restricted: Restricted randomization

In yevgenryeznik/restricted: Restricted Randomization Routine for Clinical Trials

Description

Simulates a clinical trial one time or multiple times with retricted randomizaion procedure

Usage

restricted(w, nsbj, procedure, parameter = NA, nsim = 1)

Arguments

w	vector of fixed allocation ratio ((ir)rational or integers with GCD = 1).
nsbj	number of subjects to randomize.
procedure	name of randomization procedure c("CRD", "PBD", "BUD", "MWUD", "DBCD", "DL", "MinQD", "MaxEnt").
parameter	tuning parameter of randomization procedure (NA for "CRD").
nsim	number of simulations (1 by default).

Value

In case of a single trial simulation (i.e. nsim equals to 1) returns a list with the following items items:

• op – a tibble (data frame) of operational characteristics: forcing index, imbalance.

• probability – a tibble (data frame) of allocation probabilities at each step.

• allocation – a tibble (data frame) of subjects' allocations at each step.

• observation – a tibble (data frame) of treatment assignments.

In case of multiple trial simulations (i.e. nsim larger than 1) returns a list with the following items items:

• op – a tibble (data frame) of operational characteristics: average forcing index (AFI), maximum imbalance (MI), momentum of probability mass (MPM), cumulative momentum of probability mass (CMPM), and "average standard deviation" of the allocation proportions (ASD).

• probability – a tibble (data frame) of unconditional allocation probabilities at each step.

• allocation – a tibble (data frame) of distrbutions of the subjects' allocations at each step.

• observation – a tibble (data frame) of distributions of treatment assignments.

Author(s)

Yevgen Ryeznik ([e-mail]), Oleksandr Sverdlov

References

Ryeznik Y, Sverdlov O (2018) "A Comparative Study of Restricted Randomization Procedures for Multi-Arm Trials with Equal or Unequal Treatment Allocation Ratios", Statistics in Medicine, Submitted.

Examples

   # a single trial simulation (without responses):
   #    target allocation ratio: 4:3:2:1 (i.e. 4 treatment groups)
   #    number of subjects involved in  trial: 100
   #    randomization procedure used: Drop-the-Loser (DL)
   #    randomization procedure parameter: 2

   restricted(c(4, 3, 2, 1), 200, "DL", 2)

   # 200 trial simulations without responses:
   #    target allocation ratio: 31:19 (i.e. 2 treatment groups)
   #    number of subjects involved in  trial: 50
   #    randomization procedure used: Maximum Entropy Constraint Balance
   #          Randomization (MaxEnt)
   #    randomization procedure parameter: 0.5
   #    number of simulations: 200

   restricted(c(31, 19), 50, "MaxEnt", 0.5, nsim = 200)

yevgenryeznik/restricted documentation built on May 26, 2019, 6:32 a.m.