The King's Clinical Trials Unit offers a competitively priced Independent Randomisation Service.  The Independent Randomisation Service ensures reliability and credibility in the randomisation process: the CTU sets up and manages a database, purpose-designed for each individual trial, after being briefed by the research team and trial statistician.

The randomisation service is based online at:

Randomisation Service

Randomisation Service- Advanced

Each researcher is issued with a unique username and password to use for their study and can log in via the web to randomise their patients. Once the randomisation has been requested, patients are allocated to a treatment arm or medication pack number instantly, with emails automatically generated confirming the randomisation has occurred. If any errors are made in the request (eg incorrect patient initials, date of birth or stratification factors), the trial manager should email randomization.request@kcl.ac.uk as soon as the error is noticed and it will be logged on the system.

A randomisation cannot be 'undone' where incorrect stratification factors have been used, so care must be taken when requesting randomisation. However information about errors can be recorded alongside the randomisation data for that patient, so that this key information is not missed when data is extracted for the study statistician. Data extracts can be requested at any time for TSC and DMC meetings or for other reports. Cost depends on the size and complexity of the study. Please contact the Clinical Trials Unit for a free quote.

RANDOMISATION METHODS:

1. Simple randomisation
Randomization without restriction. In a two-group trial, it is analogous to the toss of a coin.

Pros: simple to understand and implement.
Cons: imbalance likely for smaller trials (N<2000).

2. Block randomisation
Block randomisation allows the trial investigators to control the numbers of subjects allocated to each group during the enrollment phase of the study by using 'blocks' of sequences of allocation assignment wherein each letter represents an intervention or arm of the trial, e.g. blocks of four might consist of: AABB, ABAB, ABBA, etc. This example shows an equal distribution of participants to each arm of the study but this might not always be the case. Also, the blocks can be of any size, but ideally the size should correspond to a multiple of the number of groups in the study, e.g. for a study with three groups blocks of six or nine would be used. E.g. AABBCC, ABCABC, etc. Altman DG. Practical statistics for medical research. London: Chapman & Hall, 1991.

Pros: Balances the number of patients allocated to each treatment group.
Cons: Doesn't balance patient characteristics. Deterministic (predictable) towards the end of the block.

3. Stratified block randomisation
Stratified block randomisation can further restrict chance imbalances to ensure the treatment groups are as alike as possible for selected prognostic variables or other patient factors. A set of permuted blocks is generated for each combination of prognostic factors. This ensures that treatments are balanced at the end of every strata block.

Pros: Balances the number and characteristics of patients allocated to each treatment group.
Cons: Fairly complex to implement without computer software. Not suitable for very small trials

4. Minimisation
Minimisation is used as an alternative method to stratified randomisation to provide treatment groups that are similar for several variables. Stratified randomisation on several variables can lead to large numbers of strata. This can lead to a large number of incomplete blocks and to a substantial imbalance overall which minimisation can avoid.

The principle behind minimisation: People are allocated (with a probability greater than ½) to the treatment group that would minimise the imbalance between the groups with respect to the variables of interest.

Criteria for method: This method depends on the characteristics of the person entering the trial (the first person is entered by simple randomisation) and therefore no pre-arranged list can be constructed.

Example: Suppose that 20 people have been entered into the trial in this centre and that the next person is ready to receive the treatment assignment who is <10, male and <2 years since diagnosis. A table can be constructed as follows for a similar table/explanation:

The treatment allocation for this person is determined by looking at the number of people already in the study with these characteristics by treatment group, i.e.

Therefore to make the groups as similar as possible allocate this person (with probability >½) to the placebo group.

Pros: Guarantees balance, can be used in very small trials and with multiple stratification variables.
Cons: Complex to implement without software, sometimes criticised on methodological grounds due to lack of random element.