OBJECTIVES: A control group of clinical trial using real-world data (RWD) is desirable and ethically sound. Challenges include internal validity. As the first step to verify the internal validity, we tried to reproduce the control groups in the past two RCTs using RWD and evaluate the difference of the outcomes between them on each of the two RCTs.

METHODS: We selected the KEYNOTE-426 phase III trial of pembrolizumab for renal cell carcinoma and VIALE-A phase III trials of venetoclax for acute myelogenous leukemia. Overall survival (OS) was one of the endpoints of all the two trials. We reproduced the control groups using health insurance claims data, one of RWD, provided by DeSC Healthcare Inc. Kaplan–Meier OS curves reproduced by RWD and the control group of original RCTs were compared. The survival time in the KEYNOTE-426 trial were examined at the point of 60%.

RESULTS: OS curves of original control groups were between upper and lower limits of 95% confidence interval (CI) of OS curves of reproduced control groups in the KEYNOTE-426 trial. The 60% survival time in the reproduced and original control groups in the KEYNOTE-426 trial was 26.1 months (95% CI;18.8-32.7) and 30 months. In the VIAL-A trial, the OS curve of original control group was not between upper and lower limits of 95% CI of OS curve of the reproduced one.

CONCLUSIONS: Though significant difference was found between original OS curves and those produced by RWD in the VIAL-A trial, it was not found in the KEYNOTE-426 trial. Furthers studies are needed to judge the credibility of the control group using RWD.