GSD_allocation {randomizeR}

Calculates the Type I error for different randomization sequences from a randomization procedure for a group sequential design

Description

Calculates the Type I error for different randomization sequences from a randomization procedure for a group sequential design

Usage

GSD_allocation(
  n,
  reps,
  sfu,
  K,
  rp,
  seed = 42,
  ui = "No",
  rb = 4,
  mti = 3,
  p = 2/3
)

Arguments

n

total sample size

reps

number of simulations to be conducted

sfu

Group sequential design used (currently available: "Pocock" - Pocock, "OF" - O'Brien & Fleming, sfLDPocock - Lan & DeMets with Pocock like alpha spending function, sfLDOF - Lan & DeMets with O'Brien & Fleming like alpha spending function)

K

number of stages

rp

the randomization procedure used (currently available: '"CR"', '"RAR"', '"BSD"', '"CHEN"', '"PBR"', '"MP"')

seed

Randomization seed

ui

for Lan & DeMets design. Update critical values after each stage according to allocation ratio observed if set to "yes".

rb

Block size for randomization procedure PBR.

mti

Maximum tolerated imbalance for randomization procedure BSD and MP.

p

Probability p in favor of the treatment with fewer allocations for EBC and CHEN.

Value

A list consisting of a vector of Type I errors for each randomization sequence generated from the randomization procedure and a S4 object of the class of the randomization procedure.

Examples

#Simulate a group sequential design according to O'Brien and Fleming's design with 24 patients,
#10 simulation runs,3 Stages using Random Allocation Rule as a randomization procedure.
GSD_allocation(n=24, reps=10, sfu="OF", K=3, rp="RAR")
#Simulate a group sequential design according to Lan and deMets design with a Pocock
#like alpha spending function with 18 patients, 10 simulation runs,
#3 Stages using Permuted Block Randomization with block size 4
#as a randomization procedure without updating the critical values after each stage.
library(gsDesign)
GSD_allocation(n=18, reps=10, sfu=sfLDPocock, K=3, rp="PBR", ui="no", rb=4)