two_stage_design: Two-stage design

In dominicmagirr/gsdelayed: Design a group-sequential trial when anticipating a delayed effect.

Two-stage design

Description

Two-stage design for a modestly-weighted log-rank test

Usage

two_stage_design(t_star = NULL, model, recruitment, dco_int, dco_final,
  events_int = NULL, events_final = NULL, alpha_one_sided = 0.025,
  alpha_spend_f = ldobf, length_t = 18, rho = NULL, gamma = NULL)

Arguments

t_star	Parameter of the modestly-weighted log-rank test. Setting t_star=0 corresponds to a standard log-rank test.
model	The piecewise hazard model. A list containing the change_points and lambdas.
recruitment	List of recruitment information. Containing Sample size on control, n_0 Sample size on experimental, n_1 Recruitment period, r_period Recruitment parameter for power model, k
dco_int	The time of the interim analysis in time-units since start of the trial.
dco_final	The time of the final analysis in time-units since start of the trial.
events_int	May be specified instead of dco_int. The number of events that triggers the interim analysis.
events_final	May be specified instead of dco_final. The number of events that triggers the final analysis.
alpha_one_sided	One-sided alpha level.
length_t	Number of cutpoints to use when approximating the distribution of the MWLRT-statistic. Default is 18. Can be increased for greater accuracy.
alpha_spending_f	The alpha-spending function. Default is the Lan-DeMets O'Brien-Fleming function.
{rho}	rho parameter in a Fleming-Harrington test. Default is NULL. Only used if F-H test used instead of MWLRT.
{gamma}	gamma parameter in a Fleming-Harrington test. Default is NULL. Only used if F-H test used instead of MWLRT.

Value

A list describing the design.

dominicmagirr/gsdelayed documentation built on May 15, 2024, 8:24 a.m.