Randomisation
- Randomisation
- Background
- Main features
- Common types of randomisation
- Simple randomisation
- Block randomisation
- Cluster randomisation
- Methods of randomisation
- Envelopes / containers
- Telephone
- Web
- Stratification
- Allocation concealment
- References
The information on randomisation is designed to be used by
novice researchers and is friendly and simple to use. MRSC
can provide a randomisation service. Currently there are three people
at the MRSC who can assist with the details of trial
randomisation. They are Vicki Flenady, Ibinabo Ibiebele and Kristen Gilshenan. The best way to make
contact is to send emails to
Kate Reynolds.
Assistance can be given in the form of email correspondence,
phone advice if the questions is simple (how do I use the
randomisation template) or an informal consultation. We can also
offer advice on all stages of research coordination and
implementation including recruitment, data collection,
assistance with research assistants etc.
Background
- Main reason for using randomisation is to allocate treatments to patients in a controlled trial in a way which minimises biases.
- Want to compare the outcomes of treatments given to groups of patients which do not differ in any systematic way.
- Statistical theory is based on the idea of random sampling. In a study with random allocation the differences between treatment groups behave like the differences between random samples from a single population.
- The term random does not mean the same as haphazard but has a precise technical meaning.
- Random allocation means that each patient has a known chance, usually an equal chance, of being given each treatment, but the treatment to be given cannot be predicted.
- If clinicians or participants can select which group they participate in then bias can occur. For example, clinicians may assign patients who have less hopeful prognosis to the experimental group, trying to give them the ‘best possible chance’, thereby making the treatment group look less effective than it really is. Randomisation minimises such bias.
- Randomisation does not eliminate selection bias. Clinicians can still make professional judgements on who is approached for participation in a clinical trial.
- An inability to speak English, perceived low intelligence and social or emotional problems may be used by clinicians to deny patients participation in a clinical trial.
Main features of the randomisation process:
- Generation of an unpredictable assignment sequence based on a random sequence and;
- Concealment of that sequence until allocation occurs.
Types of randomisation
-
Simple Randomisation
- If there are two treatments the simplest method of random allocation gives each patient an equal chance of getting either treatment; it is equivalent to tossing a coin.
- In practice most people use either a table of random numbers or a random number generator on a computer.
- Simple randomisation does not ensure that there will be equal distribution of patient characteristics between treatment groups.
- In small trials there may be substantial differences in group sizes that will reduce the precision of estimates of the difference in treatment effect and hence efficiency of a study.
- Possible modifications include block randomisation, to ensure closely similar numbers of patients in each group, and stratified randomisation, to keep the groups balanced for certain prognostic patient characteristics.
Example of simple randomisation
Program to produce random integers (simple randomisation)
This can be accessed through the MRSC web pages at here.
Hit ‘Make random seed’ (this just gives the computer somewhere
to start its randomisation process). Record the random seed
number as this can be used to generate the same randomisation
schedule if the first one is misplaced.
To generate simple random numbers table
- Insert min value (1)
- Insert max value (maximum number of groups being compared eg if comparing 2 groups; type 2, eg treatment drug vs standard treatment, if comparing 3 groups; type 3, eg treatment drug1 vs treatment drug2 vs standard treatment drug)
- Insert number required (sample size)
- Hit ‘Make random numbers’
- The randomisation schedule can then be copied and pasted into an excel spreadsheet for easier access.
Eg: Randomised controlled trial comparing 3 groups. Sample size 100.
Important: The smaller the sample size the more chance that the numbers in each group will be uneven. Therefore you may get not only wide variation in the number in each group but also you may get an uneven distribution of patient characteristics in each group.
-
Block Randomisation
- The numbers in the two groups at any time can never differ by more than half the block length.
- Block size is normally a multiple of the number of treatments.
- Care should be taken to choose block lengths that are sufficiently short to limit possible imbalance, but that are long enough to avoid predictability towards the end of the sequence in a block.
- Investigators and other relevant staff should generally be blind to the block length; the use of two or more block lengths, randomly selected for each block, can achieve the same purpose.
Example of block randomisation
Program to produce block randomisation
This can be accessed through the MRSC web pages at here
Hit ‘Make random seed’ (this just gives the computer somewhere to start its randomisation process). Record the random seed number as this can be used to generate the same randomisation schedule if the first one is misplaced.
To generate block randomisation numbers
Eg: Randomised controlled trial comparing 3 groups with a sample
size of 60 and a block size of 6.
Therefore: number of groups 3, block size of 6, sample size 60.
- Number of blocks required ( sample size divided by block size, eg 60 / 6 = 10)
- Insert min value (1)
- Insert max value (block size)
- Hit ‘Generate random blocks’
Eg: Randomised controlled trial comparing 3 groups. Sample size 60. Block size 6
Converting to workable blocked random numbers
- The randomisation schedule must then be copied and pasted into an excel spreadsheet to finish the blocking
- Ignore the bottom numbers under ‘count of each value (cols) in each array element (rows)
- Copy and paste the top numbers to an excel spreadsheet
- Convert to modular form to get your groups by placing the cursor in a spare cell and typing =MOD(A1,3). A1 being where the first number is and 3 being your group size.
- The randomisation is now completed in ROWS. Note each row now
has 2 of each group (randomly arranged) ie For every 6 recruits,
2 participants will be randomised to treatment 0, 2 participants
to treatment 1 and 2 participants to treatment 2.
-
Cluster randomisation
- A trial where the individuals are randomised in groups (the group is randomised not the individual) eg hospitals, villages, schools etc.
- Mostly used to avoid contamination eg GP patients randomised with contamination from health promotional material in waiting room or waiting room ‘gossip’ or for convention eg water supply to a village.
- An example of possible contamination is a trial looking at patients receiving a behavioural intervention to reduce smoking rates. Individuals randomised to the treatment arm may talk to control patients who then may adopt the experimental treatment.
Methods of randomisation
-
Envelopes or containers
- Mostly used for small local single centred trials.
- Random numbers are generated and the trial number and allocated treatment group placed sequentially onto the envelopes. (eg. either Treatment A or Treatment B if blinded or Treatment A or Standard care if not blinded).
- The outside of the envelope should only have the trial name and the study number written on it.
- Envelopes need to be opaque and sealed so that the group allocation cannot be read prior to opening the envelope.
- The envelopes need to be taken sequentially.
Medications are sealed in sequentially numbered identical
containers.
-
Telephone
- Used extensively for larger multicentred trials.
- Usually a 24 hour service.
- Researchers dial into a centre telephone number and go through a process of identifying inclusion and exclusion criteria before the patient is randomised to a study number and treatment group.
- Operated by central coordinating centre or commercial randomisation
services.
-
Web based
- Used for large multicentred trials.
- Similar to telephone randomisation except done through a computer website operated by the main coordinating centre.
Stratification
- Whilst randomisation minimises bias from the allocation procedure, it does not guarantee, for example, that the individuals in each group have a similar age distribution.
- In small studies some chance imbalance will probably occur, which might complicate the interpretation of results.
- Stratified randomisation can be used to achieve approximate balance of important characteristics that may influence prognosis or treatment responsiveness without sacrificing the advantages of randomisation.
- The method is to produce a separate block randomisation list for each subgroup (stratum).
- Some characteristics which researchers may stratify for include sex and disease severity or any other variable that may be expected to have an important influence on the outcome of a clinical trial.
- Stratification should be done prospectively.
- If performed retrospectively it is called subgroup analysis.
- Stratified treatment allocation must be based on block randomisation within each stratum rather than simple randomisation; otherwise there will be no control of balance of treatments within strata, so the object of stratification will be defeated.
- Stratified randomisation can be extended to two or more stratifying variables. With multiple strata some of the combinations of categories may be rare, so the intended treatment balance is not achieved.
- In small studies it is not practical to stratify on more than one or perhaps two variables, as the number of strata can quickly approach the number of subjects.
Examples
When comparing two alternative treatments for breast cancer it
might be important to stratify by menopausal status. Separate lists of
random numbers should then be constructed for premenopausal and
postmenopausal women.
In a multicentre study the patients within each centre will need to be
randomised separately. Thus “centre” is a stratifying variable, and
there may be other stratifying variables as well.
Importance of Allocation Concealment
- Generation of a random sequence of numbers does not in itself assure proper randomisation.
- If there is any means for researchers to subvert the randomisation process the resulting effect may be an inaccurate assessment of the effects of the treatment.
- For example, if the researcher correctly generated a random allocation sequence but then pins the allocation sequence on the wall in the office this would allow anyone involved in the recruitment process to select which allocation certain potentially eligible recruits received.
- Thus if a researcher thought one group was better they may
allocate a certain type of patient into that group thereby
making subsequent interpretation of the results difficult.
Details of the randomisation that facilitate predictability (e.g. block length) should not be contained in the trial protocol. - The randomisation schedule itself should be filed securely by the sponsor or an independent party in a manner that ensures that blindness is properly maintained throughout the trial.
- Access to the randomisation schedule during the trial should take into account the possibility that, in an emergency, the blind may have to be broken for any subject.
- The procedure to be followed, the necessary documentation,
and the subsequent treatment and assessment of the subject
should all be described in the protocol.
Allocation concealment should not be confused with blinding which is the concealing of the identity of the treatment allocated.
References:
European Medicines Agency. 1998. Note for Guidance on Statistical Principles for Clinical Trials (CPMP/ICH/363/96)
http://www.emea.eu.int/pdfs/human/ich/036396en.pdf [Accessed 25 September 2006].
European Medicines Agency. 2003. Points to Consider on Adjustment for Baseline Covariates. http://www.emea.eu.int/pdfs/human/ewp/286399en.pdf [Accessed 25 September 2006].
Roberts C, Torgerson D. Randomisation methods in controlled trials. http://bmj.bmjjournals.com/cgi/content/full/317/7168/1301 [Accessed 9 August 2006].