Skip to contents

RMSTpowerBoost: Power and Sample Size Calculations for RMST-Based Trials

🔍 Overview

As clinical trials and observational studies increasingly move beyond the restrictive proportional hazards (PH) assumption, researchers need flexible alternatives for analyzing time-to-event data. One such approach—Restricted Mean Survival Time (RMST)—has gained traction due to its clear clinical interpretation and robustness in non-PH settings.

Yet, while RMST-based analysis is growing in popularity, tools for study design and power calculation under this framework have lagged behind.

RMSTpowerBoost bridges that gap—offering a powerful suite of methods to design, simulate, and plan RMST-based trials, even under complex censoring mechanisms or nonlinear treatment effects. Whether you prefer point-and-click simplicity or script-based control, RMSTpowerBoost has you covered.

✨ Key Features

✅ Broad Model Support

Design studies under a range of realistic conditions:

  • Non-Stratified Models

    • Linear: Simple, interpretable modeling.
    • GAM (Generalized Additive Models): Flexible for nonlinear covariate effects.

  • Dependent Censoring Models

    • Account for censoring that’s related to baseline covariates.

  • Stratified Models

    • Additive or Multiplicative: Ideal for multi-center trials with center-specific treatment effects.

🧠 Dual Estimation Methods

For many models, you can choose between:

  • Analytical Approach: Based on asymptotic variance formulas. Fast and efficient for large-scale simulations.
  • Bootstrap-Based Method: Simulation-driven, providing greater robustness, especially in smaller or complex datasets.

🖱️ Interactive Web App

No R experience? No problem. A full-featured Shiny web application is available, making advanced RMST-based design accessible through an intuitive graphical interface.

📦 Installation

Install the latest development version directly from GitHub:

# If not already installed
install.packages("remotes")

# Install RMSTpowerBoost
remotes::install_github("UTHSC-Zhang/RMSTpowerBoost-package")

🌐 Try the Shiny App

Explore the tool without writing a single line of code:

👉 Launch the Interactive Web App

Perfect for teaching, preliminary analyses, or rapid prototyping.

If you need to design a study using RMST, whether for a randomized trial, observational cohort, or simulation study, RMSTpowerBoost delivers both flexibility and precision—helping you move confidently beyond the hazard ratio.