Real-world data (RWD) derived from electronic health records (EHR) can be used to create external comparator cohorts. This exploratory study assessed whether EHR-derived patient cohorts can emulate the control arms of published clinical trials that supported FDA approvals of anticancer therapies across multiple tumor types. Researchers evaluated the impact of specific real‐world cohort construction analytic decisions on observed endpoints and found that results were variable depending on specific analytic decisions
Researchers first identified 15 key regulatory enabling trials and interrogated multiple large, EHR-derived disease-specific databases to determine the feasibility of emulating the control arm of these clinical trials using “off-the-shelf” RWD. They found that clinical trial control arm emulation with RWD was feasible and, as shown in the post-hoc study, step-wise analytic choices influenced the closeness of real world and clinical trial control arm outcomes. The results of this exploratory work also identify trial eligibility and clinical trial cohort factors that may critically impact the variability and fidelity of results from this type of exercise without customization of datasets and visibility into patient level data in the trial cohort.
Why this matters
There have been numerous recent reports on the utilization of RWD to complement and contextualize evidence from clinical trials. This study takes an expansive approach to address this important question.
This exploratory study is important because it is among the first to demonstrate the potential for EHR-derived RWD as a source of external comparators for single-arm oncology clinical trials. In addition, this work pushes the field of RWE science by helping to pave the way for a systematic and transparent methodological approach to clinical trial control arm emulation using RWD. This progress is a critical step toward integrating evidence from disparate sources to support rigorous and efficient efforts to bring better treatment options to cancer patients.