Why Communication Rules Matter
Interim reviews can involve information that is important for trial oversight but sensitive to disclose. A recipient who learns comparative results may change recruitment, care, outcome assessment, or other behaviour, even without intending to do so. The resulting risk is not limited to a table labelled by treatment arm: a verbal hint, an unbalanced summary, or an answer to a seemingly routine question can also convey comparative information.
A communication protocol specifies who may receive which information, for what purpose, through which channel, and how questions are handled. It complements role-based access and safe interim reporting by governing the human interactions around a review. It should be agreed before interim results are available, consistent with the trial’s oversight arrangements, and sufficiently clear that staff do not have to invent a response under pressure.
Classify the Information Before It Is Discussed
A useful protocol separates information by its disclosure risk and intended use. The categories below are a practical starting point; the trial’s governance arrangements determine the actual categories and recipients. A label such as “pooled” or “masked” is not, by itself, proof that an item is safe to circulate.
| Information category | Examples | Communication rule |
|---|---|---|
| Routine operational information | Overall recruitment progress, data query status, or review meeting logistics | May be shared with the teams who need it, after checking that the content does not indirectly reveal comparative results. |
| Non-comparative interim summary | Overall counts or pooled summaries prepared for a specified operational purpose | Share only with approved recipients and only after checking whether context, small groups, or other known information could permit an arm-level inference. |
| Comparative interim information | Outcomes or safety data separated by treatment group, or a clear signal about which group is doing better | Restrict to recipients authorized to review comparative results for the stated oversight purpose. |
| Review deliberations and decisions | Discussion of a recommendation, its rationale, or the timing of a decision | Communicate only the approved decision and necessary implementation details to each recipient group; do not circulate restricted deliberations by default. |
The key question is not simply whether a report contains treatment labels. Ask whether a recipient could combine the proposed content with information already available to infer a treatment difference. For example, a pooled number may be revealing when the recipient knows the outcome for one group, when one group’s data are almost complete and the other’s are not, or when a subgroup contains very few participants.
The protocol should name roles rather than rely on vague phrases such as “the study team.” Depending on the trial’s approved oversight structure, recipients might include a data monitoring committee, its independent statistician, a designated unblinded statistician, a sponsor representative with a defined oversight function, or a blinded operational team. Not every trial uses the same bodies, and membership alone does not establish a need to receive every type of information. For each role, specify the purpose for which it may receive information and the content it may discuss onward.
Worked Example: Assigning Content to Recipients
Suppose a hypothetical two-group trial has an independent review committee and a blinded site operations team. At a scheduled review, the committee must assess comparative safety information; the site operations team needs to know whether the next scheduled recruitment period is proceeding.
The protocol can authorize the committee and its designated unblinded statistician to receive the treatment-group safety tables needed for that review. The blinded team can receive the operational decision, such as “continue recruitment under the current procedures,” if that is the approved communication and it does not disclose the committee’s comparative deliberations. The site team does not need the tables to carry out that instruction.
Before sending the operational message, the sender checks that it contains no arm-specific counts, comparative wording, or extra rationale that would reveal the result. The message records the decision, effective date, and contact for implementation questions. This separates the information needed to act from the restricted information used to reach the decision.
Write Rules for Meetings, Messages, and Questions
A protocol should cover more than the final report. Disclosure can occur in meeting invitations, agendas, screen sharing, slide decks, email subject lines, informal conversations, or follow-up notes. Set rules for each channel that will actually be used. For example, restrict calendar titles to neutral logistics, confirm attendee lists before discussing comparative material, and check that presentation screens and shared folders are appropriate for the current audience.
The meeting chair or designated communication lead should know which agenda items are open to which attendees. If a discussion moves from operational matters to restricted comparative review, the chair should pause, confirm who is present and authorized, and only then proceed. When a person is not authorized for the next topic, the transition should be handled neutrally, without summarizing the restricted discussion in their presence.
Questions need an equally explicit route. A recipient may ask “Is one group having more events?”, “Why was recruitment paused?”, or “Did the committee see a concerning result?” Even a yes-or-no answer could disclose information. Staff should not improvise, hint, confirm a guess, or deny a difference unless the approved communication specifically permits that content for that recipient.
Determine whether the question asks for comparative results, restricted deliberations, or an operational fact that is approved for the requester.
Use the protocol’s recipient rules. Do not infer authorization from seniority, familiarity, or an urgent tone.
If the request is outside the recipient’s permitted information, use a neutral response and route the question to the designated contact.
Document the request, the response, any escalation, and the information ultimately authorized for release, using the trial’s established records.
A neutral response should be truthful and should not imply that a difference exists or does not exist. For example: “I cannot discuss comparative interim information. I can refer your question to the designated trial contact, who will determine whether an approved response is available.” If an approved operational instruction is available, provide that instruction without adding an unapproved explanation. This is a communication boundary, not a claim that the question is unimportant.
Worked Example: Responding to a Site Question
In a hypothetical trial, a site coordinator hears that an interim review has occurred and asks the study manager, “Did the new treatment have fewer serious events? We are deciding how to counsel the next participant.” The coordinator is authorized to follow the approved consent and care procedures, but is not an authorized recipient of comparative interim results.
The manager should not answer “yes,” “no,” “probably,” or “I cannot confirm that,” because even a denial or hint may communicate information. A suitable response is: “I cannot discuss comparative interim information. Please continue to use the current approved procedures; I will route your question through the designated trial channel.” The manager then alerts the designated contact and documents the request and response.
If the trial’s approved oversight process later provides an instruction that changes participant-facing procedures, the site receives the approved instruction and its effective date. It does not automatically receive restricted comparative results or deliberations that were not needed to implement the instruction.
Why Pooled Information Can Still Reveal a Difference
A summary can be pooled and still be disclosive. Consider two groups with outcome counts \(Y_1\) and \(Y_2\). The pooled total is \(Y=Y_1+Y_2\). If a recipient already knows \(Y_1\), releasing \(Y\) reveals the other count exactly: \(Y_2=Y-Y_1\). The same logic applies to some rates and averages when denominators or one group’s values are known. Therefore, “we removed the group labels” is not a sufficient review of disclosure risk.
More generally, the pooled mean is a weighted average of the group means. If group sizes are \(n_1\) and \(n_2\), then
When \(n_1=n_2\), this reduces to \(\bar{x}_{\mathrm{pooled}}=(\bar{x}_1+\bar{x}_2)/2\). If a recipient knows the pooled mean and one group mean, the other group mean can be calculated as \(\bar{x}_2=2\bar{x}_{\mathrm{pooled}}-\bar{x}_1\). This derivation explains why a pooled mean can disclose a comparison when combined with outside knowledge. It does not mean all pooled statistics are unsafe; it means their safety depends on the full information available to recipients.
Worked Example: Checking a Pooled Summary for Leakage
Suppose, hypothetically, that a blinded operations team is told the pooled mean symptom score is 6.0 points across two equally sized groups. A staff member already knows that the mean for the first group is 5.2 points from an unauthorized conversation. The pooled value is not safe to repeat to that person: the second group’s mean follows from
The arithmetic is correct because the group sizes are equal. The communication lesson is that the apparently pooled figure, combined with the recipient’s prior knowledge, reveals a 1.6-point difference between the group means. The appropriate response is to avoid further discussion of the pooled statistic with that recipient and refer the matter to the designated communication lead. The lead can determine whether a different operational summary is genuinely necessary and safe for the intended audience.
Build in Checks and Recognize Limits
Before releasing any interim-related communication, check the recipient list, purpose, content, and channel. Confirm that the message matches the approved decision or report; remove unnecessary comparative details; and consider what recipients may already know. For oral discussions, verify attendance and stop if an unapproved person joins. Afterward, confirm that notes and follow-up messages preserve the same boundary as the meeting itself.
Warning signs include informal requests for “just a quick look,” discussion of which group is “ahead,” explanations that go beyond the authorized decision, unexpected forwarding of materials, or uncertainty about who may answer a question. These signals call for pausing and consulting the designated contact, not for guessing whether a disclosure would matter. Communication rules cannot guarantee that no one will infer a result, and they cannot replace appropriate access controls, approved reporting, or oversight. They provide a consistent way to reduce avoidable disclosure and respond when a boundary is tested.
A communication protocol is preferable to relying on individual judgment alone because it makes recipient and response rules repeatable. It is also more precise than a blanket rule to “say nothing”: some roles need timely operational instructions, and an unhelpful refusal can delay legitimate work. The protocol’s task is to authorize the minimum information needed for each purpose while preserving restrictions on comparative results.
Worked Example: Choosing a Safe Follow-Up
After a hypothetical review, an operations lead is authorized to tell sites that a scheduled visit window has changed from days 28–35 to days 28–42. A site manager asks whether the change was made because one treatment group had more missed visits.
The lead can share the approved visit-window change, including its effective date and the procedure for participants already in the window. The lead should not attribute the change to a comparative pattern unless that explanation is expressly approved for the site. The question is routed to the designated contact, and the request is documented. This response gives the site what it needs to implement the change while avoiding an unsupported inference about the review’s findings.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions below.
- A blinded site team requests the treatment-group event table because it wants to plan next week’s staffing. What should the communication lead check, and what information should ordinarily be sent?
- Why can a pooled total reveal a group-specific count? State the algebraic relationship.
- A colleague asks, “I heard the committee recommended continuing. Does that mean the treatments performed similarly?” Give an appropriate response.
- Name two communication channels, other than a formal report, that should be covered by the protocol.
- What is one limitation of communication rules, even when recipients and content are carefully specified?
Solutions. (1) Check the team’s authorization and the stated purpose; staffing needs do not by themselves authorize access to comparative event data. Send only an approved operational staffing instruction or summary that has been reviewed for indirect disclosure. (2) If \(Y=Y_1+Y_2\), then knowing \(Y\) and \(Y_1\) gives \(Y_2=Y-Y_1\), so the pooled total can disclose the other group’s count. (3) Say that comparative interim information and restricted deliberations cannot be discussed, and refer the question to the designated contact; do not imply that continuation proves similarity. (4) Examples include email, meeting invitations, screen sharing, verbal conversations, and follow-up notes. (5) Rules cannot guarantee that recipients will not infer information from other knowledge, nor can they replace access controls and oversight; they reduce avoidable disclosure and standardize responses.