Drug Evaluation Committee Metrics for Evaluating Survival-Time-Type Responses (Version 2) Understanding RMST (restricted mean survival time)
Data Science Expert Committee
March 2019
The Drug Evaluation Committee's Data Science Subcommittee, Task Force 4, 2017-18, summarized the evaluation metrics for survival time-type responses.
In recent years, clinical journals have begun to discuss the use of restricted mean survival time (RMST) as one alternative measure when proportional hazards nature is not valid. However, there is no book that focuses on RMST, and we thought that a practical study of its nature is necessary for its actual application.
This report explanation includes information not only on the theoretical aspects of RMST, but also on the practical aspects such as programs. In addition, since a certain level of understanding by clinical personnel is also necessary when selecting an evaluation indicator that is considered appropriate for the situation, we have organized each evaluation indicator in Chapter 2 using diagrams so that it can be intuitively understood Not limited to RMST, when interpreting the results of clinical trials, it is important to translate the statistical interpretation of analysis results into clinical interpretation. It is important to translate the statistical interpretation of the analytical results into the clinical interpretation when interpreting the results of clinical trials. For this purpose, it is considered important to select and evaluate evaluation indices that can be appropriately interpreted according to the situation.
We hope this report will help in this regard.
In the second edition (released in March 2019), we have introduced the latest examples of RMST and added SAS macros for implementing information volume calculation and survival time data reconstruction, as well as descriptions of RMST analysis implementation in SAS/STAT 15.1.