Allocation Concealment: The Randomized-Trial Safeguard That Works Before Assignment
Anas H. Alzahrani, MD PhD MPH
Department of Preventive Medicine and Public Health
Faculty of Medicine, King Abdulaziz University
A trial uses a computer-generated random sequence. The recruiter also keeps the list in an unlocked spreadsheet. When the next assignment is control, a borderline patient is asked to return tomorrow; when it is the new treatment, the same patient is enrolled immediately. The sequence is random. Entry into the sequence is not.
Allocation concealment prevents recruiters and participants from knowing the next assignment before enrollment becomes irreversible. It protects the decision about who enters which group. Without it, an unpredictable sequence can be subverted—consciously or unconsciously—before the intervention is ever delivered.
Random Sequence Generation Is Necessary, Not Sufficient
Randomization succeeds in practice only when two protections work together. First, the sequence must be unpredictable: computer-generated random numbers can do this, while alternation, date of birth, chart number, or day of the week cannot. Second, the next assignment must remain hidden until the participant has been accepted into the trial.
These are separate failure points. A predictable rule fails even if the recruiter never sees a list. A truly random list fails if the recruiter can open it, hold an envelope to the light, or ask a central service for the assignment before confirming eligibility and consent.
Concrete takeaway
Do not accept “participants were randomized” as a method. Ask who knew the next assignment, at what moment, and what irreversible step had already occurred.
Concealment Is Not Blinding
| Safeguard | When it operates | What it protects |
|---|---|---|
| Sequence generation | Before enrollment | Unpredictability of the assignment order |
| Allocation concealment | Before and until assignment | Selection into treatment groups |
| Blinding or masking | After assignment | Care, behavior, outcome assessment, and analysis from knowledge of treatment |
Blinding may be impossible in a surgery-versus-physiotherapy trial. Allocation concealment can still be implemented. Conversely, identical tablets do not repair a trial in which the recruiter knew the next assignment before enrolling the participant. “Double blind” therefore does not answer the concealment question.
What Adequate Implementation Looks Like
Central or third-party assignment
A secure web, telephone, pharmacy, or coordinating-center system reveals the assignment only after required participant information confirms enrollment. The recruiter cannot browse future assignments or reverse the request after seeing the result.
Sequentially numbered identical containers
A pharmacy prepares treatments according to the sequence in externally indistinguishable, numbered containers. Staff use the next container only after eligibility and consent are complete.
Sequentially numbered, opaque, sealed envelopes
Envelopes can work when they are genuinely opaque, securely sealed, opened in order, protected against replacement or premature opening, and audited. Writing participant details before opening creates a trace. The acronym is not the safeguard; the operational controls are.
Method labels are not enough. “Sealed envelopes” omits whether they were opaque, sequential, tamper-evident, and opened only after the enrollment decision. “Central randomization” omits whether staff could request, reject, or repeat an assignment. Good reporting describes the mechanism and the people who implemented it.
How Foreknowledge Produces Bias
Recruiters often know prognostic details before assignment: disease severity, frailty, likely adherence, access to follow-up, or a clinician's preference. If the next treatment is known, even small changes in eligibility interpretation, timing, or persuasion can create systematic differences between groups.
The bias is created at entry, so adjustment for a short table of measured baseline variables is not a guaranteed repair. The feature that influenced enrollment may be unmeasured, poorly recorded, or visible only as a combination of clinical judgments. Baseline imbalance can raise suspicion, but apparent balance does not prove that concealment worked. Randomized trials can show chance imbalances, and compromised allocation can sometimes leave familiar covariates looking similar.
Meta-epidemiological studies have found that trials reporting inadequate or unclear concealment tend, on average, to report larger treatment effects than trials reporting adequate methods. That evidence describes an average association across collections of trials; it does not quantify the bias in any one trial. For an individual paper, the correct conclusion may be “risk of bias is uncertain” when the mechanism is not reported—not “concealment definitely failed.”
The 90-Second Allocation Audit
- Who generated the allocation sequence, and could recruiters access it?
- When did eligibility and consent become irreversible relative to revealing the assignment?
- What mechanism delivered the next assignment: central service, pharmacy, numbered container, or envelope?
- If envelopes were used, were they sequentially numbered, opaque, sealed, monitored, and opened only after enrollment?
- Who enrolled participants, who assigned the intervention, and were those roles separated from sequence generation?
- Do baseline counts, exclusions before assignment, or enrollment patterns suggest that foreknowledge could have influenced entry?
CONSORT 2025 asks authors to report the mechanism used to implement the random sequence, steps taken to conceal it until assignment, and who generated the sequence, enrolled participants, and assigned interventions. Review those details as one process rather than three disconnected checklist entries.
Worked Example: The Envelope That Is Technically Sealed
A single-center trial reports “computer randomization with sealed envelopes.” A coordinator generates the list, prints thin envelopes, stores them in an unlocked drawer, determines eligibility, and opens an envelope before the consent form is signed.
The random-number generator addresses sequence generation. It does not protect the remaining process. The envelopes may be readable against light; the coordinator has incompatible roles; access is uncontrolled; and enrollment is still reversible when assignment becomes known. The review should request the full implementation details and treat the risk of bias as unresolved rather than upgrading the method because two reassuring words appear in the manuscript.
Sources and Evidence Maturity
- CONSORT 2025 explanation and elaboration — official item-level guidance on sequence generation, concealment, implementation, and blinding.
- Cochrane Handbook, Chapter 8 — official risk-of-bias guidance for intervention assignment in randomized trials.
- Wood and colleagues, BMJ 2008 — a meta-epidemiological study examining how reported concealment and blinding relate to estimated intervention effects.
Evidence note: the direction and size of bias cannot be inferred mechanically for a single trial. Incomplete reporting can reflect either incomplete conduct or incomplete description; protocols and author clarification may reduce uncertainty.
Where Aqrab Fits
Aqrab helps reviewers turn “randomized” into an inspectable chain of claims. Paste a trial's methods section into Aqrab Try and ask it to separate sequence generation, concealment, implementation, and blinding. For repeatable methods review, use the developer documentation to add the allocation audit to a structured workflow.
Keep reading
Don't stop at one method.
Good methods judgment comes from contrast. Read the neighboring guides, see where the assumptions diverge, and avoid treating every observational problem like it needs the same hammer.
Hierarchical Testing in Clinical Trials: When a Significant Secondary Endpoint Is Still Descriptive
A practical hierarchical testing guide for clinical researchers. Reconstruct the prespecified testing path before treating a small p-value on a secondary endpoint as confirmatory evidence.
Randomized Withdrawal Trials: Why a Relapse-Prevention Win Is Not a New-Patient Effect
A practical randomized withdrawal trial guide for clinical researchers. Audit the run-in, responder enrichment, withdrawal contrast, safety, and target population before generalizing a maintenance-effect claim.
Cluster-Trial Recruitment Bias: When the Clinic Knows the Assignment Before the Patient Enters
A practical guide to post-randomization recruitment bias in cluster trials. Audit timing, allocation awareness, eligibility, consent, patient denominators, and the limits of statistical adjustment.