CMStatistics 2017: Start Registration
View Submission - CMStatistics
B0590
Title: Restricted mean survival time in time-to-event analysis: From individual level data to aggregate data Authors:  Yinghui Wei - Plymouth University (United Kingdom) [presenting]
Abstract: Restricted mean survival time refers to the mean survival time up to a specific time point. As an alternative measure to the widely used hazard ratio, the between-group difference in restricted mean survival time avoids the proportional hazards assumption and allows the effect measure to vary with time. Examples will be given to illustrate the use of restricted mean survival time in individual study. In aggregate data meta-analyses of survival outcomes from multiple studies, an essential step is to extract the estimated effect measures and their variances from publications. However, the relevant information, including restricted mean survival time, may not be reported in publications. We develop a tool to convert the published Kaplan-Meier curves to survival data at individual level, which contain the event indicator as well as the time to event for each individual. The reconstructed data open the possibility to estimate the restricted mean survival time, and hence enable the use of this alternative measure in aggregate meta-analyses of time-to-event outcome.