Generate a block randomization for a clinical trial

This function creates random assignments for clinical trials (or any experiment where the subjects come one at a time). The randomization is done within blocks so that the balance between treatments stays close to equal throughout the trial.

distribution, design , datagen
blockrand(n, num.levels = 2, levels = LETTERS[seq(length = num.levels)], id.prefix, stratum, block.sizes = 1:4, block.prefix, uneq.beg=FALSE, uneq.mid=FALSE, uneq.min=0, uneq.maxit=10)
The minimum number of subjects to randomize
The number of treatments or factor levels to randomize between
A character vector of labels for the different treatments or factor levels
Optional integer or character string to prefix the id column values
Optional character string specifying the stratum being generated
Vector of integers specifying the sizes of blocks to use
Optional integer or character string to prefix the column
Should an unequal block be used at the beginning of the randomization
Should an unequal block be used in the middle
what is the minimum difference between the most and least common levels in an unequal block
maximum number of tries to get uneq.min difference

This function will randomize subjects to the specified treatments within sequential blocks. The total number of randomizations may end up being more than n.

The final block sizes will actually be the product of num.levels and block.sizes (e.g. if there are 2 levels and the default block sizes are used (1:4) then the actual block sizes will be randomly chosen from the set (2,4,6,8)).

If id.prefix is not specified then the id column of the output will be a sequence of integers from 1 to the number of rows. If id.prefix is numeric then the id column of the output will be a sequence of integers starting at the value of id.prefix. If id.prefix is a character string then the numbers will be converted to strings (zero padded) and have the prefix prepended.

The block.prefix will be treated in the same way as the id.prefix for identifying the blocks. The one difference being that the will be converted to a factor in the final data frame.

If uneq.beg and/or uneq.mid are true then an additional block will be used at the beginning and/or inserted in the middle that is not balanced, this means that the final totals in each group may not be exactly equal (but still similar). This makes it more difficult to anticipate future assignments as the numbers will not return to equality at the end of each block.

For stratified studies the blockrand function should run once each for each stratum using the stratum argument to specify the current stratum (and using id.prefix and block.prefix to keep the id's unique). The separate data frames can then be combined using rbind if desired.


A data frame with the following columns:
A unique identifier (number or character string) for each row
Optional, if stratum argument is specfied it will be replicated in this column
An identifier for each block of the randomization, this column will be a factor
The size of each block
The treatment assignment for each subject


Schulz, K. and Grimes, D. (2002): Unequal group sizes in randomized trials: guarding against guessing, The Lancet, 359, pp 966--970.

See Also

plotblockrand, sample, rbind

  • blockrand

## stratified by sex, 100 in stratum, 2 treatments
male <- blockrand(n=100, id.prefix='M', block.prefix='M',stratum='Male')
female <- blockrand(n=100, id.prefix='F', block.prefix='F',stratum='Female') <- rbind(male,female)

## Not run: 
# plotblockrand(,'mystudy.pdf',
#    top=list(text=c('My Study','Patient: %ID%','Treatment: %TREAT%'),
#             col=c('black','black','red'),font=c(1,1,4)),
#    middle=list(text=c("My Study","Sex: %STRAT%","Patient: %ID%"),
#                col=c('black','blue','green'),font=c(1,2,3)),
#    bottom="Call 123-4567 to report patient entry",
#    cut.marks=TRUE)
# ## End(Not run)

Documentation reproduced from package blockrand, version 1.3, License: GPL-2

Community examples

Looks like there are no examples yet.