Getting started with survSampleSize

Overview

survSampleSize provides an interactive Shiny application for sample size and power calculation in clinical trials with a survival (time-to-event) endpoint, under general design conditions.

Two complementary methods are available:

• The Lu (2021) weighted log-rank method (via the lrstat package), which supports non-proportional hazards, delayed treatment effects, unequal allocation, dropout, non-inferiority testing, and Fleming-Harrington weighted log-rank statistics.
• The classic Freedman (1982) method (via the powerSurvEpi package) for the proportional-hazards setting.

Launching the application

The package exposes a single user-facing function, run_app(), which launches the Shiny application in your default browser:

library(survSampleSize)
run_app()

The interactive app relies on several packages declared in Suggests. If any are missing, run_app() reports which ones to install. You can install all of them up front with:

install.packages(c(
  "lrstat", "powerSurvEpi", "DT", "ggplot2", "bslib", "plotly"
))

Workflow

Inside the app, the typical workflow is:

1. Choose method and direction. Select either the Lu (2021) or Freedman (1982) method, and whether to solve for the sample size N given a target power, or to solve for the power given a fixed N.

2. Statistical design parameters. Set the significance level (alpha), target power, one- vs. two-sided test, allocation ratio, and – for the Lu method – an optional non-inferiority margin.

3. Time parameters (months). Set the accrual duration and the follow-up time after accrual ends. The Freedman method instead uses a single total study duration.

4. Survival and effect parameters. Set the control-arm median survival, the target hazard ratio, and – for the Lu method – the delayed-effect (DTE) time, the annual dropout rate, the accrual rate, and the Fleming-Harrington weighting.

5. Calculate. The results panel reports the total sample size, expected number of events, study duration, and/or estimated power. Additional tabs show the theoretical survival curves, a calendar-time event-prediction timeline, and a side-by-side comparison of the two methods.

References

• Freedman, L. S. (1982). Tables of the number of patients required in clinical trials using the log-rank test. Statistics in Medicine, 1(2), 121-129.
• Lu, K. (2021). Sample size calculation for logrank test and prediction of number of events over time. Pharmaceutical Statistics, 20(2), 229-244.