randotools (version 0.2.4)

randolist: Generate randomisation lists

Description

Randomisation lists are central to randomised trials. This function allows to generate randomisation lists simply, via (optionally) stratified block randomisation

Usage

randolist(
  n,
  arms = LETTERS[1:2],
  strata = NA,
  blocksizes = 1:3,
  pascal = TRUE,
  ...
)

Value

object of class randolist which is a dataframe with additional attributes ratio (randomisation ratio, e.g. 1:1, 2:1), arms (arm labels), stratified (logical whether the list is stratified), and stratavars (the stratification variables)

Arguments

n

total number of randomizations (per stratum)

arms

arms to randomise

strata

named list of stratification variables (see examples)

blocksizes

numbers of each arm to include in blocks (see details)

pascal

logical, whether to use pascal's triangle to determine block sizes

...

arguments passed on to other methods

Details

blocksizes defines the number of allocations to each arm in a block. For example, if there are two arms, and blocksizes = 1, each block will contain 2 randomisations. If blocksizes = 1:2, each block will contain either one of each arm, or two of each arm. Total block sizes are therefore blocksizes * length(arms).

By default, frequency of the different block sizes is determined using Pascal's triangle. This has the advantage that small and large block sizes are less common than intermediate sized blocks, which helps with making it more difficult to guess future allocations, and reduces the risk of finishing in the middle of a large block.

Unbalanced randomization is possible by specifying the same arm label multiple times.

To disable block randomisation, set blocksizes to the same value as n.

Examples

# example code
randolist(10)
# one stratifying variable
randolist(10, strata = list(sex = c("M", "F")))
# two stratifying variables
randolist(10, strata = list(sex = c("M", "F"),
                            age = c("child", "adult")))
# different arm labels
randolist(10, arms = c("arm 1", "arm 2"))

# unbalanced (2:1) randomization
randolist(10, arms = c("arm 1", "arm 1", "arm 2"))