Randomisation

Question 1

What steps must be undertaken before randomisation can occur?

Answer 1

Screening
Information provision to eligible candidates
Receiving of consent
Assignment of treatment

Question 2

What is the main limitation of random block permutation?

Answer 2

There is an element of predictability as the regular balance occurs at the end of each block.

Question 3

How can predictability be mitigated in block permutated randomisation?

Answer 3

Avoid block sizes that are too small
Keep block size and method of randomisation secret from investigators
Vary size of block

Question 4

What is the problem with large blocks in randomisation?

Answer 4

If the trial recruitment is stopped mid-block then large blocks can introduce an imbalance

Question 5

How does replacement of the randomisation list help with randomisation?

Answer 5

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

Question 6

What is dynamic catch-up

Answer 6

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

Question 7

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

Answer 7

Stratified randomisation

Question 8

What is stratified randomisation?

Answer 8

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

Question 9

What is the issue with over-stratification?

Answer 9

Separate randomisation lists must be produced for each strata. Each strata is multiplied together to produce the number of lists required.

Question 10

What scenario lends itself best to stratification?

Answer 10

Best kept to one or two carefully considered prognostic factors in small to medium sized trials

Question 11

If you want to achieve good balance between treatments on a number of variables across trial recruitment what method can be employed?

Answer 11

Minimisation

Question 12

List other types of randomisation which have been advocated/used

Answer 12

Unequal randomisation
Response-adaptive designs 
Cluster Randomisation 
Zelen Design (randomised consent design)

Question 13

List methods of allocation concealment

Answer 13

SNOSE (sequentially numbered opaque sealed envelopes)
Web-based
Telephone based
Pharmacist

Question 14

What is the aim of randomisation?

Answer 14

To enable a fair and unbiased comparison of treatments

Question 15

What are the two key principles of the randomisation process?

Answer 15

A known (usually equal) chance of receiving either treatment and unpredictability of the next treatment allocation