Steps Toward the Use of Real World Evidence

Real World Evidence“Real-world evidence” (RWE) is data on the use and accrued long-term benefits/risks of a drug or device that is gathered outside of a clinical trial — generally after the product has been approved. Clinical trials don’t often reflect the breadth and diversity of patient experience with a particular treatment that can be seen once the product is widely used outside of the trial’s tightly monitored setting. RWE can complement controlled clinical trials to give a more complete picture of the benefits and risks associated with specific therapies. It can also help researchers identify best uses for existing drugs (i.e. optimal dosing, ideal length of treatment, characteristics of patients most likely to benefit or be harmed by the therapy), suggest new indications, and flag rare adverse events.

 

The U.S. Congress and the Food and Drug Administration (FDA) recognized the importance and possible uses of real world data for improving the oversight of new medical products and for better responding to the needs and concerns of patients and physicians, when Congress passed the 21st Century Cures Act and reauthorized the Prescription Drug User Fee Act (PDUFA). However, making use of such data — which may be buried with varying formats in patient electronic health records (EHRs) and insurance claims — is challenging, requiring a validated framework for data collection and a standard data curation process.

 

In July, the non-profit organization Friends of Cancer Research held a public meeting on “The Future Use of Real World Evidence,” convening a variety of stakeholders from the worlds of science, policy, and regulation — patient advocates, regulators, drug developers, and bi-partisan Congressional champions. Their aim: to help determine how endpoints from real world data correlate with overall survival and key indicators of disease burden from prior clinical trials, and to propose a standard approach for data collection and investigation into the long-term value of products, including their safety and effectiveness. The result of the meeting was the white paper: “Establishing a Framework to Evaluate Real-World Evidence.”

 

The report discusses the need to integrate data from such disparate sources as EHRs, clinical decision-making and support records, hospital data, billing and claims databases, patient registries, longitudinal cohort studies, and patient-reported outcomes. Making such data useful will require the validation of methods to aggregate and integrate multiple, often widely varying, data sets. But if effectively achieved, the applications of real-world data could be many — supporting regulatory decision-making, clinical usage, coverage and reimbursement decisions, determinations of therapy effectiveness, pharmacovigilance and adverse-event reporting, as well improving clinical trial design and clinical practice guidelines.

The white paper includes the report of a pilot project: a retrospective analysis of data from EHRs and insurance claims, mined from six large sources of data, with the aim of describing demographic and clinical characteristics of patients with non-small cell lung cancer (NSCLC) who were treated with immune check-point inhibitors. The research team evaluated the ability to generate real-world endpoints in NSCLC, segmented by clinical and demographics, and to assess their performance as surrogate endpoints for overall survival. There were several key challenges in mining the EHR and claims data; deaths were rarely noted in EHR records and measuring whether a patient’s disease had progressed was also difficult. It was also a challenge to determine when a finding from such “noisy” data was good enough to make judgments about clinical practice.

 

However, while further validation is needed, the team demonstrated that they could extract several standardizable endpoints from the EHR and claims data. They also showed that survival of patients assessed through real-world data fell within the range of median overall survival values reported in several immune check-point clinical trials. In addition, they found that efficacy endpoints extracted from EHR and claims data were relatively consistent across a variety of individual patient characteristics, such as age and sex. The findings furthermore showed a correlation between real-world endpoints and overall survival, with no significant variation across age groups.

 

The FDA is now seeking funding to tap into the medical records of 10 million Americans to build a medical data enterprise that would incorporate EHR and other sources to enable the agency to more fully evaluate products in the post-market setting.

 

Given the lack of data standardization within the US health care system, how can health care data be made more interoperable? How can RWE be expanded beyond EHR and claims data to incorporate patient-reported outcomes and health data? How can RWE be used by payers in decision making, especially in the context of an accelerated approval or breakthrough therapy designation? And how important is RWE for the development of value-based payment or outcomes-based payment agreements? We’ll be watching.