Statistical issues in survival analysis (Meta-analysis & RMSE)
February 26, 2025 The authors have used network meta-analysis (NMA) to compare multiple treatments simultaneously using restricted mean survival time (RMST) with individual level data (IPD). Since the NMA was mainly used with aggregate data, they had to extend this technique to work with IPD and also RMST. NMA has been an extension of pairwise meta-analysis. They can leverage information about both “direct” and “indirect” comparisons of treatment. There are two stages for the IPD-NMA. The two stage one first estimates treatment effects within each study and then combines study-level effects across studies using NMA while a one stage approached combines the IPD of all studies using a hierarchical model. There are various benefits of using one approach versus the other. In the use of RMST with IPD-NMA, people were using a twostage approach where the RMST is evaluated on IPD in the first stage and then the data is aggregated together in the second stage, but they state this method didn’t properly adjust for covariates in the first stage. Therefore, the authors have tried to develop a one stage IPD-NMA using RMST. They also still proposed advanced two stage methods besides one stage methods for time-to-event outcomes estimated by RMST.