Randomisation Flashcards
What steps must be undertaken before randomisation can occur?
Screening
Information provision to eligible candidates
Receiving of consent
Assignment of treatment
What is the main limitation of random block permutation?
There is an element of predictability as the regular balance occurs at the end of each block.
How can predictability be mitigated in block permutated randomisation?
Avoid block sizes that are too small
Keep block size and method of randomisation secret from investigators
Vary size of block
What is the problem with large blocks in randomisation?
If the trial recruitment is stopped mid-block then large blocks can introduce an imbalance
How does replacement of the randomisation list help with randomisation?
Advantages: simple to arrange, reasonable balance can be achieved, no possibility of allocation prediction
Limitations: imbalances may still occur at interim analyses and is affected by sub-optimal recruitment
What is dynamic catch-up
Changing probability of randomisation to one arm based on an existing imbalance to allow catch-up of the lagging arm. Rarely used. Issues with guessing allocation of treatment in unblinded trials
Suggest a method of randomisation that not only achieves adequate randomisation but also can be made to rationalise the baseline of different arms of a trial
Stratified randomisation
What is stratified randomisation?
Randomisation with restriction is carried out separately within each of two or more subsets of participants to ensure that patient characteristics are closely balanced within each arm. E.g. disease severity or study centres
What is the issue with over-stratification?
Separate randomisation lists must be produced for each strata. Each strata is multiplied together to produce the number of lists required.
What scenario lends itself best to stratification?
Best kept to one or two carefully considered prognostic factors in small to medium sized trials
If you want to achieve good balance between treatments on a number of variables across trial recruitment what method can be employed?
Minimisation
List other types of randomisation which have been advocated/used
Unequal randomisation Response-adaptive designs Cluster Randomisation Zelen Design (randomised consent design)
List methods of allocation concealment
SNOSE (sequentially numbered opaque sealed envelopes)
Web-based
Telephone based
Pharmacist
What is the aim of randomisation?
To enable a fair and unbiased comparison of treatments
What are the two key principles of the randomisation process?
A known (usually equal) chance of receiving either treatment and unpredictability of the next treatment allocation