data from the patient medical record into a different platform. This critical step allows for the collected data to be curated by the prescribing HCP and to be “fit for purpose.” Only the data that is needed is collected, but the HCP can also report behavioral information as to “why” he/she did or did not do something of interest.
Understanding the ”why” is a crucial step in the process, and very often the databases that provide the “what” or the” how” are disconnected from the market research that can provide the reasoning. The advantage of abstraction over just extraction is the addition of the “why” to the collection of data about “what” and “how” a drug or therapeutic was used in a specific way. With Evidera’s approach to data abstraction, we now also have information on what decisions were made in analyzing the data — why physicians made the decisions they did, why they used the medications, when they did, and under what circumstances.
Enormous Advantages Afforded by RWE in the Pre- and Post-Marketing Arena
The RWE generated from such abstracted RWD is not only of significant use to R&D, medical, and health economic outcomes research (HEOR) teams. Commercial teams have also found RWD to have enormous advantages in the pre- and post-marketing arena, such as for surveying the competitive landscape, understanding treatment decisions for specific indications, and better understanding exactly how products are being used in current clinical practice.
RWD has potential use for market research and can also be used to inform product and marketing teams about how a particular drug is being prescribed and used, which in turn can guide commercial and marketing strategies. The insights can also critically inform further clinical development programs.
Value of Global Physician Networks
While there is strong consensus that there are significant opportunities for RWD and RWE use in R&D, data suggest that many organizations have yet to make the most of its potential in this area. Companies are looking for solutions that allow them to streamline cost and synergize databases across various departments.
With Evidera’s LiveTracker®, data are tracked over time so that insights can be gained, not only at specific time points but over a defined period, to better understand disease progression and treatment decisions. In addition, the same panelists (physicians) that abstract the data from their electronic medical records also provide information on the reasons behind their treatment decisions — the “why.” Not only do these studies provide actionable insights, they do so in a very timely manner and eliminate the need to rerun market research programs.
It is important to note that retrospective studies designed to analyze pre-existing data are subject to numerous biases as a result. Biases are not a problem as long as they are under control and taken into consideration. Therefore, the physicians participating in these studies are rigorously screened and profiled before they randomly report patient charts.
Tracking Pembrolizumab Use by Oncologists
One recent LiveTracker program involved tracking the use of first-line programmed death-ligand 1 (PD-L1) inhibitors by oncologists across multiple European countries between January 2019 and June 2020. This observational, retrospective cohort study was based on RWD collected on a monthly basis.
PD-L1 is speculated to play a major role in suppressing the adaptive arm of the immune system. Immunotherapies targeted against PD-L1 and its receptor (PD-1) have shown improved survival in a subset of patients with cancers who have high levels of PD-L1, such as non-small cell lung cancer (NSCLC) and melanoma. Because of this, PD-L1 protein expression has also emerged as a biomarker that predicts which patients are more likely to respond to immunotherapy.
The number of clinical trials testing PD-1/PD-L1 checkpoint inhibitors has tripled from 2017 to 2020, totaling 4,400. This growth was led by a surge of combination immunotherapy and chemotherapy trials; 90% of the new trials started in 2020 were combination strategies.
In this particular program, Evidera’s Real-World Data Solutions team used prescribing data from LiveTracker to determine first-line (1L) and second-line (2L) prescribing trends for the treatment of NSCLC and melanoma based on the level of PD-L1 protein expression. Patient data were continuously collected from our global network of HCPs who report EMRs and treatment decision information monthly into the LiveTracker database.
Analysis of the RWD found that, for HCPs treating NSCLC and melanoma, the label restriction of a PD-L1 inhibitor in NSCLC patients who had PD-L1 expression equal to or above 50% was extrapolated to melanoma, while the label of the product does not show any PD-L1 expression level restriction in melanoma. In addition, it was determined that new immunotherapy asset uptake can plateau in as little as three months post-launch.
In essence, the data showed increased market pressure for 2L treatment of NSCLC due to a higher use of a combination of immunotherapy and chemotherapy in 1L and increased competition in 2L across all treatments. The study also underscored that oncology is a fast-moving market with high competitive pressure and that real-time market reactions are critical to enable the adjustment of promotional investments. Continuously updating treatment data as new combinations and assets enter the PD-L1 market and leveraging these live market readouts to adjust promotional investment is essential.
- Balogh E.P. et al. “Challenges and Opportunities to Updating Prescribing Information for Longstanding Oncology Drugs.” The Oncologist. 25:e405-e411 (2020).