Why roles matter at an interim review
An interim review may produce information that is essential for a planned adaptation but unnecessary for many trial-team tasks. If a person who schedules sites, manages recruitment, or assesses outcomes learns comparative results that could influence those tasks, the review can affect trial conduct as well as analysis. A sound access plan therefore specifies not just what information exists, but which people may receive it and what they are authorized to do with it.
The earlier material on role-by-information matrices provides a way to lay out those assignments. Here the emphasis is on defining the roles and distinguishing four things that are often mistakenly bundled together: access to interim data, access to treatment codes, access to comparative outputs, and authority over operational decisions. These permissions should be chosen for the specific trial, its adaptation rules, and its governance documents; there is no universal job title that should always receive a given level of access.
The plan should distinguish information from authority. Seeing an arm-level comparison does not by itself authorize someone to change recruitment, and authority to implement a pre-specified adaptation does not automatically require access to the underlying comparative results. Similarly, being responsible for trial oversight does not mean that every member of an oversight group needs identical access.
Four information and authority categories
The following categories help make an assignment precise. Their boundaries depend on how the trial is designed and what the interim outputs contain.
| Category | What it includes | Access question |
|---|---|---|
| Interim data | Patient-level or summarized information available before the final analysis, such as accrued outcomes or follow-up status | Does this function need the data, and at what level of detail? |
| Treatment codes | The mapping that identifies which coded group corresponds to which treatment | Does this function need to know the treatment identities, or is coded information sufficient? |
| Comparative outputs | Summaries that contrast groups, such as arm-specific outcomes, estimated differences, or comparative safety information | Does this function need a pooled summary, a coded comparison, or a treatment-identified comparison? |
| Operational decisions | Actions affecting trial conduct, such as implementing an approved change or continuing recruitment under specified procedures | Who recommends, approves, communicates, and carries out each action? |
These categories are not interchangeable. A coded comparison can still reveal a difference between groups even when the treatment identities are hidden. A pooled summary can sometimes support monitoring without exposing an arm contrast, but pooling does not guarantee that no comparative information can be inferred from other available details. The access plan should describe the actual outputs and their context rather than rely on labels such as “blinded report.”
Defining trial roles by function
A role description is useful when it states the task, the information needed for that task, and the action the role can take. The labels below are common examples, not mandatory trial positions. A particular trial may combine functions, add functions, or assign them to an independent group.
- Blinded operational team: carries out routine trial conduct, such as site support and recruitment coordination, using information that does not disclose comparative interim results. Its decisions follow the approved procedures for those tasks.
- Unblinded statistician or analyst: prepares or interprets specified interim analyses when the review requires access to comparative outputs, and may have access to treatment codes if the analysis requires identified treatments. This does not automatically give the analyst authority to change trial operations.
- Independent review group: reviews specified interim evidence and provides a recommendation or decision within the authority assigned by the trial’s governance documents. Membership, information access, and authority must be defined rather than assumed from the group’s name.
- Decision authority: the person or body designated to approve a specified action. It may receive a recommendation rather than detailed results, depending on the trial’s rules and the decision required.
- Implementation role: carries out an approved action, such as applying an authorized change to trial procedures. The implementer may need the action and its effective date without needing comparative results or treatment codes.
- Data custodian: maintains or supplies data according to authorized processes. Custodianship alone does not imply permission to interpret comparative results or make adaptation decisions.
Names alone are inadequate. “Statistician” may refer to a person who prepares blinded descriptive summaries or to an analyst who sees treatment-identified comparisons. “Sponsor representative” may describe a range of functions with different information needs. The plan should resolve these differences at the level of individual functions and assigned people.
Worked Example: Assigning roles for an interim review
Worked Example: Assigning roles for an interim review
Consider a hypothetical two-group trial comparing a new treatment with control. At an interim review, an independent review group must assess safety and whether a pre-specified rule supports an adaptation. The site team continues recruitment unless it receives an authorized operational instruction. A statistician prepares the review outputs, and a trial manager implements approved instructions. Determine a defensible division of access and authority.
Step 1: Identify the required work. The analyst must calculate the specified comparisons, so the analyst needs the interim data required by the analysis and the treatment-code mapping if the outputs must identify new treatment versus control. The independent group needs the outputs needed to assess the rule. If its charter requires treatment-identified safety comparisons, it needs those comparisons; if coded comparisons are sufficient for its task, treatment identities need not be included.
Step 2: Separate review from implementation. The independent group makes or recommends the decision assigned to it. The trial manager receives the authorized instruction and its implementation details. Unless the manager has a separate review function, access to the underlying comparative outputs is not required merely to carry out that instruction.
Step 3: Protect routine operations from unnecessary comparative detail. The site team receives routine trial information and any operational instruction needed for its work. It does not receive interim arm comparisons or treatment codes in this scenario. The instruction should state what to do, not disclose the result that motivated it unless the approved procedure requires disclosure.
Conclusion. Access follows the task: the analyst prepares the necessary analysis; the independent group reviews the designated outputs; the manager implements an authorized action; and the site team follows the resulting operational instruction. The plan should still specify exactly which outputs the reviewer receives and what decision authority it has.
A justification for separating viewing and acting
The distinction between information access and operational authority can be justified by tracing what each task requires. Suppose the analysis function needs comparative output \(C\) to evaluate a pre-specified rule \(R\). The implementation function needs an authorized instruction \(I\) to carry out a decision. If the governance process determines \(I\) from \(R\) and the review, then the implementer’s task may require \(I\), but not necessarily \(C\). Giving the implementer \(C\) adds information without being necessary to perform the defined implementation task.
This is a task-based justification, not a claim that every trial can or should use this exact sequence. The review group’s authority, the information needed to make a safe decision, and applicable governance requirements may require a different assignment. The important check is to ask whether each recipient needs the information itself, or only a decision or instruction derived from it.
Worked Example: A coded comparison is still comparative information
In a hypothetical interim report, Group A has 18 events among 120 participants and Group B has 29 events among 118 participants. The treatment-code mapping is withheld from a recruitment coordinator, who sees only the labels A and B. Does withholding the mapping mean the coordinator has no comparative information?
No. The reported event proportions are \(18/120=0.1500\), or 15.00%, in Group A and \(29/118\approx0.2458\), or 24.58%, in Group B. The coded difference is approximately \(0.2458-0.1500=0.0958\), or 9.58 percentage points. The coordinator can see that the outcomes differ between coded groups, even without knowing which group received the new treatment.
Therefore, hiding treatment identities does not make the comparison non-comparative. If the coordinator’s task does not require the group contrast, the plan should not assume that a coded comparison is an appropriate routine operational output. Whether any particular summary is sufficiently non-revealing depends on the whole information available to the recipient.
Worked Example: Resolving an ambiguous decision role
Worked Example: Resolving an ambiguous decision role
In another hypothetical trial, a committee charter says that a review committee “oversees the adaptation,” while a sponsor operations lead is responsible for changing site procedures. The analyst provides an arm-level interim summary. The documents do not say whether the committee decides, recommends, or only reviews. What should be clarified before access is assigned?
First, state the decision explicitly: for example, whether the committee recommends an adaptation under a specified rule, approves it, or merely assesses the evidence. Second, identify the designated approver and the permitted basis for that approval. Third, define the operations lead’s implementation task and the instruction required to perform it. Finally, match each role to its information need: the committee may require comparative outputs to assess the rule; the operations lead may require only the authorized change and its implementation details.
The ambiguity cannot be resolved just by giving both parties the full report. That would provide information but would not establish who has authority. The governance documents and access plan must align on who reviews, who decides, and who acts.
Diagnostics, alternatives, and limitations
A role assignment may be going wrong if a recipient cannot explain why a particular output is needed for their task, if “oversight” is used without defining decision authority, or if routine staff receive comparative results despite having no assigned review function. Another warning sign is a report described as blinded even though it contains coded arm contrasts that could affect behavior. These are reasons to revisit the assignment, not proof by themselves that bias occurred.
A role-based plan is one way to organize access decisions. An alternative is to specify permissions person by person or by each individual interim review. Person-specific assignments can make accountability clear, but may be hard to maintain if responsibilities change. Review-specific assignments can reflect the needs of a particular decision, but can become inconsistent across reviews unless they are recorded and checked against trial governance. A role-based plan supports consistency, but can be too broad if a role combines people with different tasks. The suitable approach depends on the trial; each should state the actual information and authority assigned.
Role definitions also have limits. They do not establish that an output is safe to disclose, that a decision rule is statistically valid, or that a recipient will interpret information as intended. People can infer treatment differences from indirect signals, and circumstances can change after the plan is written. The plan therefore needs to be consistent with the protocol and relevant oversight documents, and access decisions should be revisited when roles, outputs, or trial procedures change.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions.
- A data custodian supplies interim records to an analyst. Does that function alone establish that the custodian may interpret arm-level results? Explain.
- A coordinator sees outcomes for coded Groups A and B but not the treatment mapping. What can the coordinator still learn?
- A committee is described as “responsible for adaptation.” Name two distinct responsibilities that description could mean and one clarification needed for each.
- A manager is instructed to implement an approved change. Give an example of information the manager may need and information that may not be necessary for implementation.
- Why might a role-based assignment be insufficient when one job title covers both routine operations and interim analysis?
Solutions. 1. No. Supplying records is a data-handling task; permission to interpret comparative results must be separately specified. 2. The coordinator can learn the difference between coded groups, even if the treatments corresponding to A and B are unknown. 3. The committee might recommend an adaptation or approve it. Clarify which body has the final authority and what the committee’s role is under the governance documents. 4. The manager may need the approved change, effective date, and implementation instructions; the comparative results that motivated the change may not be needed. 5. The two functions may have different information needs and authorities, so a single broad assignment may disclose comparative results to people whose routine tasks do not require them.