Back to search

Boston University

New statistical methods with the restricted mean survival time for randomized controlled trials

Abstract

dc:description.abstract

In randomized trials with time-to-event outcomes, the hazard ratio (HR) is the most commonly used measure of treatment effect. However, the HR relies on proportional hazards over time and can be difficult to interpret. An alternative measure of treatment effect is the difference in restricted mean survival times (RMSTD). The RMSTD has a clinically meaningful interpretation on the time scale. Furthermore, it does not rely on model assumptions, such as proportional hazards. I introduce new methods with the RMSTD for randomized trials and their syntheses. First, I propose a framework to design non-inferiority trials. I show how to determine the non-inferiority margin for RMSTD equivalent to a given HR margin and introduce a simulation-based method for sample size determination. I apply the method in a wide range of theoretical scenarios and by redesigning 35 published non-inferiority trials. Seemingly large margins for the HR translate to relatively small margins for the RMSTD. Using the RMSTD can result in considerable reductions in required sample size. Second, I present a method for surrogate endpoint evaluation using a counterfactual mediation model when both the surrogate and clinical outcome are right censored. I implement an illness-death model for risk prediction and define the direct and indirect effects and the proportion of mediation on the RMSTD scale. I derive confidence intervals using perturbation-resampling. I illustrate the method using data from the Scandinavian Prostate Cancer Group Study Number 4 randomized trial. Third, I introduce a multivariate meta-analysis model for RMSTD at multiple time points. I derive the within-trial covariance matrix of RMSTDs, enabling the synthesis of all data by borrowing strength. In a simulation study, I find improved statistical performance of the proposed method over that of two univariate models. I demonstrate the methods in a meta-analysis of randomized trials of transcatheter aortic valve replacement. Finally, I investigate the interpretation of randomized trials with time-to-event outcomes using a randomized experiment of corresponding authors of RCTs and medical residents/fellows. I garner evidence that misinterpretation of HR is common and that participants judge experimental treatments to be less beneficial when presented with the RMSTD as compared with HR.

Degree

thesis:*
Name thesis:degree_name
Doctor of Philosophy
Level thesis:degree_level
doctoral
Discipline thesis:degree_discipline
Biostatistics
Grantor
Boston University
Year dc:date.issued
2019

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Weir, Isabelle Ruth
Advisor dc:contributor.advisor
  • Trinquart, Ludovic
ORCID dc:identifier.orcid

Subjects

dc:subject × 5

Rights

Language dc:language.iso
en_US

Identifiers

dc:identifier.*
Handle dc:identifier.uri
https://hdl.handle.net/2144/39589

Chain of custody

source
Harvested from
Boston University
Base URL
open.bu.edu/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Weir, Isabelle Ruth. New statistical methods with the restricted mean survival time for randomized controlled trials. doctoral thesis, Boston University, 2019. https://hdl.handle.net/2144/39589