Randomisation and allocation concealment
- Concept
Chance draws up the list of who goes to each group; concealment stops anyone from seeing it before the patient is enrolled.
What it is
Randomisation produces an unpredictable sequence that decides each participant's group. It can be simple (like tossing a coin), in blocks (to keep group sizes equal) or stratified (blocks within each centre or risk group). Allocation concealment protects that sequence until the patient is already enrolled; it is different from blinding, which hides the treatment after allocation. Without concealment, the recruiter can choose whom to enrol and the benefit of chance is lost.
Worked example
- A trial will enrol 40 patients in blocks of 4, each block with 2 A and 2 B allocations.
- There are 6 possible orders per block: AABB, ABAB, ABBA, BAAB, BABA and BBAA.
- 10 blocks are drawn from those 6 orders: at the end there are exactly 20 patients in A and 20 in B.
- Problem: if the recruiter knows blocks are of 4 and has seen A, A, B, the fourth must be B.
- Fix: blocks of varying size (4 and 6) and the list kept in a central system.
- The system reveals the allocation only after the enrolled patient is registered.
- Reading: chance balances the groups and concealment stops anyone from choosing which patient lands in each.
Markers
- Blocks
- Strata
In practice
- In risk-of-bias assessment, sequence generation and its concealment are judged together in the randomisation-process domain.
- Multicentre trials are often stratified by centre, because patients and practice vary between hospitals.
Common pitfalls
- Confusing allocation concealment with blinding: they happen at different moments.
- Allocating by weekday, record number or birth date: it is predictable and not random.
- Using clear or unnumbered envelopes that can be opened and swapped.
Test yourself
Think of the answer, then tap the question to see it.
What is the difference between allocation concealment and blinding?
Concealment protects the sequence before allocation; blinding hides which treatment each person receives after allocation.
What is block randomisation for?
Keeping group sizes equal throughout recruitment, for example 2 A and 2 B in every block of 4.
Why is allocation by odd or even record number unacceptable?
Because it is predictable: the recruiter can know the group in advance and choose whom to enrol.
Can allocation be concealed in a trial of surgery versus medical treatment?
Yes: concealment is always possible, even if the surgeon and the patient cannot be blinded afterwards.
Connections
Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.
Sources: Chapter 8: Assessing risk of bias in a randomized trial (Cochrane, 2024); Allocation bias | Catalog of Bias (Catalogue of Bias Collaboration, 2017).
Spotted a mistake?We are always improving for the community. Tell us what is wrong and our team will review it.Report a mistake
Representative model, for educational purposes only.