des_opt: Design an optimal group-sequential clinical trial for a...

In mjg211/OptGS: Optimal and Near-optimal Group-sequential Designs for Clinical Trials with Continuous Outcomes

Description

des_opt() determines optimal group-sequential clinical trial designs assuming the primary outcome variable is normally distributed, using the approach proposed in Wason et al (2012).

Usage

des_opt(
  J = 2,
  alpha = 0.05,
  beta = 0.2,
  delta = 0.2,
  sigma0 = 1,
  sigma1 = sigma0,
  ratio = 1,
  w = c(1, 0, 0, 0),
  quantile_sub = FALSE,
  integer_n = TRUE,
  parallel = 1,
  popSize = 50,
  maxiter = 100,
  run = maxiter,
  seed = Sys.time(),
  summary = FALSE
)

Arguments

J	A numeric indicating the chosen value for J, the maximal allowed number of stages. Must be an integer greater than or equal to 2. Defaults to 2.
alpha	A numeric indicating the chosen value for α, the desired type-I error-rate. Must be strictly between 0 and 1. Defaults to 0.05.
beta	A numeric indicating the chosen value for β, the desired type-II error-rate. Must be strictly between 0 and 1. Defaults to 0.2.
delta	A numeric indicating the chosen value for δ, the treatment effect to power the trial for. Must be strictly positive. Defaults to 0.2.
sigma0	A numeric indicating the chosen value for σ0, the standard deviation of the responses in the control arm. Must be strictly positive. Defaults to 1.
sigma1	A numeric indicating the chosen value for σ1, the standard deviation of the responses in the experimental arm. Must be strictly positive. Defaults to sigma0.
ratio	A numeric indicating the chosen value for r, the allocation ratio to the experimental arm relative to the control arm. Must be strictly positive. Defaults to 1.
w	A numeric vector of length 4 indicating the weights to use in the optimality criteria. All elements must be greater than or equal to 0, and at least one of the first 3 elements must be strictly positive. See wason (2015) for further details. Defaults to c(1, 0, 0, 0).
quantile_sub	A logical variable indicating whether quantile substitution should be applied to the identified stopping boundaries. Defaults to FALSE.
integer_n	A logical variable indicating whether the computed values for n0 and n1, the group sizes in the control and experimental arms, should be forced to be whole numbers. Defaults to TRUE.
parallel	See ga.
popSize	See ga.
maxiter	See ga.
run	See ga.
seed	A variable to be passed to set.seed, to make the calculations reproducible.
summary	A logical variable indicating whether a summary of the function's progress should be printed to the console. Defaults to FALSE.

Value

A list with additional class "OptGS_des". It will contain each of the input variables (subject to internal modification), relating them to the outputs of the various group-sequential design functions in OptGS, along with additional elements including:

• CovZ: A numeric matrix giving Cov(Z), the covariance between the standardised test statistics for the identified design.

• e: A numeric vector giving e, the efficacy stopping boundaries for the identified design.

• f: A numeric vector giving f, the futility stopping boundaries for the identified design.

• GA: A list containing the output from the call to ga and each of the ga specific input parameters.

• I: A numeric vector giving I, the vector of information levels for the identified design.

• n: A numeric vector giving n, the vector of stage-wise sample sizes for the identified design.

• n_fixed: A numeric giving the sample size required by a corresponding fixed-sample design.

• n0: A numeric giving n₀, the group size in the control arm for the identified design.

• n1: A numeric giving n₁, the group size in the experimental arm for the identified design.

• name: A character string giving a name for the identified design.

• opchar: A tibble giving the operating characteristics of the identified design when τ = 0, τ = δ, and τ = argmax_θESS(θ).

See Also

build, des_nearopt, des_opt, est, opchar, sim, plot.OptGS_des, print.OptGS_des, summary.OptGS_des

Examples

# The optimal group-sequential design for the default parameters
des     <- des_opt()
# A three-stage optimal design
des_3   <- des_opt(J = 3)
# Optimal under the alternative hypothesis
des_alt <- des_opt(w = c(0, 1, 0, 0))

mjg211/OptGS documentation built on May 28, 2021, 3:44 p.m.