OBJECTIVES: Prior studies have attempted to ascertain the feasibility of using EHR to emulate data elements necessary in clinical trials. However, few studies have focused on assessing the utility of both structured and unstructured (i.e., abstracted) data from oncology specific EHR systems for such purposes. This study examined critical elements of recent oncology clinical trials to assess the degree to which these real-world data (RWD) can be reasonably used to retrospectively replicate the information needed in trials.

METHODS: FDA approvals of oncology drugs in 2020 were identified and matched to trial data from the Aggregate Analysis of ClinicalTrials.gov (AACT) database (downloaded 12/17/2021). Inclusion and exclusion criteria in these trials were tabulated from AACT, and the degree to which they could be ascertained as structured data (available in a database, such as age, sex, diagnosis, stage, treatment, biomarkers, etc.) and/or unstructured data (such as free text fields, scanned documents, outside diagnostic reports, etc.) in an oncology specific EMR system were qualitatively assessed by experienced investigators.

RESULTS: Among the 53 trials identified, 20 were phase III studies and had necessary information available in AACT. Of the 459 eligibility criteria reviewed, the median per study was 21 (range 7-51). Median inclusion was 8.5 (range 2-31) and exclusion was 12 (range 2-25). Overall, 81% of eligibility criteria was likely available in the oncology EHR (50% in structured and/or unstructured and 31% in unstructured only). Approximately 19% of criteria was not generally available in EHRs outside of a clinical trial (e.g., signed informed consent, agreements for use of birth control, etc.)

CONCLUSIONS: Oncology-specific EHR data can be leveraged to emulate approximately 80% of inclusion/exclusion criteria of clinical trials. However, it is important to differentiate eligibility criteria that are available as part of real-world patient care versus those applied specifically for the purpose of a trial.