Why Platform and Arm Priorities Can Collide
A platform trial coordinates multiple treatment comparisons under shared governance and infrastructure. That structure can improve consistency and efficiency, but it also creates competing demands. A platform-wide objective, such as maintaining reliable shared data processes, may conflict with an arm’s need for a timely answer to a population-specific question. Conversely, an arm’s request for priority access to staff or shared resources may undermine the platform’s ability to meet obligations to other arms and participants.
These disputes are not resolved well by asking which committee has the strongest preference, which sponsor is most influential, or which option seems most efficient in isolation. A defensible resolution begins by identifying the decision, its constraints, and the authority to make it. The governance charter and the established decision-rights structure provide the basis for that work; a dispute is not an opportunity to invent authority or bypass an agreed process.
A Constraint-First Method
A useful resolution method separates non-negotiable constraints from priorities that may be balanced. Constraints include applicable ethical and regulatory requirements, participant safety, approved protocol commitments, data integrity, and the boundaries of delegated authority. They cannot be set aside merely because a competing goal is important. Within those boundaries, decision-makers can compare options and make proportionate trade-offs.
The ordering below is a practical governance framework, not a substitute for the trial’s own charter or applicable requirements. If those sources specify a different authorized process, follow that process. Where they do not settle the dispute, use criteria that are recorded before the decision rather than choosing criteria to favor a preferred outcome.
State what must be decided, who is affected, and what each party is asking for. Separate evidence about the need from the proposed solution.
Check participant-protection obligations, approved commitments, data and operational requirements, and the charter’s delegation map. Do not proceed with an option that violates a constraint or exceeds the decision-maker’s authority.
Consider urgency, likely impact on participants and scientific validity, effects on other arms and shared operations, feasibility, resource use, and reversibility. Apply the same criteria to each option.
The authorized body chooses among feasible options. Escalate unresolved or out-of-scope matters to the designated authority; record the rationale, any dissent, implementation owner, and conditions for review.
The criteria help distinguish a real priority from a preference. For example, urgency should be supported by a consequence of delay, not merely by the fact that one party asks first. Impact should include effects on participants and the reliability of the evidence, not only convenience for the requesting arm. Feasibility includes staff, data, and operational capacity. Reversibility matters because a temporary, reviewable arrangement may be preferable when uncertainty is high and it does not breach a constraint.
Why Constraints Come Before Trade-Offs
The ordering has a simple justification. Suppose each proposed option is first screened against a set of binding requirements. Any option that fails a requirement is not an admissible choice, regardless of its score on efficiency or scientific value. Only the remaining options are compared on balancing criteria. This prevents an attractive aggregate benefit from being used to excuse a breach of participant protections or authority.
Here, \(\mathcal{A}\) is the set of proposed options, \(C_j(a)=1\) means option \(a\) satisfies binding constraint \(j\), and \(\mathcal{F}\) is the set of feasible options. The authorized decision-maker compares options only within \(\mathcal{F}\). If \(\mathcal{F}\) is empty, the right response is not to waive a constraint informally; it is to revise the options or escalate through the proper process.
A weighted score can help organize discussion among feasible options, but it should not be treated as a universal formula. Different governance settings may not have agreed weights, and a precise-looking total can conceal a value judgment. When weights are not established in advance, a transparent comparison by criterion, including recorded reasons for any trade-off, is more defensible than an improvised numerical ranking.
Worked Example: Allocating a Shared Data-Management Team
In a hypothetical platform, a shared data-management team has capacity for 12 staff-days next week. Arm A requests 8 days to resolve queries before a planned internal data review; Arm B requests 7 days to complete routine cleaning. The platform team estimates that Arm A needs at least 6 days to meet its stated review timeline, while Arm B can complete its work over the following two weeks without disrupting scheduled participant visits or the reliability of its data. The charter gives the executive committee authority to allocate shared operational capacity within the approved work plan.
Step 1: Define the decision. The committee must allocate 12 staff-days between two documented requests. Neither request automatically takes precedence because it is arm-specific or platform-wide.
Step 2: Check constraints and authority. The facts specify no safety, regulatory, data-integrity, or approved-commitment requirement that would be breached by scheduling Arm B’s routine cleaning later. The allocation is within the executive committee’s stated authority. If either condition were different, it would need to be addressed before balancing convenience or efficiency.
Step 3: Compare feasible allocations. Giving Arm A 6 days and Arm B 6 days uses all 12 available days. It meets Arm A’s minimum need for its review timeline and allows Arm B to begin its work, with the remaining 1 day of its request scheduled in the following two weeks. The delay to Arm B has a stated operationally acceptable window, whereas fewer than 6 days for Arm A would miss the stated timeline.
Decision and record. Under these assumptions, allocating 6 days to each arm next week is a proportionate response. The decision record should identify the capacity estimate, the reason for the short-term priority, the date by which Arm B’s remaining work is expected, and who will check that the schedule remains feasible. If the estimate of Arm B’s delay impact changes, the committee should reconsider rather than treating this allocation as a permanent precedent.
Applying the Method to Different Disputes
Worked Example: An Arm Requests an Earlier Scientific Review
In a hypothetical platform, an arm’s investigators request a scientific review two weeks earlier than the usual schedule because they believe emerging evidence may affect the relevance of their research question. The scientific committee’s calendar is shared across arms. The request includes a concise evidence summary, but the committee has not established that a two-week delay would compromise participant protection or an approved commitment.
Clarify what is being requested. The request is for an earlier review, not an automatic endorsement of a particular scientific conclusion. The relevant question is whether the urgency and expected value justify displacing other scheduled work.
Apply common criteria. The committee compares the evidence and consequences of waiting with the impact on other scheduled reviews, available expertise, and whether the review can be conducted adequately at the proposed time. It should ask the other affected parties for relevant information, not allow the requesting arm alone to define the impact.
Resolve within authority. If the charter authorizes the scientific committee to set its own agenda and the earlier review is feasible without compromising another material obligation, it may reprioritize and record why. If reprioritization would displace a decision reserved to another body or threaten an approved commitment, the committee should escalate that part of the request. The outcome may be an earlier review, a review at the original time, or a request for further information; the evidence and authority determine which option is defensible.
Worked Example: A Request for an Exception to a Shared Process
In a hypothetical trial, one arm asks to send a data extract to its analysts before a platform-wide quality check is complete. The arm says the extract is needed to prepare a scheduled scientific discussion. The shared process is intended to identify material data inconsistencies before interim data are circulated. The request does not identify an approved exception, and the person receiving it does not have authority to change the process.
Screen for constraints. The possibility that preliminary data may be misleading or inconsistently checked is directly relevant to data reliability. The request also exceeds the recipient’s authority. Convenience or timing cannot make an unauthorized release a feasible option.
Identify alternatives and escalation. The recipient should not grant the exception informally. They can ask whether a version of the discussion can proceed using information already approved for that purpose, or refer the request to the body empowered to consider it. Any authorized decision should specify what information may be shared, with whom, under what safeguards, and whether the decision is limited to this circumstance.
Explain the outcome. On the facts given, the extract should not be sent by the unauthorized recipient. The request may be considered through the proper route, but that is distinct from an assumption that the exception will be granted. The record should preserve the request, the authority check, and the basis for the next step.
Decision Records, Diagnostics, and Failure Modes
A useful decision record is concise but specific enough that another authorized reviewer can understand how the outcome was reached. It should identify the issue and parties affected, the governing constraints, the options considered, the criteria applied, the decision owner, the rationale, any dissent, and the person responsible for implementation. Where uncertainty or changing conditions matter, record what information would trigger reconsideration and when the matter will be reviewed.
Several warning signs indicate that the process may be going wrong:
- Criteria appear only after a preferred option is known. This can make the analysis look tailored. State the criteria before comparing options, or explain openly why a new criterion became relevant.
- One arm’s urgency is accepted without evidence. Ask what consequence follows from delay and compare that consequence with impacts on other arms and shared operations.
- Authority is assumed rather than checked. Refer to the charter and decision-rights structure. A committee’s expertise does not automatically grant it decision authority.
- An exception becomes an informal precedent. Record whether a decision is case-specific, what safeguards apply, and which authority can approve any broader change.
- Dissent or affected parties are omitted. A decision may still be reached, but missing information or unrecorded disagreement weakens its accountability and can conceal foreseeable harm.
An unresolved conflict should be escalated when the charter assigns it to another body, the decision exceeds delegated authority, material constraints are unclear, or authorized decision-makers cannot agree. Escalation should include the question to be decided, the options, relevant evidence, constraints, and the specific authority sought. Sending an issue upward without a focused question can delay resolution; deciding it locally without authority is not a valid shortcut.
Choosing a Resolution Approach
Three approaches are often proposed. A platform-first rule gives shared objectives priority in every dispute. It is simple, but may dismiss a legitimate arm need even when a limited adjustment is feasible and has little platform-wide cost. An arm-first rule has the opposite problem: repeated local exceptions can erode consistency and disadvantage other arms. Case-by-case balancing is more responsive, but without defined constraints, criteria, and authority checks it can become inconsistent or vulnerable to influence.
The constraint-first method retains the useful parts of balancing while preventing avoidable governance failures. It does not promise that every party receives its preferred outcome. It requires that comparable requests receive comparable consideration, that hard obligations remain intact, and that the authorized body explain any trade-off. Its main limitations are that criteria may require judgment and that incomplete evidence can make impacts uncertain. Those limitations should be made visible, not hidden behind an arbitrary score.
Common Mistake
Check Your Understanding
Answer each question using the constraint-first method. Brief solutions follow the questions.
- A committee receives two competing requests for shared staff time. What should it establish before ranking which request is more urgent?
- A proposed option would meet an arm’s timeline but exceed the committee’s authority. Can the committee choose it if the platform-wide benefit is substantial? Explain.
- Why is a weighted score not automatically a fair way to settle an arm-versus-platform dispute?
- A request is urgent, but the evidence about the cost of delay is incomplete. Name two defensible next steps.
- What should a decision record include if the committee chooses a temporary arrangement that may need review?
Solutions. (1) Define the decision and affected parties; check binding constraints and authority; then compare the requests using stated criteria. (2) No. The option is outside the committee’s authority and is not a feasible choice for that committee; it should seek the designated authority. (3) A score can conceal unagreed weights, omit important constraints, or make value judgments appear objective. Use it only where the criteria and weights are justified and appropriate; otherwise compare criteria transparently. (4) Request information that clarifies the consequences of delay, and consider a reversible interim option if it satisfies constraints and is within authority; otherwise escalate. (5) Record the rationale, scope and duration, safeguards, implementation owner, review date or trigger, and any dissent.