Development of an External Control Arm Using Electronic Health Record-Based Real-World Data to Evaluate the Efficacy of COVID-19 Treatment

To protect people from severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2) infection, tremendous research efforts have been made toward coronavirus disease 19 (COVID-19) treatment development. Externally controlled trials (ECTs) may help reduce their development time. To evaluate whether ECT using real-world data (RWD) of patients with COVID-19 is feasible enough to be used for regulatory decision making, we built an external control arm (ECA) based on RWD as a control arm of a previously conducted randomized controlled trial (RCT), and compared it to the control arm of the RCT. The electronic health record (EHR)-based COVID-19 cohort dataset was used as RWD, and three Adaptive COVID-19 Treatment Trial (ACTT) datasets were used as RCTs. Among the RWD datasets, eligible patients were evaluated as a pool of external control subjects of the ACTT-1, ACTT-2, and ACTT-3 trials, respectively. The ECAs were built using propensity score matching, and the balance of age, sex, and baseline clinical status ordinal scale as covariates between the treatment arms of Asian patients in each ACTT and the pools of external control subjects was assessed before and after 1:1 matching. There was no statistically significant difference in time to recovery between ECAs and the control arms of each ACTT. Among the covariates, the baseline status ordinal score had the greatest influence on the building of ECA. This study demonstrates that ECA based on EHR data of COVID-19 patients could sufficiently replace the control arm of an RCT, and it is expected to help develop new treatments faster in emergency situations, such as the COVID-19 pandemic.