Why Arm-Opening Criteria Need to Be Prespecified
A platform trial may add a treatment arm as new interventions or scientific questions become relevant. The decision to open an arm is not merely a scheduling choice: it determines which intervention is offered, to whom, under what protocol, and from what date. If the conditions for opening are left vague, teams may make inconsistent decisions, enroll participants before essential safeguards are in place, or mistake a promising scientific rationale for operational readiness.
Prespecification means defining the evidence and approvals required before the opening decision is made. The criteria should be specific enough that an independent reviewer can determine whether each one has been met, while allowing the protocol to say how genuine exceptions are handled. They should also separate two different questions: Is the proposed comparison scientifically justified? and Can the arm be implemented safely and consistently now? Both must be answered before recruitment begins.
Specify Distinct Gates
An opening rule is clearest when it separates scientific requirements from operational and governance requirements. Passing one category does not compensate for failing another. A strong rationale cannot make an unavailable or inadequately characterized intervention ready; a fully stocked site cannot make an unsupported scientific question worthwhile.
- Scientific rationale and question. State the clinical or scientific question the arm addresses, why the intervention is appropriate for that question, and what evidence supports evaluating it in this trial. The question and proposed treatment contrast must be sufficiently defined to justify the planned arm.
- Intervention readiness. Specify the required evidence that the intervention can be supplied, prepared, stored, prescribed, administered, and monitored as intended. Include any product authorization, manufacturing or supply arrangements, pharmacy procedures, dose instructions, and safety information required by the applicable protocol and oversight bodies. Criteria should be intervention-specific rather than assumed from another arm.
- Eligibility definition. State who may be considered for the new arm and how eligibility will be assessed. The definition should be operationally usable, clinically coherent with the scientific question, and supported by an adequate screening and documentation process. Eligibility criteria should not be left to informal interpretation at individual sites.
- Control availability. Define what must be in place for the planned control allocation or comparison to operate when the arm opens. This may include a control intervention, the ability to deliver the control under the protocol, and a documented process for enrolling and assigning eligible participants. This is an opening-readiness condition; it does not by itself establish which control observations belong in a later analysis.
- Protocol and oversight approval. Identify which protocol documents, participant materials, data systems, safety procedures, and institutional or regulatory approvals must be effective before enrollment. Name the authority that confirms completion and the evidence retained in the trial record.
- Site and staff capability. Specify which participating sites are permitted to open the arm and what training, equipment, laboratory capability, or referral arrangements they need. The protocol should distinguish a platform-level decision to add an arm from authorization for a particular site to recruit.
Each condition should have a verifiable pass standard. “Supply is adequate” is difficult to audit unless the protocol or operational plan identifies the required supply, where it must be available, and who confirms it. “Sites are trained” is clearer when it points to a completed training record for named roles. Thresholds and evidence will differ by intervention and trial; they should be justified and documented rather than borrowed mechanically from another arm.
A Logical Rule for Opening
Represent each required condition as a gate. Let \(G_j=1\) if gate \(j\) is verified as satisfied, and \(G_j=0\) otherwise. If all listed gates are mandatory, the opening indicator is
Here \(O=1\) means the arm may open, and \(O=0\) means it may not yet open. This compact expression is a decision rule, not a statistical model. It makes the non-compensatory logic explicit: if any mandatory gate is zero, the product is zero. A protocol may include conditional or arm-specific requirements, but it should state which conditions apply and how their completion is verified.
The conjunction rule follows directly from the meaning of “all required conditions.” For binary gates, a product equals one only when every factor equals one; if any factor is zero, the product is zero. By contrast, a weighted readiness score could allow an excellent score on supply to offset a missing approval. That is inappropriate when each requirement is essential. If a trial wants to distinguish essential gates from desirable but nonessential indicators, it should label them separately and define what happens when a desirable indicator is not met.
Worked Example: Testing a Proposed Opening Rule
A hypothetical platform plans to add an oral treatment arm. Its prespecified rule requires four conditions: (1) the scientific rationale and treatment question are approved by the steering committee; (2) the medication supply for the first planned recruitment period is confirmed at the participating pharmacy; (3) eligibility screening instructions and staff training are complete; and (4) the revised protocol and participant materials have all required approvals. The trial records the gates in that order.
At review, the committee has approved the rationale, the pharmacy has confirmed stock, and the screening materials and training are complete. However, approval of the revised participant materials is still pending. The gate vector is therefore \((1,1,1,0)\).
Calculation. Applying the rule gives \(O=1\times1\times1\times0=0\). The arm must not open yet, even though three requirements are satisfied. Once the remaining approval is documented, the vector becomes \((1,1,1,1)\), so \(O=1\) and the arm may open, subject to any additional site-specific authorization in the protocol.
The decision is about documented readiness, not a judgment that the pending approval is unimportant because no other issue has arisen. The record should identify the outstanding item, who is responsible for resolving it, and the date on which the decision is reconsidered.
Worked Decisions in Different Settings
Worked Example: A Biologic With Site-Specific Requirements
Suppose a hypothetical platform proposes a biologic that must be stored within a specified temperature range and administered at sites with infusion capability. The protocol defines these opening conditions: a confirmed product supply route; documented temperature monitoring and excursion procedures; trained staff at each opening site; an approved infusion and safety-monitoring procedure; and completed scientific and protocol approvals.
The supply route is confirmed and the scientific question has been approved. Two candidate sites have infusion facilities, but only one has documented temperature-monitoring procedures and completed staff training. The other site has not yet demonstrated either requirement.
Decision. The platform-level arm may be ready to open only if all platform-level conditions have passed. The second site is not ready to recruit because its site-specific gates are incomplete. The first site may be authorized if its required records and all other applicable approvals are complete. This example shows why the opening rule should state whether a condition applies to the arm as a whole, each site, or both.
The criteria do not imply that the second site can never participate. They require the site to meet the specified conditions before recruitment there begins. A readiness checklist should preserve this distinction rather than record a single undifferentiated “arm open” status.
Worked Example: Control Not Yet Available
A hypothetical trial proposes a new arm for a disease subtype. The scientific rationale is approved, the intervention is ready, eligibility procedures are documented, and the protocol has received the required approvals. The opening rule also requires confirmation that the protocol’s control intervention can be provided to participants at each site authorized to open.
At one site, the control intervention is temporarily unavailable and there is no approved procedure for handling that situation. The other requirements are satisfied, but the control-availability gate for this site is not.
Decision. The site cannot open the arm under the stated rule. The team must resolve availability through a protocol-consistent plan and obtain any required approvals before recording that gate as passed. The scientific case for the new intervention does not remove the requirement that the planned trial procedures can be delivered. This is an operational opening decision; it does not determine the later analytic role of control observations.
Documentation, Diagnostics, and Failure Modes
A prespecified rule is useful only if its implementation can be checked. For every gate, retain the criterion, the evidence reviewed, the responsible reviewer, the decision, and its date. Record an overall opening decision separately from the date a particular site is authorized to recruit. If a condition is not met, the record should identify the deficiency rather than converting “pending” into “pass.”
Common signs that the process is going wrong include:
- Vague standards. Words such as “ready,” “adequate,” or “appropriate” appear without observable evidence or a named decision-maker.
- Post hoc criteria. Requirements are added or relaxed after the team knows which decision they favor, without a documented rationale and formal review.
- One score hides a critical failure. A high overall readiness score is used to overlook a missing approval, safety procedure, or essential supply condition.
- Arm and site decisions are conflated. A platform-level approval is treated as proof that every site has the capability to recruit.
- Availability is assumed. The plan says a control or intervention will be accessible but provides no confirmation of supply or delivery arrangements.
- No version or date is recorded. Reviewers cannot establish which protocol documents were approved or when the arm became eligible to open.
A practical audit is to ask whether two independent reviewers, given the same checklist and evidence, would reach the same pass/fail decision. If not, clarify the criterion, the evidence standard, or the authority responsible. This audit assesses reproducibility of the operational decision; it does not establish that the scientific rationale itself is correct.
Choosing a Decision Structure and Recognizing Its Limits
The all-gates-must-pass rule is appropriate when each condition is essential. It is transparent and straightforward to audit, but it can delay opening while any one mandatory item is incomplete. A staged rule can be useful when the trial distinguishes approval to add an arm from authorization for individual sites to recruit. In that case, state the separate decisions and the gates for each; do not let a platform-level “go” be mistaken for site-level readiness.
A weighted score is an alternative when the goal is to summarize several nonessential readiness indicators, but it is a poor substitute for explicit safety, approval, or delivery requirements. If scores are used for monitoring, define their purpose and threshold in advance and retain mandatory gates separately. Informal committee discretion is another alternative, but without recorded criteria it can yield inconsistent decisions and make it difficult to explain why one proposed arm opened while another did not.
No checklist can guarantee scientific value, successful recruitment, or beneficial treatment effects. Readiness can change after a decision—for example, supply arrangements can fail or site capability can lapse—so the protocol should state how such changes are escalated and whether recruitment pauses. Criteria also need to be feasible: excessive or irrelevant requirements create delay without improving participant protection or scientific coherence. The governing principle is to require evidence proportionate to the risks and logistics of the specific intervention, and to apply the stated rule consistently.
Check Your Understanding
For each question, decide what the opening rule implies and compare your reasoning with the solutions below.
- A proposed arm has approved scientific rationale, available intervention supply, and trained staff. A required protocol approval is still pending. May the arm open under an all-gates-must-pass rule? Explain.
- A platform-level arm approval is complete, but one site lacks the required equipment and training. What is the appropriate opening decision for that site?
- Why can a weighted readiness score be unsuitable when an approval is a mandatory condition?
- Name two pieces of information that should be recorded when a gate is marked as passed.
Solutions. (1) No. The pending approval means at least one mandatory gate is zero, so the product of the gates is zero. (2) The site should not recruit until it meets the site-specific requirements and the evidence is reviewed; platform-level approval does not establish site readiness. (3) A weighted score can allow strong performance on other items to offset the missing approval, even though the protocol makes that approval essential. (4) Suitable entries include the evidence reviewed, the reviewer or responsible authority, the decision, and the date; the record should also identify the criterion and protocol version being assessed.