All Times EDT
Keywords: hybrid control, EHR, weighting, clinical trial
Clinical trials with a hybrid control arm, a control arm constructed using a combination of patients randomized to receive the control treatment and real-world data on patients receiving the control treatment in standard clinical practice, have the potential to increase efficiency and decrease cost of randomized trials. Recent work on hybrid controlled trials has leveraged methods for incorporating historical data into the control arm of a clinical trial. However, data for hybrid control arms derived from electronic health records (EHR) may feature additional or different challenges compared to those encountered when using historical data. For instance, it may be difficult to fully harmonize inclusion/exclusion criteria between randomized and EHR-derived controls, patient characteristics may differ systematically between the two groups, and completeness of outcome ascertainment may also be poorer in the EHR setting. To investigate the relative performance of alternative methods for hybrid controlled trials in the presence of these real-world data challenges we conducted a series of simulation studies and provide guidance on the most promising approaches. Given the complexity of real-world data, use of appropriate methodology is key to obtaining valid results.