Sample Size Calculation — ssize

Sample size calculation using model-based discrimination measure.

Usage

ssize_D(
  fit,
  D = NULL,
  cens = NULL,
  p = 0.1,
  alpha = 0.05,
  power = 0.8,
  delta = 0.25,
  trial = c("superiority", "non-inferiority")
)

fit: an object fit of class coxph
D: discrimination measure D, defaults to output of survival::royston() for fit
cens: censoring proportion, defaults to proportion calculated from fit
p: proportion of D accepted as delta, defaults to 10%.
alpha: Type I error
power: statistical power (1 - Type II error)
delta: superiority/non-inferiority margin. The difference we wish to detect between a prior model and a new model.
trial: either "superiority" or "non-inferiority"

The final sample size needed in the new model using the supplied input parameters.

The sample size calculation uses a discrimination measured based on multivariable prognostic time-to-event models.

Derek Chiu