Sample Size and Power Calculation for Common Non-Parametric
Tests in Survival Analysis
Description
A number of statistical tests have been proposed to compare two
survival curves, including the difference in (or ratio of) t-year
survival, difference in (or ratio of) p-th percentile survival, difference in
(or ratio of) restricted mean survival time, and the weighted log-rank test.
Despite the multitude of options, the convention in survival studies is to assume
proportional hazards and to use the unweighted log-rank test for design and
analysis. This package provides sample size and power
calculation for all of the above statistical tests with allowance for
flexible accrual, censoring, and survival (eg. Weibull, piecewise-exponential,
mixture cure). It is the companion R package to the paper by Yung and Liu (2020)
. Specific to the weighted log-rank test, users may
specify which approximations they wish to use to estimate the large-sample mean
and variance. The default option has been shown to provide substantial
improvement over the conventional sample size and power equations based on Schoenfeld
(1981) .