The Cox proportional hazards model, proposed by Dr. Cox in 1972, has traditionally been employed in disease prevention studies with time-to-event outcomes as the primary endpoint. The hazard ratio (HR) is used to summarize the magnitude of the intervention effect in these studies. However, the limitations of using Cox’s HR as a summary of intervention effect magnitude have been widely discussed. Restricted Mean Survival Time (RMST), first introduced in 1949 by Dr. Irwin, has gained attention since approximately 2013 as an alternative to address the limitations of the traditional approach. There is now a growing body of literature demonstrating the applications of RMST in various prevention research contexts. 

This presentation reviews the limitations of Cox’s HR and introduce the utilization of RMST as a robust alternative that calculates the mean survival time within a specified time window, offering a more intuitive interpretation of intervention effects. Dr. Uno then discusses several recent methodological advancements in RMST-based analysis. Furthermore, Dr. Uno introduces essential software tools for RMST analysis, including the survRM2 package in R, the RMSTREG procedure in SAS, and the strmst2 command in Stata. Use case examples are also provided to illustrate these concepts effectively.

Dr. Hajime Uno is an Assistant Professor of Medicine at Harvard Medical School. He is a Principal Biostatistician and the Director of the Statistical Programming Core in the Division of Population Sciences at the Dana-Farber Cancer Institute. He received his Ph.D. in biostatistics from Kitasato University in Tokyo, Japan, and completed postdoctoral training under the direction of Prof. LJ Wei at the Harvard School of Public Health. Since then, he has been working on various methodological and clinical research projects. 

One of his notable methodological contributions is a modification of the C-index to quantify risk prediction models for censored time-to-event data. His paper published in Statistics in Medicine in 2011 was ranked one of the 10 most cited Statistics in Medicine papers in the following two years. His method has been called “Uno’s C” and is employed in the SAS PHREG procedure. The number of citations of this paper is over 1,300 as of May 2024. One of his current research interests is to improve the everyday practice of survival data analysis in clinical research as well as to disseminate better alternatives to improve the quality of informed decisions about interventions.